The U.S. subprime mortgage-backed securities market has attracted intense attention during and after the widespread financial turmoil of 2007-2009. Most commentators have reached a consensus that an underestimation of the risks associated with these products led to problems for the entire financial system. Therefore, this thesis seeks to provide a better understanding of these complex financial instruments and their role in the propagation of the crisis. We focus on three main issues. The first issue (Chapter 3) analyses the risk factors underlying this market. The goal is to identify key variables that may potentially explain its decline. Measures of the U.S. real estate market, interbank liquidity and counterparty risk, as well as market volatility are all found to play a role. Furthermore, we find that the importance of these risk factors changed as the crisis evolved from a real estate problem to a broader global credit crisis. The second issue (Chapter 4) concerns identifying interdependencies and contagion within this market during the crisis. We adapt the vector autoregressive (VAR) framework of Longstaff (2010) to estimate the intra-market relationships using a spliced ABX dataset and two traded ABX vintages. We find contagious effects during the subprime crisis, emanating mainly from shocks to the higher-rated assets. Finally, the third issue (Chapters 5 & 6) examines contagion from the subprime mortgage-backed securities market to several other asset markets using both the original and an extended version of Longstaff's (2010) VAR framework. Using the original specifcation, evidence of contagion is found but we uncover important differences between the choice of index used to proxy for the subprime mortgage-backed securities market. Furthermore, employing a more innovative econometric tool, namely a timevarying transition probability Markov-switching VAR, shows that although contagion played a role in transmitting the shock to other markets, it may not have been as prevalent as suggested by the original VAR framework. We conclude that accurately dating the crisis is crucial to the analysis.