Comments to "The Determinants of Household Debt Default"

Transcription

1 Comments to "The Determinants of Household Debt Default" Félix Villatoro Superintendence of Banks and Financial Institutions (SBIF) September 2010 Félix Villatoro (Institute) Determinants of Household Default 09/10 1 / 12

2 Roadmap Main results Regarding the EFH Results and comments Final Remarks Félix Villatoro (Institute) Determinants of Household Default 09/10 2 / 12

3 Main Results Mortgage default a ected by level of income, but not by LTV. Consumer debt default a ected by level of income and (sometimes) DSR. Threshold default probabilities consistent with F-RT (2009). Larger portions of debt are from borrowers that are less vulnerable to negative shocks ) relatively robust nancial system. Félix Villatoro (Institute) Determinants of Household Default 09/10 3 / 12

4 The EFH Rich dataset regarding households characteristics. Data is self-reported. National representativeness achieved through the use of expansion factors. Félix Villatoro (Institute) Determinants of Household Default 09/10 4 / 12

5 Methodology Probit models, with self-selection bias adjustment. For consumer debt only those w/o mortgage debt are considered. How many households are left out? Are they similar? Will change in EFH (2009). SE are estimated by boostrapping. Little clarity regarding signi cance levels used. Félix Villatoro (Institute) Determinants of Household Default 09/10 5 / 12

6 Mortgage Debt Default Mortgage debt default is negatively a ected by: level of income; education. Economic signi cance? Default not a ected by: sex; marital status; age; number of income earners; DSR and LTV. Implies that level of equity invested in property is not meaningful when making default decisions. However, international and Chilean evidence show that LTV does a ect probability of default and severity of loss given default. In fact, regulators in EEUU and Australia have proposed to adjust risk weights according to LTV. Perhaps won t change in 2009 survey. Challenge/limitation for self-reported data. Félix Villatoro (Institute) Determinants of Household Default 09/10 6 / 12

7 In uence of LTV on Mortgage Debt Default Relationship between LTV and PD, LGD for Chile. Huerta and Pacheco (2009) study the risk pro le por mortgage debt in the chilean banking system. Data from 1999 to 2007 from a 3MM debtors universe. LTV turns out to be signi cant in terms of PD and LGD. Félix Villatoro (Institute) Determinants of Household Default 09/10 7 / 12

8 In uence of LTV on Mortgage Debt Default Relationship between LTV and PD, LGD for Chile. Source: Huerta and Pacheco (2009) LTV (%) Sample Size PD (%) LGD (%) [0 40) 86, [40 60) 52, [60 70) 15, [70 80) 5, [80 90) 2, , Total 163, Félix Villatoro (Institute) Determinants of Household Default 09/10 8 / 12

9 Consumer Debt Default Consumer debt default is a ected by: level of income; bank account; age; number of income earners ("wrong" sign); and DSR (for some datasets). Nice rationalization for "number of income earners sign". Should be checked with the EFH data. Economic signi cance? Default not a ected by: level of education; sex; and marital status. Odd since education is highly correlated with income. It would be interesting to segment sample by labor market status (employed versus unemployed) and a liation to unemployment insurance (presumably would a ect LGD). Félix Villatoro (Institute) Determinants of Household Default 09/10 9 / 12

10 Consumer Debt Default Félix Villatoro (Institute) Determinants of Household Default 09/10 10 / 12

11 Some Final Remarks The use of surveys such as the EFH are a valuable tool since they: Provide a rich characterization of households; Help to identify key variables that in uence default decisions. Nevertheless, in order to have robust and reliable results, improvements should be made: Measurement of property value. Moreover, a richer set of questions regarding employment characteristics and history would be valuable. Although nancial system as a whole may seem robust, this doesn t necessarily apply to individual institutions. Is there enough data to identify nancial institutions involved? Félix Villatoro (Institute) Determinants of Household Default 09/10 11 / 12

12 References 1 Huerta, S. and Pacheco, D. (2009), "Estimación de Ponderadores de Riesgo para la Cartera Hipotecaria Residencial del Sistema Bancario en Chile", Documento SBIF. Preliminary version available at: Félix Villatoro (Institute) Determinants of Household Default 09/10 12 / 12