Abstract
This paper reviews policy tools that have been used and/or are available for policy makers in the region to lean against the wind and review relevant country experiences using them. The instruments examined include: (i) capital requirements, dynamic provisioning, and leverage ratios; (ii) liquidity requirements; (iii) debt-to-income ratios; (iv) loan-to-value ratios; (v) reserve requirements on bank liabilities (deposits and nondeposits); (vi) instruments to manage and limit systemic foreign exchange risk; and, finally, (vii) reserve requirements or taxes on capital inflows. Although the instruments analyzed are mainly microprudential in nature, appropriately calibrated over the financial cycle they may serve for macroprudential purposes.
V. Loan-to-Value Ratio Caps as a Macroprudential Tool
A. Introduction
Loan-to-value ratio (LTV) caps limit borrowers’ leverage in asset purchases. LTV restrictions set an upper limit to the amount of loan a borrower can take out from the lender to purchase an asset, which usually serves as the collateral of the loan. The restrictions—typically expressed as a percent of the value of the purchased asset or collateral—are often applied to the housing market, though they are also sometimes imposed in other asset markets (e.g., auto loans). In the housing market—where the asset prices are highly procyclical—the denominator of the restrictions often refers to the lower of the market value and the professional surveyor’s assessment value, which tends to fluctuate less along the cycle.
Such limits could be a useful policy tool to help moderate the asset and credit cycles, and mitigate risks to the financial system. While banks are likely to have risk management mechanisms in place that implicitly restrict LTVs, such internal control may become too loose during booms, thus giving rise to the need for an appropriate regulatory limit. LTV caps can create frictions on either leg of the cycle, which often features strong feedback loops between loan growth and asset price/demand dynamics. On the upturn, by dampening the former, a tight LTV cap could also put a drag on the latter. On the downturn, the presence of LTV limits during the preceding boom could reduce defaults and hence banking stress—for instance, in the housing market a lower LTV ratio would imply fewer incidences of negative home equity for any given decline in house price, and there is evidence that those households still with positive home equity are less likely to default (e.g., Bhutta et al., 2010, and Wong et al., 2004).2
B. LTV Rule Implementation3
Targeted focus and transparency are main advantages of LTV restrictions vis-à-vis other tools in addressing asset cycles. LTV rules have first-order influence on credit expansion in the targeted sectors, while leaving the economy-wide liquidity and cost of credit largely unaffected. In addition, since LTV restrictions bind only for new and highly leveraged (hence likely riskier) asset purchases, they entail few “collateral damages”—existing borrowers and less leveraged buyers are not directly affected. They also convey very simple and transparent signals to the public of the authorities’ concerns about asset and credit bubbles.
Implementation of LTV caps can take various forms, notably in a countercyclical fashion to lean against swings in credit supply and asset demand. In particular, many countries have recently either started imposing LTV caps or tightened the existing restrictions as part of their broader toolkit to manage the ongoing upswing.4 To yield sharper focus on specific pressure points in the market, LTV caps can also be applied differentially to loans of different characteristics—e.g., for mortgages, owner-occupied vs. buy-to-lets, high-end vs. lower-priced, “hot” vs. peripheral locations, local currency vs. FX mortgages, and longer duration vs. shorter-term loans. In some cases, there are active government-supported programs that provide mortgage insurance, with which borrowers are allowed a limited breach of the LTV caps (e.g., Canada, Hong Kong SAR).
And often LTV rules are complemented with other regulatory measures to address concerns beyond the degree of leverage. For example, LTV caps on mortgages are sometimes applied along with ceilings on debt servicing to income ratio (DTI) to limit risks from borrowers’ cash flow stress (e.g., Korea, Hong Kong SAR), and/or with higher stamp duty to discourage “flipping” and speculation (e.g., Singapore, Hong Kong SAR).
The use and effectiveness of LTV restrictions, however, is sometimes constrained by practical issues. As with other macroprudential tools, measurement difficulty and potential loopholes give rise to implementation challenges. For instance, high LTV loans may migrate to the parts of the financial sector that are beyond the authorities’ direct prudential oversight (including offshore banks). LTV rules may also be evaded through backdoor arrangements of additional loans disguised as personal loans from banks, or “top-ups”. The targeted nature of the tool—while an advantage—could also create unintended spillovers by pushing financial excess to nontargeted sectors. Moreover, there is not a well-developed analytical framework underpinning the calibration of LTV restriction in accordance with the market’s cyclical position—the recent rapid reversal of housing policy stance in Korea illustrates the difficulty faced by policymakers in determining an appropriate level of LTV limits.5 The relative quantitative arbitrariness could further compound the inherent tension between an adoption of LTV restrictions in the housing market and the socio-political agenda of promoting home ownership in some countries (including through mortgage interest tax deductibility). And from an efficiency perspective, too-strict LTV caps could unduly reduce welfare by erring too much on the side of ensuring stability at the expense of financial deepening.
In light of the constraints and also reflecting a preference against absolute limits, many regulators opt to discourage, rather than strictly rule out, high LTVs. In particular, instead of instituting a LTV cap, some countries apply higher capital charges or provisioning requirements to mortgage loans of higher LTVs (e.g., Norway, Switzerland, the UK, Spain, and Israel). And in some cases, tight LTV limits are applied to only mortgages that are used as collaterals of covered bonds (e.g., Peru, Germany, and Switzerland).
C. Existing Evidence on Effectiveness
Despite their prevalence, the empirical literature on the effectiveness of LTV restrictions is relatively limited, albeit growing. Gerlach and Peng (2005) show that following the introduction of housing LTV caps, credit expansion in Hong Kong SAR has become less sensitive to property prices, hence implying possibly less pronounced credit cycles. Similarly, in a panel of (mostly) advanced economies, Almeida et al. (2006) find that the sensitivity of housing prices and mortgage credit to income shocks is lower when the LTV limits are tighter. Meanwhile, Crowe et al. (forthcoming) find that maximum LTV limits are positively correlated with house price appreciation between 2000 and 2007 in a cross-section of countries, and Wong et al. (2011) show that banks in those countries that have explicit LTV restrictions seem more shielded from house price and macro developments. Using micro survey data, Igan and Kang (forthcoming) find that Korea’s LTV policy—along with many other measures—might have helped moderate the housing cycle by influencing households’ expectations of future house price increase, although the policy’s net social benefits are not assessed. However, the scattered pieces of favorable evidence notwithstanding, there has yet to be a full agreement on the tool’s effectiveness—of those authorities having employed LTV caps, some are convinced of the tool’s usefulness, while others report mixed analytical results, not least reflecting the empirical difficulty of identifying the various factors simultaneously at work (CGFS, 2010 May).
D. Recent Examples in the Americas
Canada
In Canada high-LTV mortgages are required by law to be guaranteed by mortgage insurance. The LTV limit on conventional mortgages is set at 80 percent,6 and lenders are required to obtain insurance against borrower default for mortgages with LTVs above this threshold. Even then, mortgage insurance—whether provided by the government agency (CMHC) or private companies—is allowed to back mortgages of LTVs only up to 95 percent.7 Reflecting the relatively conservative LTV restriction on conventional mortgages and the active government support in the mortgage insurance market, insured mortgages represent as much as 47 percent of total outstanding mortgage loans held by chartered banks.8
In a precautionary policy move, in April 2010 the Canadian government tightened the LTV limits for insured mortgages that are refinanced or are used for buy-to-let purchases, while keeping the LTV limits for conventional mortgages and for other insured mortgages unchanged. There are few overt signs of overheating in the housing market (e.g., house prices increased by a moderate 4 percent y/y in April 2010); but in light of the financial crisis experience and recognizing that the current record low interest rates have the potential of breeding excessive risk taking, the government took early actions to prevent undesirable trends from developing. Accordingly, to discourage home equity-financed consumption and promote larger buffers against any housing downturns, the LTV limit on refinanced insured mortgages was lowered from 95 to 90 percent. Also, to dampen speculative activity, the limit on insured mortgages for buy-to-let properties was reduced from 95 to 80 percent. As a complementary measure, the minimum DTI criterion was also tightened—to qualify for mortgage insurance, borrowers are now required to meet the income standards for a five-year fixed rate mortgage even if they choose a mortgage with a lower interest rate and/or a shorter term.
Canada: House Price Inflation (percent)
Citation: IMF Working Papers 2011, 159; 10.5089/9781455297726.001.A005
Source: Statistics CanadaIt has been announced that the LTV limit for refinanced insured mortgages will be again tightened in March 2011. Extending the April 2010 measures, the government announced in January that the LTV limit on refinancing of insured mortgages will be further lowered, to 85 percent. In another preemptive step to mitigate any consumption bonanza fueled by a housing boom, the government will also withdraw its existing insurance backing on non-amortizing home equity lines of credit. In addition, the maximum amortization period allowed for new insured mortgages will be reduced from 35 to 30 years, so that the cost of home purchase is better reflected in the borrower’s monthly payments.
Brazil
With the mortgage market still small (albeit rapidly growing), the fast expansion of consumer loans—particularly auto loans—is seen as a relatively more important source of vulnerability in Brazil. At 140 billion reais (or 4½ percent of GDP) and rising at about 50 percent y/y, auto loans account for nearly half of the 29 percent year-on-year growth of total non-earmarked loans to individuals. The recent lengthening in the average maturity of auto loans (to 19 months) and reduction in the lending spreads have added to the worries that the boom of auto loans has been in part driven by a laxer risk control by the lenders and could well prove to be unsustainable.
Brazil: Auto Loans
Citation: IMF Working Papers 2011, 159; 10.5089/9781455297726.001.A005
Source: Banco Central do Brasil.1 Refers to only credit operations with non-earmarked funds.In Brazil there has been long-standing capital regulation to discourage high-LTV auto loans; in early December 2010 the authorities tightened the regulation to further restrain those loans. Specifically, for any given maturity, the new rule stipulates a greater risk weight on loans that carry high LTVs. For instance, a risk weight of 150 percent (vs. 100 percent before the change) is now imposed on auto loans with LTVs higher than 80 percent for the 2 to 3-year tenor, or loans with LTVs higher than 70 percent for the 3 to 4-year tenor, or loans with LTVs higher than 60 percent for the 4 to 5-year tenor. Other regulatory measures introduced at the same time to contain credit growth include a heavier capital charge on long-duration payroll-deducted personal loans and a higher reserve requirement.
The macroeconomic implications of the new auto loan measure are qualitatively similar to those of tightening LTV limits on a booming housing market. In either the mortgage or auto loan market, rapid credit growth increases risks of excessive domestic demand, although the exact channels differ.9 To the extent that car prices are relatively insensitive to economic cycles, auto loan providers may be less exposed than mortgage lenders to corrections of the value of the underlying collaterals. Nevertheless, with car values depreciating quickly, providers of high-LTV and long-duration auto loans are likely to be significantly at risk from the borrowers’ repayment ability. As such, a sharp rise in credit in either the mortgage or auto loan market would similarly leave the lenders vulnerable to adverse macroeconomic shocks. Aimed to restrain the expansion of auto loans (especially the higher-risk ones), the recent measure by the Brazilian authorities could thus help lean against the domestic demand pressures and mitigate the underlying fragility of the financial sector.
In a tentative suggestion of its effectiveness, auto loan interest rates rose and domestic vehicle sales slowed immediately following the introduction of the new measure. The average auto loan interest rates increased by 2½ percentage points in the same month of the rule change, and the year-on-year change in the volume of domestic car sales was flat in January 2011, falling markedly from 24 percent in December 2010. A more complete assessment of the measure’s effectiveness, however, awaits further data, including on the volume and average maturity of auto loans.
E. Conclusion
Where the housing sector bears greater systemic consequences for the macroeconomy, LTV restriction on mortgages is more likely to be a useful macroprudential tool. In countries where the housing sector is viewed as particularly prone to boom-busts and closely tied to economic activity (due to, e.g., high share of household wealth in housing, prevalent practice of home equity withdrawal, large capital inflows to the housing sector), LTV restrictions could be a useful tool to reduce the amplitude of housing cycles and weaken their macroeconomic impacts. In many Latin American countries, the residential mortgage market is relatively shallow. And while some countries in the region already have LTV rules in place at least for the regular mortgage products (e.g., Chile, Colombia, El Salvador, Guatemala), the restrictions have not been typically adjusted along the cycles.10 Looking ahead, however, a very rapid mortgage market expansion—especially if fueled by the surge in capital inflows and/or accompanied with a fall in lending standard—would still give rise to significant risks to financial stability; by requiring lenders to hold larger collateral during the boom to buffer against the downturn, countercyclical LTV regulation could have an important prudential role to play in that context. Meanwhile, since mortgage LTV caps narrowly target the housing market, complementing this tool with wider measures to rein in financial excesses would be needed if the exuberance is widespread and beyond just the residential real estate.
Moreover, LTV rules could be a useful tool even in non-housing asset markets, as highlighted by the auto loan example in Brazil. Measures to disallow or discourage high-LTV loans in other asset markets (e.g., cars, commercial real estate) work in a similar way to mortgage LTV restrictions in moderating domestic demand cycles and helping to shield the lenders from economic downturns.
The appropriate level of LTV restriction depends on the specific structure and trends of the credit market. For instance, in countries where the lenders’ financial soundness is less strong (due to, e.g., reliance on wholesale or foreign funding, low profitability) or mortgage loans are of nonrecourse nature, there may be a more compelling case for a stricter LTV limit given the lenders’ greater vulnerability to shocks. While stronger signs of overheating likely warrant a steeper LTV tightening, abrupt and aggressive policy moves might lead to an excessive correction in the asset market.11 Adjusting the LTV caps in gradual increments, on the other hand, could yield a more efficient outcome as the authorities would be better able to assess the impact of the measure and the evolving market trends before proceeding to next steps.
Annex 1. Case Study of Hong Kong SAR
By way of a case study on Hong Kong SAR, this annex seeks to highlight some practical considerations relevant for the implementation of housing LTV rules. The choice of Hong Kong SAR owes to its long experience with LTV rules, and to the strong link between the housing market and the macroeconomy there.
Hong Kong SAR faces high volatility but is armed with relatively limited macroeconomic policy tools. Hong Kong SAR has a currency board arrangement and a small government, hence for macroeconomic management purposes monetary policy is absent and the fiscal tool is constrained. This provides a more prominent role to macroprudential measures. Besides its relatively small economic size and openness to international capital markets, Hong Kong SAR’s close tie to mainland China is also a key factor contributing to the economy’s volatility—on one hand, international investors’ desire to acquire financial exposures to China prompts large volumes of “proxy” investment flow to Hong Kong SAR; on the other hand, an increasing diversification of mainland Chinese’s wealth to Hong Kong SAR has helped make the territories’ housing market a speculation hotspot. Banks in Hong Kong SAR are typically highly capitalized, with low loan-to-deposit ratio. Residential mortgages represented about 27 percent of bank loans in 2010.
The LTV policy has a relatively long history in Hong Kong SAR. The restriction, at 70 percent, was first introduced in 1991 on a voluntary basis, followed by formal guidance in 1994. The denominator of the cap refers to the lower of the actual transaction price and professional surveyor’s valuation (the latter tends to be significantly less than the former during booms). The LTV cap is complemented by a DTI limit at 50–60 percent (with the upper limit applied to high earners) and guidance to banks urging against excessive property market exposures.12 The cap applies to both newly originated mortgages and refinancing, except for those refinancing cases involving negative home equity.13
Hong Kong Monetary Authority (HKMA) is responsible for the formulation and enforcement of the LTV cap. HKMA enjoys undiluted regulatory power as it is the sole prudential overseer of banks and mortgage products. While the LTV cap is not statutory, violations would result in HKMA questioning the bank’s risk management practice—a threat seen to be serious enough to make the cap essentially a rule in practice. Enforcement is done through on-site spot checks and off-site reviews. And HKMA has proved to be nimble at fine-tuning bank regulation to block loopholes—e.g., it has effectively discouraged illicit “top-ups” by moving quickly to limit the drawdown window and maturity of personal loans.14
The LTV cap has been actively managed in a countercyclical fashion. HKMA has made adjustments to the LTV rule over the years in response to housing sector developments, including a tightening of the cap in the run-up to the 1997–98 Asian crisis. More recently, in October 2009, against the backdrop of rapidly rising property prices, especially in the high-end segment, HKMA lowered the LTV cap to 60 percent for homes above HK$ million (or US$2.5 million, or about 6 times the average home price in Hong Kong SAR). In August 2010, HKMA broadened the tighter 60 percent cap to properties above HK$12 million and all buy-to-lets. As complementary measures, HKMA has also reduced the DTI limit to 50 percent for all borrowers, and required banks to grant loans only to those whose DTI would stay below 60 percent even if mortgage interest rates rise 2 percentage points.15 In November 2010, HKMA further tightened the LTV cap, to 50 percent for properties above HK$12 million and all buy-to-lets, and to 60 percent for properties between HK$8 million and HK$12 million.16
There is some early suggestive evidence that the new LTV measures might have helped slow mortgage credit growth and housing turnover, although their effects on housing prices seem less clear. Since the new measures, the weighted average of new loan LTVs has fallen—to the lowest level since at least 2001, when tracking of such data became available—probably reflecting a dropout of high LTV loans for high-end purchases (Figure A1-Chart 1). There is some suggestion that the measures might have also helped push down mortgage credit expansion and market turnover (Figure A1-Chart 2), with the targeted high-end segment particularly affected (Figure A1-Charts 3 and 4). Similarly, while the rise in average home prices has continued unabated (Figure A1-Chart 5), the relative increase in high-end home prices seems to have moderated somewhat following the new LTV rules (Chart 6). Needless to say, however, high volatility of the data and simultaneous developments of many other relevant factors suggest cautions in interpreting the outturns.17
Hong Kong SAR: Housing Sector Developments
Citation: IMF Working Papers 2011, 159; 10.5089/9781455297726.001.A005
Sources: Hong Kong Monetary Authority and Hong Kong Mortgage Corporation.1/ Due to data limitations, high-end transactions do not fully correspond to the market segment to which the tighter LTV applies.Annex 2. Recent Country Examples of LTV Restrictions
| Country | Dates of new Rules | Latest rules |
|---|---|---|
| Korea | Jul ‘09; Oct ‘09 | 40–50 percent for mortgages for the capital region; looser limits for (also DTI at 40–50 percent for the capital region) |
| Hungary | Mar ‘10 | 75 percent for local currency mortgages; 45–60 percent for FX mortgages |
| China | Apr’10; Jan ‘11 | 70 percent for large first homes (>90 sqm), and 40 percent for second homes |
| Norway | Mar ‘10 | 90 percent for all new mortgages |
| Sweden | Oct ‘10 | 85 percent for all new mortgages |
| Malaysia | Nov ‘10 | 70 percent for third homes |
| Hong Kong SAR | Oct ‘09; Aug’10; Nov ‘10 | 50–60 percent for high-end purchase (>HK$8 million) and 50 percent for buy-to-lets |
| (also tighter DTI at 50 percent, and bank stress test required on interest rate rise) | ||
| India | Dec ‘10 | 80 percent for higher-priced homes (>Rs 20 lakh); 90 percent for others (also higher risk weight for large mortgage loans (>Rs 75 lakh)) |
| Brazil | Dec ‘10 | Higher capital charges on longer-duration, higher-LTV auto loans |
| Singapore | Feb ‘10; Aug‘10; | 80 percent for buyers with no existing mortgages; 60 percent for other (also higher stamp duty for quickly resold properties) |
| Thailand | Jan ‘11 | 95 percent for low-rise homes; 90 percent for most condos (< 10 million bahts) |
| Canada | Apr’10; Mar ‘11 | 85 percent for refinancing of insured mortgages; 80 percent for buy-to-(also shorter maximum amortization period) |
| Country | Dates of new Rules | Latest rules |
|---|---|---|
| Korea | Jul ‘09; Oct ‘09 | 40–50 percent for mortgages for the capital region; looser limits for (also DTI at 40–50 percent for the capital region) |
| Hungary | Mar ‘10 | 75 percent for local currency mortgages; 45–60 percent for FX mortgages |
| China | Apr’10; Jan ‘11 | 70 percent for large first homes (>90 sqm), and 40 percent for second homes |
| Norway | Mar ‘10 | 90 percent for all new mortgages |
| Sweden | Oct ‘10 | 85 percent for all new mortgages |
| Malaysia | Nov ‘10 | 70 percent for third homes |
| Hong Kong SAR | Oct ‘09; Aug’10; Nov ‘10 | 50–60 percent for high-end purchase (>HK$8 million) and 50 percent for buy-to-lets |
| (also tighter DTI at 50 percent, and bank stress test required on interest rate rise) | ||
| India | Dec ‘10 | 80 percent for higher-priced homes (>Rs 20 lakh); 90 percent for others (also higher risk weight for large mortgage loans (>Rs 75 lakh)) |
| Brazil | Dec ‘10 | Higher capital charges on longer-duration, higher-LTV auto loans |
| Singapore | Feb ‘10; Aug‘10; | 80 percent for buyers with no existing mortgages; 60 percent for other (also higher stamp duty for quickly resold properties) |
| Thailand | Jan ‘11 | 95 percent for low-rise homes; 90 percent for most condos (< 10 million bahts) |
| Canada | Apr’10; Mar ‘11 | 85 percent for refinancing of insured mortgages; 80 percent for buy-to-(also shorter maximum amortization period) |
Prepared by Man Keung Tang. This note has benefited from comments from various members of the group.
See also the simulation of a financial accelerator model in IMF (2008), which shows that macro procyclicality increases with LTV ratio.
Annex 1 on a case study of Hong Kong SAR highlights some practical considerations in LTV rule implementation, and also discusses the effects of the recent LTV tightening there.
See Annex 2 for some recent examples of LTV rule application.
The Korean authorities tightened LTV restrictions in the second half of 2009 to stem the fast rising housing prices; but as the housing cycle turned soon after, the authorities reversed policy stance in August 2010, introducing stimulating measures to support the cooling housing market (e.g., suspending DTI limits for owner-occupied properties until March 2011, subsidizing first-time home buyers, and reducing transaction taxes on some properties).
The limit was raised from 75 percent in April 2007.
The limit was lowered from 100 percent in October 2008.
Even most privately provided mortgage insurances are backed by the government (subject to a deductible equal to 10 percent of the loan value).
The effect of rapid mortgage credit growth on domestic demand may operate more through the liquidity (e.g., home equity withdrawal) and wealth (higher asset prices drive stronger consumption) channels. On the other hand, the effect of rapid auto loan growth on domestic demand may be more a result of a direct increase in net car purchases (as car financing becomes more abundant) and greater liquidity (car buyers have more free cash flows for non-car purchases).
One important exception is Chile, where the LTV requirements were lowered in 2009 for highly rated banks.
This is suggested by the recent experience in Korea, which saw a sharp reversal in the housing market dynamics following the tightening in LTV policy.
It is “recommended” that each bank keeps mortgage loans below 40 percent of its total loans.
As in some other countries (e.g., Canada), borrowers are allowed to exceed the LTV cap by a limited margin if they purchase mortgage insurance. In both Hong Kong SAR and Canada, government agencies are key providers of mortgage insurance (HKMC and CMHC, respectively). In Hong Kong SAR, LTV was allowed to go up to 90 percent if accompanied by mortgage insurance (vs. 70 percent without). Standard premium on insurance provided by HKMC is about 3 percent on total loan value for mortgages with 90 percent LTV. In comparison, in Canada LTV for new mortgages is allowed to go up to 95 percent with purchase of mortgage insurance (vs. 80 percent without). Standard premium of insurance provided by CMHC is 2 percent on total loan value for mortgages with 90 percent LTV and 2.75 percent for those with 95 percent LTV. In both places, the insurance premium can be amortized over the life of the mortgage loan.
Specifically, HKMA disallows personal loans to be available before the borrower closes any pending property transaction, and requires personal loans to be fully repaid within a period much shorter than the typical maturity of mortgage loans.
Most Hong Kong SAR homebuyers opt for variable mortgage interest rates due to their tendency to resell quickly.
Also in Nov 2010, the authorities imposed stricter restriction on the use of mortgage insurance to bypass the LTV cap (HKMC suspended its provision of mortgage insurance for properties above HK$6.8 million).
Some believe that strong demand from mainland Chinese was a key factor boosting the prices of high-end properties in Hong Kong SAR. As mainland Chinese buyers do not usually take out mortgage loans from Hong Kong SAR banks, the LTV measures might not have directly affected their demand.
