eThesis - Kauppakorkeakoulun sähköiset gradut

The purpose of this thesis is to find out the most significant financial ratios in determining credit ratings and construct an estimated rating model based on these ratios. Further the model is first tested in-sample and then out-of-sample on the case company Finnvera plc's data. Thus, another objective of this thesis is to compare the accuracy of Finnvera's rating model with the Standard & Poor's rating model and give suggestions for Finnvera plc, on how to improve its current rating process in terms of choosing and emphasizing the most significant variables.

The rating model estimated in this study is based on a sample of financial statements from 543 Standard & Poor's rated North American companies for the year 2013 with revenue in between 50 million and 5 billion US dollars. The financial statement data of Finnvera, to which the model was tested, included 57 companies from different countries and industries. The linear regression model was used as a statistical model in this study.

The findings are similar to prior academic literature. Profitability and coverage ratios and the size indicators were found to be statistically significant in explaining credit ratings in a positive fashion. Solvency ratios (leverage) were found to be negatively statistically significant in relation to predicting credit ratings. Liquidity ratios proved to be statistically insignificant in the estimation of credit ratings. In terms of specific financial ratios - operating profit margin, net profit margin, return on assets, natural logarithm of total assets and leverage - were found to be the most significant determinants of credit ratings in this research.

The findings of this thesis also suggest that there is significant correlation between Standard & Poor's ratings and Finnvera plc's ratings. Thus, this research indicates that Finnvera as a state owned company uses a rating system, which is commonly reliable. The reliability and correlation of Finnvera's rating process to other commonly used rating processes (e.g. Standard & Poor's, Moody's or Fitch) has not been tested earlier, which increases the value of this study.