Macro Stress Testing of Danish Households

This chapter analyses the financial vulnerability of the Danish household sector on the basis of microdata and presents a method to simulate the effect of higher unemployment and interest-rate increases on the households' ability to service their debt. This method is part of Danmarks Nationalbank's overall stress testing project, cf. the chapter on macro stress testing of the financial system.

The analysis shows that even in the event of strong increases in unemployment in combination with interest-rate increases the household sector will not inflict losses on the banking sector that might threaten financial stability.

Households with relatively high incomes account for most of the household sector's debt. Only a small proportion of the financial institutions' lending to households is concentrated in financially vulnerable households, and by far the largest share is mortgage-credit debt, which is collateralised to a high degree.

The results presented in this chapter are sensitive to changes in the assumptions on which the analysis is based. In view of data limitations, the assets of the households are excluded from the analysis. These assets can serve as a buffer and thereby protect the financial institutions against losses.

INTRODUCTION

The purpose of the analysis is to quantify the debt raised by financially vulnerable households, and to analyse the effect of higher unemployment and interest-rate increases. Due to the scale of calculation of the scenario-determined simulation experiments, the calculations are based on a 10 per cent random sample of data on all Danish households.

The data basis is the income statistics from Statistics Denmark. The analysis is prepared on the basis of data for the 2005 fiscal year. The unit in the applied database is the household^[1] defined as one or more individuals residing at the same address, and who have certain mutual relations.

Since only employed persons risk losing their jobs, only wage-earner households (with no children over the age of 18 living at home) are included. The descriptive statistics for the data material applied are shown in Table 5.

DEFINITION OF FINANCIALLY VULNERABLE HOUSEHOLDS

Definitions of when a household has financial difficulties are hard to establish. There is no data available on the private consumption, including housing occupancy, of the individual households.

A measure of the financial margin of households^[2] is defined by Norges Bank and Sveriges Riksbank in similar analyses. The financial margin is compiled as the households' disposable income less a measure of the basic cost of living and housing costs, including maintenance, etc. of the home and servicing of debt. Norges Bank's analysis includes repayments on loans, which are excluded from Sveriges Riksbank's analysis.

Below, a budget method is also used to define the financial margin of Danish households, cf. Table 6. A financial margin of zero should not be perceived as a poverty line. The financial margin serves to indicate whether a household, with a given amount at its disposal (the household's disposable income), is able to maintain a basic level of consumption while also paying housing costs, excluding repayments on loans. A positive financial margin indicates that the household has the financial scope for consumption beyond the basic level, or e.g. for savings or investments. A household with a negative financial margin is classified as financially vulnerable.

Consumption budgets
Forbrugerinformationen's^[3] household budget (standard budget) provides general information on what is characterised as a normal, fairly average, standard of living in Denmark. The standard budget thus exceeds a basic level of consumption. The breakdown of consumption for a discount budget is the same as for the standard budget, but all purchases are made in discount stores.

The Centre for Alternative Social Analysis, CASA, has defined a basic budget that is based on Forbrugerinformationen's standard budget.^[4] The basic budget does not include savings for consumer durables, or spending on holidays and leisure activities. On the other hand, the basic budget does not take into account that households may shop in discount stores. The differences between the three budgets are shown in Chart 41.

In this analysis, the basic budget will be applied as an expression of a standardised consumption budget for Danish households. In order to analyse the sensitivity to choice of consumption budget, the two other budgets are included in the analysis below. In the sensitivity analysis, high-income households are assumed to apply the standard budget, while medium-income households are allocated the discount budget, and low-income households retain the basic budget. The analysis does not take into account that the households can adjust their consumption if their finances deteriorate. The annual standard, discount and basic budgets by household category are shown in Table 7.

In addition, the households' housing costs from Statistics Denmark's publication Consumption Survey 2003-2005^[5] are taken into account. Repayments of housing loans or other debt are thus not included.

METHOD FOR MACRO STRESS TESTING OF DANISH HOUSEHOLDS

The finances of the households are influenced by many factors. If income is no longer earned, or interest expenses increase, this can have major financial consequences for the individual household. Social events such as a divorce or illness also have a number of financial consequences. The analysis below focuses on changes in the households' ability to service their debt, in order to determine the impact on financial stability.

Choice of scenarios
Partial scenarios comprising higher unemployment and interest-rate increases are modelled, cf. Box 11. The scenarios generally correspond to those prepared by the IMF in connection with its FSAP for Denmark in 2006.^[6] The consequence of an increase in unemployment by 1-5 percentage points and in interest rates by 1-4 percentage points are analysed. Since the early 1970s, the maximum annual increase in unemployment has been almost 3 percentage points (1974-75), while the equivalent maximum annual increase in the average bond yield has been almost 3.5 percentage points (1973-74).^[7] Scenarios comprise combinations of increased unemployment and higher interest rates as well. Historically, there are no indications of correlation between 1-year changes in unemployment and interest rates. Nevertheless, there have been periods in which unemployment and interest rates rose simultaneously. In the period 1980-82, unemployment rose by almost 3.5 percentage points, and the average rate of interest by just over 3 percentage points. In the specified scenarios, second-round effects of changed behaviour among the financial institutions or households are not taken into account.

Results
From the viewpoint of financial stability, it is relevant to quantify the debt raised by households with limited financial scope. If the households are no longer able to service their debt, the financial institutions may lose the loan amounts less any collateral pledged. Most of the households' debt is raised by households with relatively high incomes. The debt of working households with disposable incomes in the 5th (upper) quintile constitutes 33 per cent of the total debt, while working households with disposable incomes in the 1st (lower) quintile account for almost 8 per cent of the total debt, cf. Chart 42. In the 2nd to 4th quintiles, households with high incomes have relatively more debt than low-income households, since the average debt burden increases with income.

In the analysis, households with financial problems are limited to households with a negative financial margin. A negative financial margin implies that the household is unable to pay both the sum of its housing costs and consumption equivalent to the budgets stated in Table 7.

As a baseline, just over 4 per cent of the total debt of the working households has been raised by households with negative financial margins. For some of the debt, the home is pledged as collateral, since three quarters of the total debt is mortgage-credit debt, cf. Table 5. In addition, any financial wealth will also serve as a buffer against losses on lending to the households. This is not taken into account in the analysis, and all other things being equal it moderates the potential losses of the households' creditors.

Each scenario comprises 500 simulations^[8], where each iteration recalculates which working households become unemployed and which homeowners have fixed or adjustable-rate housing loans. The simulations are static in that housing and consumption costs remain constant and that any reserves such as financial or real wealth are not used. For various scenarios, Table 8 shows the proportion of the debt that is raised by households with negative financial margin.

In this analysis, the households are generally not particularly exposed to increased unemployment. Even if unemployment rises by 5 percentage points, households with a negative financial margin will account for only 5 per cent of the total debt. One of the reasons for the modest increase in the number of financially vulnerable households and their debt in the event of large increases in unemployment is the size of unemployment benefits compared to the tight basic budget. An isolated increase in interest rates by 2 percentage points would increase the proportion of the total debt that is raised by financially vulnerable households by almost 2 percentage points compared with the baseline. Doubling this interest-rate rise will increase the debt ratio of the financially vulnerable households from approximately 4 per cent to almost 9 per cent. In the most extreme scenario with an increase in unemployment by 5 percentage points and in interest rates by 4 percentage points, households with a negative financial margin have raised just over 9 per cent of the total debt.

The results in Table 8 are sensitive to the choice of consumption budget. Applying the standard budget to households in the 4th and 5th income quintiles, and the discount budget to those in the 2nd and 3rd income quintiles, the financial margin of more households becomes negative if unemployment rises. The baseline therefore differs significantly from Table 8. With the change in consumption, households with a negative financial margin as the baseline will have 15.5 per cent of the total debt of working households. An isolated increase in unemployment by 3 percentage points would increase the debt ratio of financially vulnerable households to 16.5 per cent. For each percentage point that the interest rate is raised in the sensitivity analysis, the proportion of the total debt that is attributable to financially vulnerable households increases by approximately 4 percentage points.

In its calculations of the financial margins of Norwegian households, Norges Bank finds that raising the lending rate by 2 percentage points increases the debt of households with a negative financial margin from 16 to 22 per cent of the total debt.^[9] Sveriges Riksbank has equivalently found that the debt of financially vulnerable households increases from 5.6 per cent to 7.2 per cent on an increase in interest rates by 3 percentage points. The debt ratio increases from 5.6 per cent to 6.3 per cent on an isolated rise in unemployment by 3 percentage points.^[10] The analyses are not fully comparable with the results in this chapter since they are based on different assumptions. For example, everyone with debt is included in the Swedish analysis (whether employed or not), and the unemployment benefit system and percentage of households with debt at adjustable interest rates also differ in Sweden and Denmark. The analysis of Danish households in this chapter only includes people in employment. All other things being equal, this implies an underestimation of the total interest-rate exposure.