IV. Debt to Income limits
The global financial crisis has highlighted that households, like corporations and financial institutions, are highly indebted, particularly in some of the countries that have been at the center of this crisis. This raises many questions about the sustainability of households’ indebtedness and the interconnectedness of balance sheets. It also makes it necessary to consider policy actions to address this indebtedness. This note discusses the degree of household indebtedness, its implications, and a possible action to address household indebtedness, namely the implementation of debt to income limits.
A. Debt to Income Ratios
Households are highly indebted in many advanced countries, some of which have been at the center of the global financial crisis. Table 1 shows that the ratio of gross household debt to GDP ranged from 129 percent in Ireland to 50 percent in Italy in 2010 (IMF, 2011a). The ratio of gross household debt to GDP was 91 percent and 107 percent in the United States the United Kingdom, respectively, two countries that have been at the focal point of the crisis. The ratio of gross household debt to GDP was nearly 93 percent in Canada, among the highest in this group of countries. Tempering these high ratios is the fact that households’ debt is far less than households’ holdings of financial assets. The ratio of net household debt to GDP (i.e., debt minus financial assets) was minus 230 percent in the United States and minus 184 percent in the United Kingdom. Undoubtedly, the large household holdings of financial assets have implications about the ongoing debate about the role of collateral in the extension of credit.
|Households Gross Debt1||54.8||92.8||72.4||69.4||61.5||68.2||129.0||50.4||74.0||103.4||90.1||106.5||91.1|
|Households Net Debt12||−204.3||…||−129.2||−131.3||−130.4||−55.7||−60.2||−177.7||−231.5||−126.1||−73.8||−183.9||−230.1|
Most recent data divided by 2010 GDP.
Household net debt is calculated using financial assets from a country’s flow of funds.
Most recent data divided by 2010 GDP.
Household net debt is calculated using financial assets from a country’s flow of funds.
Figures published by the Banque de France confirm the high indebtedness of households in countries that were at the center of the crisis (Table 2). These figures, which rely on different sources than the ones used to compile the gross household debt to GDP in Table 1, show that the gross household debt to GDP were highest in the United States, the United Kingdom, and Spain. The figures for the gross household debt to disposable income (DI) were also highest in the United States, United Kingdom, and Spain. As of end-September 2010, the figures for gross household debt to DI reached 149 percent in the United States, 142 percent in the United Kingdom, and 126 percent in Spain. In the context of the global financial crisis, these figures would seem to suggest that a high ratio of gross household debt to DI could be a source of major source of risk to economic stability. As Mian and Sufi (2010) note, the increase in household leverage in the United States explains well the rise in mortgage defaults, and the subsequent fall in house prices and decline in durable goods consumption. Stated differently, the deleveraging of households therefore appears to have contributed to the sharp retrenchment in economic growth in the United States.
B. Risks Arising from Household Indebtedness
Households face increasing risks as their indebtedness rises. As the experience in the global financial crisis has shown, households that borrow too much, or borrow in excess of their ability to repay, could end up facing payment obligations that are unsustainable. Households may simply not have enough disposable income to meet their obligations and could also experience a sudden decline, or worse a loss of income, or an increase in essential spending needs, which would make it difficult, if not impossible, to meet their debt obligations. In the context of an economic crisis that results in an increase in unemployment and a decrease in wealth, including the value of homes, this could have severe consequences for households. In addition, changes in credit conditions, for example as a result of an increase in interest rates that leads to an increase in payments on variable-interest loans, or a credit crunch, would make it difficult to roll over debt obligations. The interaction of these risks could be particularly difficult to manage (Elul et al. 2010).
A generalized inability by households to service their debt could pose a systemic risk for the economy. As the crisis has shown, the inability of households to repay their obligations could jeopardize the health of financial institutions, and, in the case of extreme shocks, have systemic consequences (Dynan, 2009; Mayer et al., 2009). This could adversely impact the ability of financial institutions to extend credit, which could negatively affect businesses’ capacity to continue with their ongoing investment projects or to kick off new projects. Depending on the extent of these difficulties, the government may see its finances deteriorate as a result of a decline in tax revenue and possibly its balance sheet weaken because of the need to support financial institutions in difficulty. Such developments would, no doubt, have an adverse impact on economic growth. In this light, the key question is: Is there a policy that could limit the risks arising from a generalized problem of household indebtedness?
C. Debt to Income Limits
DTI limits could help reduce the risks associated with high household indebtedness, while lessening the procyclicality in lending. Implementing DTI limits could help reduce the households’ debt-service payment difficulties, or lessen the probability of default. In doing so, these limits could also help lower the likelihood of the materialization of the systemic risk arising from the generalized inability of households to service their debt. These limits could be preventive in nature. At the same time, DTI limits could contribute to lessen the procyclicality of lending. They could serve to smooth the credit dynamics in the context of the business cycles, by tightening them in the upswings to slow credit expansion and relaxing them in the downswings to spur credit growth.
DTI limits could take on different characteristics. As Chang (2010) and Crowne et al. (2011) note, DTI limits aim at establishing a maximum percentage of a household’s income for paying debt (both principal and interest) in any one year. While the DTI limits could include only a narrow definition of debt such as mortgage obligations, they could also take into account a broad definition of debt, including taxes, insurance fees, or even utilities. The DTI limits could take the form of front-end ratios, which establish the percentage of income for paying housing costs (principal and interest, hazard insurance premium, property taxes, and homeowners’ association fees). The DTI limits could also be back-end ratios, which set the percentage of gross income for paying all recurring debt payments (Galati and Moessner, 2011). Finally, DTI limits could be either discretionary or rules based subject to credit dynamics.
The effectiveness of DTI limits would depend on several factors. First, the effectiveness of DTI limits, as with any other tool to reduce the incidence of future crises, would depend on supervisory enforcement of the limits. This would require the issuance of guidelines, regulations or laws to lay the basis for the application of DTI limits, and the creation of a comprehensive database on household debt and income to track such limits. Second, the effectiveness of these limits would depend on how well financial institutions use them to assess the creditworthiness of borrowers, which, in the case of households, focuses on past debt payment record, indebtedness, income, wealth, and collateral (Tirole, 2006). While financial institutions may well face increased costs for including the DTI limits in the assessment of creditworthiness of households, they need to weigh this increase in costs against the potential decline in the costs of collecting late debt-service payments or going through bankruptcy procedures. Third, the effectiveness of such limits would also require the participation of all financial institutions that extend credit to households. Households should not have the ability to pick financial institutions with a view to avoiding the DTI limits.
Many countries worldwide have actively used DTI limits. More than two dozen countries, including some in Latin America, require financial institutions to use DTI limits as part of their assessment of creditworthiness of households. By way of example, the United States has long used DTI limits for conventional loans and mortgage loans insured by the Federal Housing Administration. In the case of conventional loans, the DTI limits have typically been 28/36 percent. The 28 percent is applied to the housing payment, and the 36 percent is applied to the housing payment plus recurring debt. Hong Kong SAR has put in place caps on debt service to income as a condition for household lending. As part of their efforts to improve financial institutions’ risk management, lower FX credit risk and protect borrowers, particularly in the low-income groups, Poland and Serbia have put in place differentiated debt-service to income limits by currency. The fact that other countries have not used these limits may well reflect the difficulty in determining household income, particularly in economies that still face high degree of informality, and the lack of consolidated household debt
Korea has made DTI limits part of its discretionary policy tools. In 2005, Korea adopted DTI limits as part of a package of policy actions for the real estate sector aimed at increasing the supply of housing, lowering housing prices, and restructuring the property tax system. In adopting this package, Korea also sought to limit speculation in the real estate sector. Korea initially set the DTI limits at 40 percent for married borrowers whose spouse already had a mortgage and for unmarried applicants 30 years or younger seeking houses in areas that had seen sharp house price increases (e.g., speculative areas). However, later Korea began to apply differentiated lending limits within a range of 40 percent to 60 percent depending on the size of the housing unit, lending amount, borrower’s credit rating, repayment method, interest rate, and evidence of income. After introducing the DTI limits in August 2005, Korea tightened the DTI limits on seven occasions and loosened them on one occasion.
A study shows that a tightening of the DTI limits in Korea has important results (Igan and Kang, forthcoming). Preliminary econometric results in this study show that a tightening of the DTI limits results in a slowdown or even a reversal of house price increases six months after the intervention. At the same time, a tightening of the DTI limits leads to a decline in transaction activities in the housing industry over many months following the intervention, particularly in metropolitan areas. The econometric results suggest that the tightening of the DTI limits has more of an impact on transaction activities than on house prices. The tightening of the DTI limits also results in some deceleration of the growth of household debt levels. Finally, the study concludes that a tightening of the DTI limits may be helpful to curb price expectations in the housing sector.
Still, the use of DTI limits requires further analysis. The determination of the DTI limits, including whether to apply them to credit lines or contingent liabilities, needs careful consideration. The interaction of DTI limits and other leverage ratios, including the debt to liquid net worth ratio, and collateral requires further analysis. In this context, it is necessary to consider whether households that provide significant collateral, say, in the form of liquid financial assets, to borrow should be subject to the same DTI limits as households that do not provide collateral or provide collateral that is not liquid. In addition, the definition of the debt and income used to determine the DTI limits also requires careful analysis. For instance, should the debt include only mortgages or all debt? Should income exclude extraordinary income in any one year? Should the determination of DTI limits make use of household debt and income for just one year or should it rely on debt and income over a number of years to smooth out years of extraordinary earnings?
High indebtedness of households could heighten systemic risk. High indebtedness jeopardizes the ability of households to service their debt, particularly in the face of major external shocks as observed in the global financial crisis, and makes balance sheets in the economy more fragile. This makes it necessary to consider, among other policies, the use of DTI limits. DTI limits aim at preventing the excessive indebtedness of households, and lessening of procyclicality in lending. As the experience of Korea indicates, limits on the DTI ratio—complemented with other measures—could serve as a credit cycle and countercyclical moderation tool. Kyung-Hwan and Man (2010) note that DTI limits, as with the loan-to-value rate, have proven to be an effective measure for cooling off-housing demand and containing housing price increases in this country.
Prepared by Carlos Medeiros. This note has benefited from comments by and conversations with Reinout De Brock, Diane Mendoza, Camilo Tovar, other members of the WHD Macroprudential Working Group, and colleagues throughout the Fund.