ABC Finance Balance Sheet Analysis

Assets Liabilities and Equity

The balance sheet of ABC Finance is given below (market yields are in parentheses and amounts in millions):

Cash $ 20 Overnight repos (5%) $ 340

1-month T-bills (7.0%) 150 Subordinated debt 300

3-month T-bills (7.25%) 150 (7-year fixed rate, 8.5%)

2-year T-notes (7.50%) 100

8-year T-notes (9%) 50

Mortgage Loans (adjustable, 8%, reset every 6 months ) 200 Equity 30 670 670

(a) Briefly define what rate sensitive assets/liabilities are. [10%]

(b) What is the repricing gap if the planning period is 30 days? [10%]

(c) What is the impact over the next three months on net interest income if interest rates on RSAs and RSLs increase by 0.5%? [10%]

(d) What is the impact over the next two years on net interest income if interest rates on RSAs increase by 0.5% and on RSLs increase by 0.75%? [10%]

(e) List the missing directions in the NII column of the following table for the repricing model (note: ?R in the table indicates the direction of the rate change for both the RSA rates and the RSL rates). For example, row 1 says that if the CGAP is negative, the interest rates fall and the spread narrows, then the change in NII cannot be predicted without knowing the size of the CGAP and the change in the spread.

Make sure you write your answers not on this exam sheet but in the answer booklet!

(f) Briefly discuss the weaknesses of the repricing model. [32%]

You have a cross-sectional data set on loan defaults for small companies from the US with the following variables:

Default =1 if loan default, =0 otherwise

Amount Loan amount (in million $)

RealEstate =1 if loan is backed by real estate, =0 otherwise

Urban =1 if company is located in urban area, =0 otherwise

New =1 if company is a new business, =0 otherwise

You estimate two Logit models and obtain the following results:

(a) Evaluate the statistical significance of the estimated coefficients in the larger model (m1). [10%]

(b) For the larger model (m1), and for each explanatory variable, discuss whether the sign of the estimated coefficient is what you would expect. [20%]

(c) Based on the larger model (m1), what is the predicted probability of a loan default for a company that obtains a loan of 1 million $. Assume the loan is backed by real estate and the company is located in an urban area and is an existing business. [20%]

(d) Which of the two models do you prefer in the sense that it fits the data better “in-sample”. Apply different available procedures to choose between the two models. [30%]

(e) The model that fits the data best “in-sample” is not necessarily the model that fits the data best “out-of-sample”. Describe a way how you can find out which of the two models is better when it comes to “out-of-sample” forecasting of loan default probabilities. [20%] 5 [Please turn over ]

(a) Assume that you compute the following one-day ahead forecasts of the conditional volatilities and conditional covariance for daily Greggs (GRG) and J D Wetherspoon (JDW) stock returns: σˆGRG = 0.04, σˆJDW = 0.01, σˆGRG,JDW = 0.00002. Using the delta-normal approach and stating all relevant assumptions,compute the 1-day VaR for an equally weighted portfolio of the GRG and JDW stocks at the 99% confidence level assuming the current portfolio value is £1,000,000. [25%]

(b) Using relevant algebra explain how to compute an estimate of the 1-day VaR at the 99% confidence level for an equally weighted portfolio of these two stocks using the historical simulation approach. [25%]

(c) Using relevant algebra, explain how to compute an estimate of the 1- day VaR at the 99% confidence level for an equally weighted portfolio of these two stocks using the Monte Carlo simulation approach. [50%]