Challenges in Developing Risk Models for Crowdlending Platforms

In the digital financial environment, crowdlending is becoming one of the key forms of alternative financing, as it enables direct connection between investors and borrowers and promotes democratisation of access to capital. The diploma thesis 'Challenges in Developing Risk Models for Crowdlending Platforms' addresses the problem of the effectiveness of traditional credit risk assessment models for small and start-up companies and explores possibilities for their modernisation by incorporating non-financial indicators and the associated challenges. The theoretical part of the research analyses the development of crowdlending as an integral part of digital technologies and presents traditional approaches to measuring credit risk (Altman Z-score, PD/LGD/EAD models), whilst additionally demonstrating their limited applicability for companies without extensive financial history. The results show that classical models often incorrectly classify reliable borrowers as risky.

Problem

In the digital financial environment, crowdlending is becoming one of the key forms of alternative financing, as it enables direct connection between investors and borrowers and promotes democratisation of access to capital. The diploma thesis 'Challenges in Developing Risk Models for Crowdlending Platforms' addresses the problem of the effectiveness of traditional credit risk assessment models for small and start-up companies and explores possibilities for their modernisation by incorporating non-financial indicators and the associated challenges.

Solution and Method

The theoretical part of the research analyses the development of crowdlending as an integral part of digital technologies and presents traditional approaches to measuring credit risk (Altman Z-score, PD/LGD/EAD models), whilst additionally demonstrating their limited applicability for companies without extensive financial history. The empirical part includes an analysis that verifies the accuracy of models on data from a crowdlending platform.

Results

The results show that classical models often incorrectly classify reliable borrowers as risky. To address the inefficiency, a set of seven 'soft information' indicators was incorporated into the Altman Z-score, with an explanation of their correlation with the probability of default and with consideration of the challenges in collecting and interpreting these data. The linear discriminant analysis method and logistic regression were used for the calculations.

The combination of financial and non-financial information increased the predictive power of the model. The research confirms that diversity in the team is closely linked to greater innovation, better financial indicators and access to capital, which determines the long-term success of start-up companies and influences the assessment of young companies. In conclusion, it can be emphasised that the integration of quantitative and qualitative information is of key importance for the further development of risk assessment models in crowdlending.

Such an approach increases the effectiveness of credit decisions, reduces the default rate and strengthens the stability of platforms.