Valuation of Additional Tier-1 Contingent Convertible Bonds (AT1 CoCo) : Modelling trigger risk in a practical investment setting

Sammanfattning: Contingent convertible bonds (often referred to as CoCo bonds, or simply CoCos) are a relatively new financial instrument designed to absorb unexpected losses. This instrument became increasingly more common after the financial crisis of 2008, as a way to decrease the risk of insolvency among banks and other financial institutions. In this thesis, we will investigate two mathematical models for valuation of CoCo bonds, known as the credit derivative approach and the equity derivative approach, previously developed by De Spiegeleer and Schoutens [1]. We will investigate how these models can be modified in order to be applied to a large set of bonds available on the market. The effect of parameter alterations will also be studied, in order to determine which parameters that influence the pricing accuracy the most. We reach the conclusion that by estimating market triggers, conversion prices and by computing a continuous interest rate from a discrete rates table, the models are indeed executable on a large set of bonds available on the market. However, these parameter estimations come at the cost of reduced accuracy. In general, both investigated models produces prices which follows the overall movements of the market prices quite well, but at the same time with a relatively large absolute distance from the market prices. In other words, the correlation with the market is often high, but the absolute error (measure by root mean square error) is often large. The sensitivity analysis of the parameters shows that the market trigger is the most influential parameter in both investigated models. The fact that we had to estimate the market trigger in order to be able to price a large number of bonds is believed to be the main cause of reduced accuracy. By utilizing a more bond-specific parameter estimation, the accuracy of the investigated models could most likely be improved. We can conclude that there is a trade-off between being able to price a large set of bonds with a mediocre accuracy, or being able to price a few bonds with high accuracy.