The restriction of banker's credit distribution caused many debitor candidates find another way on financing to adequate their needs. An alternative that used to choose is offering bonds. One thing that can not be avoid on investing is default risk. So that investor have to be selective to choose their investing instrument?s before put their money on it. To measure a risk of a bond, investor can see from its rating. The bond ratings issued by an institution called credit rating company. Publised creding announcement from the credit rating company sometimes already responded by the market before, so it becomes too late to used.
This research has a purpose to analyze the prediction of the bond ratings and how they are determined, so we can predict the next credit rating. Unlike previous study,this research include both of accounting factors and accounting factors as a bond rating predictors and test explanatory power of the following five variables : (1) Leverage ratio, (2) Current ratio, (3) Size, (4) Growth and (5) Maturity structure.
This research is a quantitative research which apply the dichotomous logistics regression model on the data to test the influence of five variables above to predict the bond ratings. The data consist of the manufacture companies listed in Indonesia Stock Exchange and PT.Pefindo on 2004-2006. The research question is asked whether accounting factors and non accounting factor can predict bond ratings if used together.
The result of this research shows that leverage ratio is the best variable that represent accounting factor which had classification power to predict bond ratings. That is because the leverage ratio shows the proportion of the liabilities to the assets, where bond is one of liabilities component. The maturity that represent non accounting factor can not predict bond ratings because the data of maturity are homogenous.
Based on the result, this research conclude that acconting factors and non accounting factor can not perform well if used together. The next research should be add more variables as the predictor and apply another models to predict bond rating, like Multi Discriminant Analysis (MDA) or Probit Model.