DEBT AND FINANCIAL TROUBLE: evidence from Denmark
12 Feb 2020
About 5 percent of all Danes are late with paying the dues on their loans at any point in time. This rate is surprisingly stable across time and moved only little during the recent financial crisis suggesting that financial trouble is not merely related to being affected by unemployment or other unfortunate events. One hypothesis is that differences in financial behavior can explain why some people end up in financial trouble while others do not. We test this hypothesis by examining whether financial behavior is shared within the family.
We leverage a new administrative registry data set provided by the Danish Tax Agency. The data set holds information about the universe of loan accounts held by Danish residents in Danish banks, covering more than 31 million personal loan accounts for the period 2004-2011. The data document loan defaults, measured as being more than 60 days late with payments on the loan. The loan data are linked to the population register, which reveals the identity of parents.
Figure 1 shows the propensity to be in default in 2011 by age for individuals whose parents are also in default in 2011 and for individuals whose parents are not in default in 2011. The figure shows a striking pattern. The propensity to be in financial trouble is strongly correlated across generations. At age 30, individuals with parents in default are more than four times as likely to be in default as individuals whose parents are not in default. Moreover, the intergenerational correlation is apparent soon after children turn 18 years old where they become legally eligible to establish debt. The default rates for both groups increase until the late twenties after which they stabilize and remain almost constant at 22-23 percent for individuals with parents in default and at 4-5 percent for individuals with parents not in default.
Notes: The figure shows the mean default rate surrounded by 95% CIs for each age group in 2011. Standard errors are clustered at the child level. Each age group is categorized into two groups according to parental default in 2011. An individual is defined as being in default if having at least one delinquent loan at the end of the year.
The pattern documented in Figure 1 is stable across time, across levels of loan balances, across parental income levels, and across levels of cognitive ability as proxied by the middle school grade point average.
The large differences in behavior imply that there are systematic differences in the risk that banks face when granting loans to people. By analyzing the interest setting of banks, we find that financial institutions are unable to fully price in the systematic risk of default related to family background. Figure 2 shows the future default probability by loan specific interest rate and parental default. To construct the graph, we have selected out all loans of persons who were not in default on any loan in 2004 and divided them into two groups dependent on whether the parents are in default or not in 2004. We then follow the loans of the individuals and compute the share of the loans that become delinquent at some time during the period 2005-2011. This is displayed as a function of the interest rate on the loans in 2004. We divide all the loans into one percentage point interest rate intervals and compute for each group of loans, and conditional on parental default, the average future default rate on the loans.
For the case of parents not in default, Figure 2 shows that the future average default rate increases gradually from 0.5 percent to 3.5 percent when going from loans with an interest rate of 2 percent to an interest of 20 percent, consistent with banks being able to predict delinquency when setting interest rates. We also obtain a clear increasing relationship between the interest rate and the future default rate for the loans of individuals where parents are in default. However, this relationship lies considerably higher in the diagram. At each level of interest rate, the future default rate is significantly higher if parents were in default in 2004. For example, for loans with an interest rate of 5 percent the probability of default within the next seven years is 0.5 percent if the parents are not in default in 2004 but 1.75 percent if parents are in default. Banks are thus unable to fully account for the intergenerational relationship in default propensities when setting the interest rates, and with a difference in the range of 0.7-4.7 percentage points across the interest rates levels, the effect not accounted for is quite large.
Figure 2. Future default probability by loan specific interest rate and parental default
Notes: The figure shows the average loan-level default rate in 2005-2011, along with 95% CIs, by loan specific interest rates in 2004, binned into one-percentage point interest rate intervals. The figure considers loan accounts of individuals who are not in default on any loan in 2004. Loan accounts are grouped by the individual-level default status of the debtor's parent in 2004. A loan is classified as becoming delinquent if the individual has defaulted on loan payments in any of the years 2005-2011.
This result points to the existence of an interest rate externality in the market for personal loans. In isolation, the results suggest that less debtor-friendly bankruptcy laws are sensible. However, this should be balanced against social insurance benefits from a more debtor-friendly system. Moreover, if the propensity to end up in financial trouble is high for some people because they are financially illiterate or have self-control problems then appropriate policy responses might be to provide financial training, prevent high-interest loans or to set limits on loan balances.
“Financial Trouble Across Generations: Evidence from the Universe of Personal Loans in Denmark” by Claus Thustrup Kreiner, Søren Leth-Petersen, and Louise Willerslev-Olsen is published in the January 2020 issue of The Economic Journal.
Professor of Economics | University of Copenhagen | +45-35-32-30-20 | firstname.lastname@example.org
Professor | University of Copenhagen | +45 35323084 | Soren.Leth-Petersen@econ.ku.dk
University of Copenhagen