Address by Lars Frisell to Irish Economy Conference: Learning from the Crisis

‘The Timing of Macroprudential Intervention: The Central Bank’s new LTV and LTI regulations’

On January 27, 2015, the Central Bank of Ireland announced the introduction of new limits to mortgage lending by financial services providers regulated in Ireland. The rules came into force on February 11. Most people in this audience are probably familiar with the regulations, but let me make a brief recap. PDH mortgages for non-first time buyers are subject to a proportionate cap of 80 per cent LTV.

• For first time buyers, the LTV cap will be 90 per cent for properties valued at up to €220,000 with an 80 per cent cap applied on any value over this amount.
• BTL mortgages are subject to a proportionate cap of 70 per cent LTV.
• PDH mortgage loans are subject to a proportionate cap of 3.5 times loan to gross income (LTI).

The word “proportionate” refers to the important degree of leeway built into the rules, according to which lenders may make loans in excess of the caps, subject to an overall ceiling on the share of their lending. Thus, for each institution, the leeway for loans in excess of the LTV caps on primary dwellings and buy-to-lets amounts to 15 and 10 per cent respectively of the euro value of granted loans, while banks may exceed the debt-to-income cap to the value of 20 per cent of new loans. This leeway for lenders is very important. Unlike microprudential rules, such as capital requirements, the new regulations do not primarily target the resilience of individual banks, but the long-term stability of the financial system. It is not that every loan above a particular threshold is risky, but that too many high LTV or LTI loans can destabilise the system as a whole. Hence, the rules are necessarily broad-brushed and one can imagine many circumstances in which it makes sense to make use of the leeway. For example, this could include borrowers who have a high and stable income but do not quite have the required deposit - or the other way around. Also, the presence of a well-designed mortgage insurance, a personal guarantee, or additional collateral might motivate a loan above the caps. In addition, the new rules do not apply to home owners in negative equity. This is important so as not to further impede the mobility of borrowers in this difficult situation.

The Central Bank regulations aim both to decrease the likelihood of property bubbles developing and to increase the resilience of households and banks in case they do. There seems to be little controversy over the general desirability of these kinds of rules in Ireland - very few of the responses to the Central Bank consultation argued that the regulations should be scrapped in their entirety. However, concerns have been raised that the loan-to-value cap in particular might significantly affect house prices and as a consequence decrease the supply of housing. The remainder of my speech focuses on these aspects. ¹

Experiences from other countries

To date, a dozen European countries have implemented restrictions on either loan-to-value and/or debt-to-income ratios. Most of these are recent regulations, put into place after the financial crisis. In South-East Asia several countries have implemented these and other restrictions – often including limits on exposures in foreign currency – based on their experiences in the Asian crisis in the late 1990s. It is of course ironic that few countries have been able to learn from previous experiences, the political impetus to take pre-emptive action only seems to materialise after a crisis in your own country. Hong Kong, who implemented their first loan-to-value regulation in the early 1990s, is a notable exception to this sorry rule.

There are a few studies that have examined the effects of loan-to-value restrictions. One should bear in mind that it is tricky to empirically separate the effect of these regulations from that of other variables, such as changes in interest rates and fiscal conditions. However, the evidence is still pretty conclusive. In summary, loan-to-value restrictions have usually had a distinct impact on market turnover – i.e., the number of transactions decreases - but only a small and temporary effect on house prices, lending, and housing supply.² This is probably the reason why the loudest opposition to macroprudential regulations often comes neither from consumer organisations nor banks, but real estate brokers. Encouragingly, transactions seem to decrease more among those who already own a property, including potential speculators, rather than first-time buyers.³

Intuition

When one considers all possible interdependencies and potential feed-back loops it seems one can’t approach the issue without a fully-fledged model of the housing and credit markets. Firstly, a restriction on the minimum deposit will constrain a number of prospective, mostly first-time, buyers, who will no longer possess the required deposit for the property they had in mind. This should lead to lower prices at the lower end of the market, though not necessarily for the cheapest objects for which demand could instead increase. On the other hand, lower prices may attract some buyers who have the required deposit, but had previously considered prices too high. In addition, some next-time-buyers could decide to take advantage of the lower prices at the lower end and trade down. This will even out the price effect across the property spectrum. Finally, to the extent that prices do fall, this will mitigate the deposit constraint that caused the price fall in the first place. At the end of it, although some buyers on the verge of buying will necessarily be frustrated, is it even clear whether first-time buyers in general will be worse or better off from the regulations? (When the Swedish FSA introduced a loan-to-value regulation in 2010 I received a bouquet of flowers with a card that read: “Thank you Lars - From two first-time buyers”. To this day I don’t know if that was sincerely or ironically meant, but the flowers were real...)

If the effect on house prices was expected to be strong and long-lasting it would decrease the profitability of construction companies and the pace of house construction should decrease. However, the loan-to-value and the loan-to-income cap are pure liquidity constraints, they force borrowers to accumulate (more) capital before buying a property but unlike changes in taxes and subsidies they do not alter households’ underlying debt-servicing capacity. Since the overall income in the economy does not change, both housing demand and supply should instead shift to rental accommodation. This means that after an adjustment process, both rents and house prices should be expected to return to their initial levels.⁴ Of course, especially in Ireland where the tradition of home ownership is so deeply rooted, this process could take some time. In the Central Bank's simulation, at their trough house prices are only 1.3 per cent lower than in the case without regulations, which occurs in the third year after the implementation. The effect on housing supply reaches its maximum in the second year, with house construction about 3 per cent lower (about 550 units) than in the alternative scenario.⁵

One does not need an intricate model to understand why the effect on prices will be limited. Take an example of a first-time-buyer household that is currently renting and has accumulated a deposit of 30.000. Assume that the household has no assets to liquidate and that there are no gifts or inheritances from the older generation. Finally assume that mortgages are amortised over 30 years. If the household before the loan-to-value cap could borrow 90 per cent of the price of a property, they would have been able to secure a loan enabling them to buy a house for 300.000. (From Central Bank data we know that the average LTV for the relatively small number of first-time-buyers who borrowed from Irish banks in 2014 was around 75 per cent, with about half of them at 90 per cent or higher.) However, with the new regulations the same deposit will only allow them to purchase a house for 260.000. This means that the potential price drop could be 40.000, or 1 3 per cent. If the original LTV allowed by banks was instead 95 per cent, the buyer could have secured a loan enabling them to buy a house priced as high as 600.000 and the price fall could theoretically be 57 per cent.⁶

But sellers would surely not sell at these prices. To see why, assume for simplicity that the household’s current rent is equal to the user cost that buying would entail.⁷ Since the household is able to amortise the loan in 30 years it follows that they would be able to save the extra deposit required in just 11 months in the first case, and in about 3 and a half years in the second case.⁸ (If rents are higher than the user cost of buying the household will need more time to accumulate the incremental deposit, and vice versa.) Realising this, most sellers will surely prefer to wait. So instead of a house price fall, we are likely to see a slowdown in the market for some time, with fewer objects for sale and more to rent, which is exactly what we have seen in other countries.

In practice, price effects are further mitigated by the fact some buyers, though far from all of course, have some assets they can liquidate or have parents or other relatives that will help them out. And since the effect on prices is expected to be small, there is no reason to expect that housing supply should contract in any significant way. Again, if anything we should instead see some shift towards construction of rental accommodation. (If it were the case that buyers initially planned to get interest-only mortgages and to save nothing henceforth - through amortisation or otherwise - then the price effect would be permanent. But then we would likely have a bigger problem looming, one which we would be happy to avoid.)

The timing of macroprudential intervention

Various critics have argued that the regulations are untimely, as Irish house prices are still below their fundamental levels and construction activity is very low. This argument is difficult to understand. If house prices are depressed because of tight credit conditions and high real interest rates - which I think is true - then the new regulations are less likely to be a binding constraint. The effects are not additive. For example, the proportion of new mortgage loans that exceeded the new LTI cap of 3.5 was over 70 per cent at the peak of the credit bubble in 2007, but had declined to less than 20 per cent by 2013.

There are probably more important factors hampering house construction, including zoning restrictions and fraught relationships between lenders and builders. It is of course important to resolve these problems, however the Central Bank regulations neither mitigate nor exacerbate them.

The fact that this kind of regulation does not seem to significantly affect either house prices or credit growth may lead some observers to conclude that they do not “work”. But that is also a fallacy. As long as changes in house prices are motivated by fundamentals – genuine demand for housing – the rules should have little effect. But if the housing market has become speculative, where credit and house prices grow partly based on expectations of further house price increases, the rules will have a significant effect. The reason is that, at that stage, buyers’ incomes will neither pass the loan-to-income rule, nor will they support sufficient savings to accumulate future deposits. Hence, the (irrational) expectation of ever-increasing prices is thwarted, and credit and house price growth will revert to a sustainable path. Hence, macroprudential regulation, by putting prudent norms in the credit market on a legislative footing, can counteract financial bubbles without significant side effects in normal times.

The dramatic increase in Dublin house prices – 42 per cent in the 18 months to in October 2014 - can easily be interpreted as a bounce-back after an overshoot on the downside. But these price rises also show how easily a bubble psychology could take hold. From that point of view it has been important to establish that the open-ended provision of lending that is associated with housing bubbles is not accepted.

Finally, it is near-impossible to determine what is the fundamental level of house prices - there are almost as many house price models as economists. Nor is it possible to say, for example, what is the “correct” loan-to-value cap. But fortunately we do not need to. Experience shows that as long as the rules do not deviate too much from (good) market practice, effects will be muted and short-lasting. The worst policy mistake is to delay implementation too long, until loan levels and house prices have reached unsustainable levels. As we know all too well in Ireland, at that stage no policy option is appealing.

____________________________________________________________________________________________________

¹ For an effective overview of macroprudential instruments and country experiences see Grace, Hallissey and Woods (2015).

²  See Igan and Kang (2011) for the case of Korea, Ahuja and Nabar (2011) for Hong Kong, Price (2014) for New Zealand and Finansinspektionen (2014) for Sweden.

³ Igan and Kang (2011)

⁴ See Kennedy and Stuart (2015) for an analytical framework.

⁵ See Cussen, O’Brien, Onorante and O’Reilly (2015)

⁶ Applying the LTV-regulation for first-timer-buyers gives that 30.000 = 0.1 * 220.000 + 0.2 * 40.000, and 220.000 + 40.000 = 260.000.

⁷ Several factors including taxes, rental regulations, and the relative propensities to let and rent may cause a wedge between rents and user costs. See Kennedy and Stuart (2015) for a discussion.

⁸ The deposit required for a 300.000 home for a first-time buyer under the new regulation is 0.1*220.000 + 0.2*80.000 = 38.000. The household was able to amortise a loan of 270.000 in 30 years, i.e. 9000 per year, so the extra 8.000 could be saved in slightly less than 11 months. The deposit required for a house worth 600.000 under the new regulation is 0.1*220.000 + 0.2 * 380.000 = 98.000. Analogously to above, the incremental deposit could be accumulated in roughly three and a half years.