companydata.dkBankruptcy analysis
Every quarter we test the Company Data Score against reality: can the score rank-order companies that later went bankrupt against those that survived? Here are the latest results.
Last updated: Q3 2026
The share of companies in the sample that subsequently went bankrupt, broken down by rating tier. The lower the rating, the higher the bankruptcy rate — as expected.
How many times higher the bankruptcy rate is in each score range compared with the sample average. These are relative separation figures, not absolute probabilities of default.
The same model drives our credit recommendation. Measured on the sample ahead of the bankruptcies:
- 79% of the companies that later went bankrupt were advised against credit ahead of the bankruptcy.
- Bankrupt companies made up 29% of the sample but only 1.8% of the total recommended credit volume.
The AUC for each individual scoring category, computed on its own. Shows which signals contribute most.
- Cases: companies that went bankrupt within the last 24 months, plus companies currently in bankruptcy proceedings. Each case is scored on data available before the distress.
- Control group: randomly selected active companies, scored as of today.
- Both groups are scored status-neutrally, so a company’s own bankruptcy status cannot leak into its score.
- AUC is the Mann-Whitney probability that a random surviving company scores higher than a random bankrupt one.
- The figures show relative separation in a sample with an artificially high bankruptcy share — not absolute probabilities of default for individual companies.
- Bankruptcy dates are currently approximated from each company’s cessation date until enough decree dates from Statstidende have accumulated.
- Owner information on dissolved companies can be incomplete, which conservatively weakens the management-track-record signal for cases.
The Company Data Score is available for every Danish company.