Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

Fin 500Q –  Quantitative Risk Management

Homework #3 Solutions

1. Suppose that the price of an asset at close of trading yesterday was $300 and its volatility was estimated as 1.3% per day. The price at the close of trading today is $298. Update the volatility estimate using:

(a) The EWMA model with λ = 0.94.

Answer: Using the EWMA model, the variance is updated to

0.94 × 0.0132 + 0.06 × (−2/300)2  = 0.00016153

so that the new daily volatility is ^0.00016153 = 0.01271 or 1.271%.

(b) The GARCH(1,1) model with ω = 0.000002,α = 0.04, and β = 0.94.

Answer: Using GARCH(1,1), the variance is updated to

0.000002 + 0.94 × 0.0132 + 0.04 × (−2/300)2  = 0.00016264

so that the new daily volatility is ^0.00016264 = 0.01275 or 1.275%.

2. Suppose that the parameters in a GARCH(1,1) model are α = 0.03,β = 0.95, and ω = 0.000002.

(a) What is the long-run average volatility?

Answer: The long-run average variance, VL , is

ω            0.000002

1 − α β          0.02

The long-run average volatility is ^0.0001 = 0.01 or 1% per day.

(b) If the current volatility is 1.5% per day, what is your estimate of the volatility in 20, 40, and 60 days?

Answer: The expected variance in 20 days is

0.0001 + 0.9820 (0.0152 − 0.0001) = 0.000183.

The expected daily volatility is therefore ^0.000183 = 0.0135 or 1.35%.  Similarly, the expected volatilities in 40 and 60 days are 1.25% and 1.17% per day, respectively.

(c) What volatility should be used to price 20-, 40, and 60-day options?

Answer: We have a = ln ( ) = 0.0202. The variance used to price 20-day options is 252 × [0.0001 + 1  (0.0152 − 0.0001)] = 0.051

so that the annual volatility is 22.61%. Similarly, the volatilities that should be used for 40- and 60-day options are 21.63% and 20.85% per annum, respectively.

(d) Suppose that there is an event that increases the current volatility from 1.5% per day to 2% per day. Estimate the effect on the volatility in 20, 40, and 60 days.

Answer: The expected variance in 20 days is now

0.0001 + 0.9820 (0.022 − 0.0001) =  0.0003.

The expected daily volatility in 20 days is therefore ^0.0003 = 0.0173 or 1.73%.  Similarly, the expected daily volatilities in 40 and 60 days are 1.53% and 1.38%, respectively.

(e) Estimate by how much the event increases the volatilities used to price 20-, 40, and 60-day options. Answer: When today’s volatility increases from 1.5% per day to 2% per day, the volatility used to price a 20-day option changes by

1 e 0 .0202×20          0.015 

or 6.88% on an annual basis, bringing the volatility up to 29.49%. Similarly, the 40- and 60-day volatilities increase to 27.63% and 26.10%, respectively.

3. The probability density function for an exponential distribution is λe λx  where x is the value of the

variable and λ is a parameter. The cumulative probability distribution is 1 − e λx . Suppose that two

Gaussian copula to define the correlation structure between V1  and V2 . You can use the file “bivar.xls”

to compute values of the cumulative bivariate normal distribution function.

(a) What is the probability that V1  ≤ 1?

Answer: The probability is 1 − e 1 .0·1  = 0.632.

(b) What is the probability that V2  ≤ 1?

Answer: The probability is 1 − e 2 .0·1  = 0.865.

(c) With a copula correlation of 0, what is the probability that V1  ≤ 1 and V2  ≤ 1?

Answer:  The probability that V1  ≤ 1 is transformed to the normal value U1  = Φ 1 (0.632) = 0.337. This probability can be calculated in Excel with the formula =NORM .INV(0.632, 0, 1).         Similarly, the probability that V2  ≤ 1 is transformed to the normal value U2  = Φ 1 (0.865) = 1.102. With a copula correlation of 0, we can use the provided Excel file to find that the joint probability is M(0.337, 1.102, 0) = 0.547 (= 0.632 × 0.865).

(d) With a copula correlation of 0.5, what is the probability that V1  ≤ 1 and V2  ≤ 1?

Answer: With a copula correlation of 0.5, the joint probability is M(0.337, 1.102, 0.5) = 0.591.

(e) With a copula correlation of −0.2, what is the probability that V1  ≤ 1 and V2  ≤ 1?

Answer: With a copula correlation of −0.2, the joint probability is M(0.337, 1.102, −0.2) = 0.531.

4. Suppose that a bank has made a large number of loans of a certain type. The one-year probability of default on each loan is 1.2%. The bank uses a Gaussian copula for time to default. It is interested in estimating a 99.97% worst case for the percent of loans that default on the portfolio.  Show how this worst case percentage varies with the copula correlation, using copula correlations of 0 .2, 0.4, 0.6, and 0.8.

Answer: The WCDR with a 99.97% confidence level is

Φ ( ) ,

where Φ is the CDF of a standard normal distribution. We compute that Φ 1 (0.012) = −2.257 (Excel:

WCDR = Φ (  −2.25 · 3.432 ) = Φ(−0.807) = 0.210,

which is computed in Excel as =NORM .DIST(−0.807, 0, 1, 1).

The table below gives the WCDR for different values of the copula correlation.

 

ρ

WCDR(%)

0.2

21.0

0.4

45.5

0.6

73.7

0.8

96.5