Abstract
Purpose of the research is to identify the socio-economic determinants of loan delinquencies among households in Kazakhstan, considering the specifics of the regional context and the spread of digital lending. Methodology is based on the analysis of cross-sectional survey data from 256 households conducted in 2023 in the Kapshagay region. To test two hypotheses, logit and probit regressions were applied: H1 — the number of loans is positively associated with the risk of default; H2 — low per capita household income increases the probability of delinquency. Originality / value of the study lies in the use of microdata that reflect actual borrowing practices in an emerging financial market, as well as in the inclusion of demographic characteristics that are rarely considered in borrower assessments. Unlike traditional models employed by banks for automated credit decision-making, this study additionally analyzes variables such as per capita household income. This makes it possible to more accurately capture social vulnerability and potential insolvency risks among different population groups. Findings show that the number of active loans significantly increases the probability of default, whereas per capita household income does not have a substantial effect. Social status and credit accessibility also proved to be significant factors. The study contributes to the literature on financial vulnerability and may be useful for regulators and microfinance organizations in developing sustainable lending policies.