Hide/Show Apps

Predicting financial distress of Turkish non-financial firms: evidence from micro data

Yılmaz, Muhammed Hasan
In this thesis, determinants of financial distress probabilities are analyzed for the period over 2006-2016 by utilizing firm-level and loan-level data sets. Based on the financial distress definition constructed from the existence of non-performing loans, univariate tests indicate that financially problematic Turkish corporates have lower liquidity, profitability and asset turnover, while they are also the ones with inferior short and long-term debt paying ability. To form multivariate specifications, variable selection techniques such as univariate logit, principal components analysis and bootstrap stepwise logit estimations are applied on a broad set of financial ratios. Multivariate logit and panel logit models indicate that stronger liquidity buffers, abundant profits and improved asset turnover decrease the probability of financial distress. On the other hand, rising interest expenses and considerable debt burden increase the likelihood of facing financial problems. The informative nature of multivariate models are evaluated for in-sample predictions and out-of-sample forecasts. Survival analysis is also performed to assess the time to financial distress. It is found that firm size, profitability, liquidity and operational performance lengthen the period until the occurence of financial problems, while leverage significantly shortens it.