As part of this project you will need to answer all the questions shown below. For questions 1, 2 and 3 you will need to consider a portfolio of sovereign loans as indicted in Appendix A. Please choose a portfolio of your choice from Appendix A, but at least one contry should be different from your original group project allocation. You should assume that your analysis is conducted on 30th April 2021. All the data you need to collect will have to be consistent with such date. State the assumptions, if any, you need to make to carry out your calculations and explain why, in your opinion, they are sensible.

1. [10% weight]. After reviewing relevant industry papers (e.g. from regulators and/or practitioners) and the financial press, describe the macro-economic challenges facing each country in your portfolio [max 200 words per country].

2. [30% weight] Calculation of the risk of a sovereign loan portfolio.

   1. With the CreditMetrics model compute 95% and 99% relative (from the mean) VaR and ES for the portfolio of sovereign loans assigned to your group (see Appendix A). Use a time horizon of 1 year, 10,000 Monte Carlo simulations and a 30% asset correlation.

      [Note: Relative VaR in CreditMetrics is computed as E(V)-V* where E(V) is the expected value of the 1-year forward portfolio value and V* is the simulated 1-year forward portfolio value at the 5%, 1% and 0.1% percentiles, respectively. Relative expected shortfall is computed as E(V)-V** where V** is the average of the 1-year forward portfolio values below V*. ]

   2. Explain why you think that your results are plausible.

   Portfolios are composed of exposures to Turkey, Argentina, Brazil and Venezuela. To implement CreditMetrics you will need the following information:

   1. Zero coupon government yields. You may obtain yields from Bloomberg, Reuters, Datastream or other services. The yields for the relevant maturities in your analysis can be derived by interpolation in case of need. Assume that all yields refer to debt denominated in US dollars.

   2. Country ratings should be sourced via the Moodys website. Simply select look up a rating from the Research and Ratings tab. You will need to register with Moodys to have access to the rating information. Registration is free.

   3. Sovereign default rates and recovery rates can be obtained from the document Sovereign Default and Recovery Rates, 1983-2019 which can be downloaded from the Moodys website. In case information is not available, you will need to make reasonable assumptions and explain their rationale.

3. [20% weight] Stress testing: Compute the 95% and 99% relative VaR and ES for your portfolio under the following stress scenarios. Consider the effect of each scenario separately and then all combined.

   Scenarios:

   1. Assume that yields increase by 20% across all maturities (i.e. if a

      current yield is 7% it will go to 17%).

   2. Assume asset correlation goes to 70%.

   3. Explain why you think that your results in all the above points

      are plausible.

4. [40% weight] Review the IMFs Regulatory reform 10 years after the global financial crisis: looking back, looking forward and other relevant material on the topic which can be drawn from academic papers, (e.g. from Science Direct) industry papers (e.g. from regulators and practitioners) and the financial press. Answer the following questions:

   1. What is your assessment of the capital and leverage regulation that was introduced following the financial crisis? [max 400 words]

   2. What is your assessment of the systemic risk regulation that was introduced following the financial crisis? [max 400 words]

   3. In your opinion, what are the main regulatory challenges going forward? [max 400 words]

   Add the word count in each sub-question. In your answer to each sub-question you should include one figure or table that best illustrates one or more of the key points you are discussing. Follow these stylistic guidelines. Figures and tables can be taken from existing published material, in which case the source needs to be acknowledged in the figure/table legend, or built by yourself with data you may obtain from Bloomberg, Reuters, Datastream or other services. Tables and figures legends should not be included in the word count.