Increased risk-taking by youngest mortgage borrowers

Driven by appreciating house prices it has become increasingly difficult for starters to enter the housing market. We find that younger people increased their risk-taking to be able to buy a house. This increased risk-taking did not translate in an increased share in arrears for the youngest age group, but there might be a substantial tail risk in this group.

Difficult housing market for starters

In recent years it has been increasingly difficult for starters to enter the housing market. Many starters are willing to take high risks to have a chance to buy a house. About one-third of the starters hide their student debt when applying for a mortgage, or plan to do so. This is twice as much as two years before when the percentage was 15% [1]. Furthermore, the total outstanding student debt has increased from 12.7 billion in 2015 to 24.4 billion in 2021 [2].

Additionally, many starters take extra consumer loans or starter loans to finance the purchase of a house. However, this problem seems to weaken as lending criteria became more stringent. AFM finds that by using these means, 20% to 40% of the starters get a higher mortgage than what is strictly deemed appropriate. This does not necessarily mean over-crediting as lenders may deviate from the maximum allowed loan, as higher loans are allowed for example to finance energy saving measures [3].

Vulnerability of starters

The high number of mortgages above the appropriate amount for starters raises concerns from AFM, as starters are more vulnerable. AFM mentions multiple reasons to highlight the vulnerability of starters. Starters often get (close to) the maximum allowed loan. They also have a flex or temporary employment contract relatively often, which gives more income uncertainty. Furthermore, starters have a smaller capital buffer. Another reason is that about half of the starters are two-income households. If the paying capacity depends on both incomes, this poses a risk in case of illness or separation. Additionally, starters can get extra starters loans, increasing their monthly payments. Moreover, there is extra room for sustainability investments in the house when taking a mortgage. Another growing concern is that student debt has become more common (and higher) for starters. This student debt is also increasingly hidden when applying for a mortgage. Finally, many starters likely have lease obligations. Mostly lease obligations are neglected when applying for a mortgage [4]. Although all of these points apply to starters, not all of the points apply exclusively to starters.

Data

If the higher risk-taking in the lowest age category translates into higher arrears, this could cause losses for mortgage investors. In this article, we examine if this increased risk for starters has translated into increased arrears in the lower age borrower group. For the analysis we use loan-level data of Dutch mortgages from European DataWarehouse, which consists of ECB eligible RMBS deals. It is important to stress that the goal of this article is not to prove any causalities as the risk factors mentioned above cannot all be observed from the data. Instead, this article aims to provide insights in differences in arrears between age groups over time.

In their publications AFM defines starters as follows:

Households that took out a new mortgage and didn’t have a mortgage in the year before;
The primary wage-earner is 35 years old or younger;
The primary wage-earner (and possible partner) are employees, so not self employed [5].

The data does not allow to distinguish by new borrowers as this field is poorly populated, and using the field would harm the sample size. Moreover, we don’t filter on employment type because the employment type distribution of each age group is fairly constant over time, and filtering on the employment type does not change the conclusions. Therefore, we only segment the data in different age groups.

Risk-taking by age group

First, we look at observable risk metrics such as the original DTI and original LTV trend by age group, to observe cross sectional differences in risk-taking between age groups over time. Figure 1 shows that the lower age group has traditionally been the group with the highest DTI. Furthermore there has been an increasing trend in the relative DTI level in the lower age group, indicating increased risk-taking in the lower age category. Income data of the highest age bucket may not be accurate as pension income may be reported differently.

Figure 1: DTI by age group

Figure 1: DTI by age group. Source: LoanClear, European DataWarehouse

Similarly the lowest age group has the highest LTV traditionally, as shown in Figure 2. The maximum allowed LTV has decreased over the years, resulting in lower LTVs over the years and less differences between the age groups. However, the percentage point difference between the lowest age group and the average is more than 8% in 2021, the highest observed difference since 2014.

Figure 2: LTV by age group

Figure 2: LTV by age group. Source: LoanClear, European DataWarehouse

The findings of Figure 1 and Figure 2 indicate a higher, and increased relative risk-taking in the lower age group. The other risk components mentioned by AFM can’t be observed from the data, but it is likely that on top of the high DTI and LTV ratios, there is some hidden risk residing in the lower age group.

Arrears by age group

To investigate whether the extra risk is reflected by higher arrears, Figure 3 shows the arrears per age group over time. Most age groups follow the same trend. The general trend has been decreasing, caused by the favorable economic environment and tighter underwriting criteria. The age groups Age <= 35, 35 < Age <= 45 and 45 < Age <= 55 are largely at the same level, the age group 55 < Age <= 65 is mostly a bit lower and the group Age > 65 is always lowest. Also, the difference between the age groups has been decreasing. Arrears in the highest age group have decreased the least compared to the other groups, but come from a substantially lower starting point and have been lowest in all periods. Also, during the corona period, arrears levels in this group remained rather stable. This seems logical as this group consists of mostly pensioners who’s incomes were least affected [6].

Figure 3: Arrears by age group

Figure 3: Arrears by age group. Source: LoanClear, European DataWarehouse

To give a better insight, Figure 4 shows the contribution of each age group to the total arrears, normalized to account for unequal group balances. The contribution of most groups has been fairly stable over time. Most noteworthy is the increase of the contribution of the highest age group gradually over the whole period. Furthermore, in 2021-Q3, there has been an increase in the contribution of the lowest age group. However, as the arrears data is a bit noisy it is to be seen if this trend continues.

Figure 4: Arrears composition by age group, normalized to account for unequal balances

Figure 4: Arrears composition by age group, normalized to account for unequal balances. Source: LoanClear, European DataWarehouse

Tail risk in youngest age group

We find that the younger age group has increased their risk-taking over time. This risk-taking, however, has not yet resulted in an increased representation of the lowest age group in the total arrears. To investigate the potential effect of an economic downturn, we analysed the debt service ratio under a stressed scenario.

Figure 5: Percentage of loans where the Debt Service Ratio exceeds the allowed limit

Figure 5: Percentage of loans where the Debt Service Ratio exceeds the allowed limit. Source: LoanClear, European DataWarehouse

As a base case, in 2021-Q3, 22.6% of the youngest age group had a debt service ratio exceeding the maximum allowed. In a stress scenario where a 10% haircut on the borrowers’ income is applied more than 50% of the youngest age group has a debt service ratio that exceeds the maximum allowed. This is siginificantly more than the average of the other age groups which is limited to 10.7% in 2021-Q3 in the stressed scenario. Although the riskier behavior of young people has not translated into higher arrears yet, there might be a significant tail risk for this group.