Estimation of the probability of default for sovereign credit risk using structural models

Abstract

Over the years, it has become imperative for creditors to know the creditworthiness of prospective borrowers. This research aims at predicting the probability of default by a sovereign state using a structural model (the Merton model), which is chosen for its logical simplicity. A major challenge with the implementation of structural models is the estimation of asset value and its volatility. To overcome this, this dissertation makes use of net foreign assets value as a proxy for sovereign assets, and the volatility of sovereign assets is estimated to be the historical volatility (standard deviation) of the net foreign assets. The Merton model is then applied on six countries to estimate their default probabilities from 2011-2020. The results obtained using the Merton model are compared against the credit ratings of the 6 sample countries; three of which are considered investment grade and three which are considered speculative grade. The investment grade countries considered are Botswana, Mexico and Bulgaria, while the speculative grade countries considered are South Africa, Brazil and Serbia. The results show that the countries rated as speculative have been close to default since 2011. Of the three investment grade countries, only Mexico has been close to default since 2011. Botswana has been unlikely to default since 2011, while Bulgaria has been unlikely to default since 2014. The findings show that although the Merton model due to certain limitations could return a higher risk-neutral probability of default than the real-world probability of default, it generally returns results which match up with the credit ratings awarded by the Big Three (Standard and Poor’s Global Ratings, Moody’s Investors Service, and Fitch Ratings) credit rating agencies and Trading Economics.