Forecasting Sub-Sovereign Credit Ratings using Machine Learning Methods

Abstract:

Recently, credit rating agencies worldwide have gained higher rate of criticisms for their influence on the economic crisis that started in 2007. These rating agencies are often criticized for their perceived inaccurate ratings and slow reactions to new information (Paudyn, 2013; Kim and Park, 2016). The role of these credit rating agencies is very influential on the financial markets in the production of credit risk information of issuer’s and its allocation to market participants. Credit rating reviews the solvency of an entity or a particular debt obligation in a single ordinal class. The credit rating can be defined as a degree of the ability of a firm to meet its debt servicing obligations in time, and gives a way to quantify how close the entity is to avoidance. The main property of ratings is to translate credit risk into a single class to make them very attractive to market participants. Credit ratings boost borrowing conditions by the cost of credit and the availability of the credit. For example, a lower bond rating would increase the risk quality demanded by the market and therefore the cost of borrowing; a lower rating could also result in the reduction of the potential market for the downgraded debt, since some investors are either incapable or not ready to hold debt below a certain rating (Cheung, 1996). In a municipal and regional setting (sub-sovereign), credit ratings are usually used to express the general financial performance. This rather difficult evaluation is based on diverse criteria than the financial performance of firms (Beck et al., 2017). Regional financial performance is influenced by various socio-economic determinants.

nsdlogo2016