You are on page 1of 39

Level I (Questions 1-20)

FRM 2010 Practice Questions Vol. I

By David Harper, CFA FRM CIPM


www.bionicturtle.com

Table of Contents
Question 1: Simulation methods [quantitative]

Question 2: Jensens Alpha measure [foundations]

Question 3: Creating Value [foundations]

Question 4: Market Structure [products]

Question 5: Bonds DV01 [valuation]

Question 6: Mean & standard deviation [valuation]

Question 7: Simple regression model [quantitative]

10

Question 8: Null hypothesis [quantitative]

12

Question 9: Hypothesis testing [quantitative]

14

Question 10: Option delta [valuation]

16

Question 11: Estimate of invoice price [products]

18

Question 12: Hedge [products]

20

Question 13: Market/Credit/Operational risk [foundation]

22

Question 14: Models for estimating volatility [quantitative]

24

Question 15: Duration of the bond [valuation]

26

Question 16: Diversification for a VaR [valuation]

28

Question 17: Probability [quantitative]

30

Question 18: Bonds [valuation]

33

Question 19: Duration and convexity [valuation]

35

Question 20: Historical simulation [valuation]

37

www.bionicturtle.com

FRM 2010 LEVEL I (QUESTIONS 1-20) 1

These sample questions are for paid members of bionicturtle.com only! Anyone else is
using an illegal, pirated copy and also violates GARPs ethical standards.
Some of the questions may have a follow-up explanation. This would be located on the
forum: http://www.bionicturtle.com/forum/viewforum/45/

Question 1: Simulation methods [quantitative]


Which of the following statements about simulation is invalid?
a) The historical simulation approach is a nonparametric method that makes no specific
assumption about the distribution of asset returns.
b) When simulating asset returns using Monte Carlo simulation, a sufficient number of
trials must be used to ensure simulated returns are risk neutral.
c) Bootstrapping is an effective simulation approach that naturally incorporates
correlations between asset returns and non-normality of asset returns, but does not
generally capture autocorrelation of asset returns.
d) Monte Carlo simulation can be a valuable method for pricing derivatives and examining
asset return scenarios.
[Please note: the additional, follow-up questions were written by David Harper. The
goal is to explore the topic in greater depth.]
1.2. Which VaR simulation approach can incorporate (handle) heavy-tailed or skewed asset
returns?
1.3. Briefly explain Linda Allens hybrid approach.
1.4. What is the motive and advantage of Jorions QMC method?

Answer: B
Explanation: Risk neutrality has nothing to do with sample size.
Topic: Quantitative Analysis, Subtopic: Simulation methods. Reference: Jorion, chapter 12.
1.2 Which VaR simulation approach can incorporate (handle) heavy-tailed or skewed asset
returns? All VaR methods, including both HS and MCS, can incorporate non-normal returns!
1.3 The hybrid approach blends HS and EWMA for the weighting scheme; Linda Allens hybrid is
still essentially a SIMULATION approach as a parametric form does not describe returns.
1.4 quasi Monte-Carlo (QMC) are faster, producing an error that shrinks at a faster rate,
proportional to close to 1/K rather than 1/SQRT(k)

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 2

Question 2: Jensens Alpha measure [foundations]


Portfolio Q has a beta of 0.7 and an expected return of 12.8%. The
market risk premium is 5.25%. The risk-free rate is 4.85%. Calculate
Jensens Alpha measure for Portfolio Q.
a) 7.67%
b) 2.70%
c) 5.73%
d) 4.27%
2.2. What is the portfolios Treynor measure?
2.3. Are Jensens and Treynor related?
2.4. What is the portfolios Sharpe measure?
2.5. What is a criticism of Jensens alpha?
2.6. Is Jensens alpha the same as Grinolds alpha?

Answer: D (4.27% or 4.28%).


Explanation: Jensens alpha is defined by: E(RP ) RF = P + P(E(RM) RF);
P = E(RP ) RF - P(E(RM) RF) = 0.128 - 0.0485 - 0.7 * (0.0525 + 0.0485 - 0.0485)= 0.0427
a. Incorrect. Forgets to subtract the risk-free rate for the excess market return.
b. Incorrect. Forgets to multiply the excess market return by beta.
c. Incorrect. Forgets to subtract the risk-free rate for both excess market return and excess
portfolio return.
d. Correct.
2.2 Treynor measure = (12.8% - 4.85%) / 0.7 = 0.114
2.3. Yes, both assume CAPM such that an exact linear relationship exists between them!
2.4. Not enough information! We need portfolio volatility.
2.5 Says Amenc, The Jensen measure is subject to the same criticism as the Treynor measure:
the result depends on the choice of reference index. In addition, when managers practice a
market timing strategy, which involves varying the beta according to anticipated movements in
the market, the Jensen alpha often becomes negative, and does not then reflect the real
performance of the manager.
2.6 No, Grinolds alpha (think hedge fund alpha) is a generalized version of Jensens alpha.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 3

Jensens alpha presumes CAPM such that the only risk factor is the market (or equity) risk
premium, and alpha is the residual return that is unexplained only by the equity risk premium:
Jensens alpha = portfolio return (ERP * beta)
Grinolds pure alpha is the residual return that is unexplained by any and all common beta
factor exposures:
alpha = portfolio return return (exposure1*factor 1+ exposure2*factor2 + exposure3*factor3 +
.... + exposure_n*factor_n). Hopefully, you can see how we can view Jensens alpha as a
SPECIAL case of Grinolds alpha where there is only one common risk factor (i.e., the
equity risk premium per the market model)

Question 3: Creating Value [foundations]


A corporation is faced with the decision to choose between the two
following projects:
Project Investment
A
100
20
B
80
55

Perpetual Annual Cash Flow


50
200

Cash Flow at Risk

Assuming that there is no systematic risk and the projects are mutually exclusive, under what
circumstances would project A be selected over project B?
a) Project A should never be chosen because it requires a larger initial investment and
generates lower perpetual annual cash flows.
b) Project A could be preferred over Project B if Project As cash flows are negatively
correlated with the firms existing cash flows while the cash flows of Project B are highly
positively correlated with the firms existing cash flows.
c) Project A should be chosen if the opportunity cost of funds is low, and Project B should
be chosen otherwise.
d) Project A should be chosen if the net present value of the project is positive.
3.2. If the confidence level is 95%, what are the implied project cash flow volatilities?
3.3 Is Stulz CFaR a relative or absolute VaR?
3.4 The question says assuming there is no systematic risk What does that imply about beta?
Are there any non-zero betas to be found?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 4

Answer: B.
Explanation: Project A should be chosen only if the cash flow at risk of the project has low or
negative correlation with the other projects the company currently has or plans. The overall
cash flow position of the firm has to be evaluated as a result.
3.2. If the confidence level is 95%, what are the implied project cash flow volatilities?
Stulz CFaR = cash flow volatility * normal deviate. In this case, CFaR = CF volatility * 1.645, such
that:
Project A CF volatility = -50/-1.645 = $30.4, and
Project B CF volatility = -200/-1.645 = $121.6
3.3 Is Stulz CFaR a relative or absolute VaR?
It is a relative VaR because it is given by Prob[E(C) C > CFaR] = 5%
i.e., it is not a shortfall relative to zero (an absolute loss) but rather a shortfall relative to
EXPECTED cash flow
3.4 The question says assuming there is no systematic risk What does that imply about beta?
Are there any non-zero betas to be found?
The question implies the market betas of the project are zero (i.e., no systematic risk). However,
this does not imply that PROJECT BETAS (i.e., the beta of the project with respect to company
cash flow) is zero. In fact, please note that Project As cash flows are negatively correlated with
the firms existing cash flows necessarily implies a NEGATIVE PROJECT BETA for project A
because:
Project Beta = Covariance (Project, Firm) / Variance (Firm) = Volatility (Project)*Volatility
(Firm) * Correlation (Project, Firm) / Variance (Firm) = Volatility (Project)*Correlation (Project,
Firm) / Volatility (Firm).
And since Volatility () must be positive, a negative Correlation () implies a negative project beta!

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 5

Question 4: Market Structure [products]


If the lease rate of commodity A is less than the risk-free rate, what is the
market structure of commodity A?
a) Backwardation
b) Contango
c) Flat
d) Inversion
4.2 Is the CONVERSE a true statement: Does a contango necessarily imply the lease rate is less
than (<) risk-free rate, for a consumption commodity?
4.3 What is the financial commodity equivalent of a consumption commodity lease rate?
4.4 (Hard) How does McDonald define lease rate in terms of growth rate and discount rate?
4.5 (Helpful) What does contango imply on the commodity roll return?
4.6 (To contemplate) Use 4.5 to explain impact of shift to contango on Metallgeschaft?

Answer: B
Explanation: 1. Contango occurs when futures prices are higher than current spot, so in this
case the risk-free rate is greater than the lease rate. 2. Backwardation occurs when futures
prices are less than spot, so in this case the lease rate is greater than risk-free rate. So, if the
lease rate is less than the risk-free rate, the futures price is above the current spot price.
4.2 Is the CONVERSE a true statement: Does a contango necessarily imply the lease rate is less
than (<) risk-free rate, for a consumption commodity?
No, not necessarily, depending on our definition of lease rate. If lease rate excludes storage
cost, then positive storage cost could overwhelm a high lease rate and create contango.
4.3 Dividend income.
4.4 (Hard) How does McDonald define lease rate in terms of growth rate and discount rate?
McDonald uses: lease rate (delta) = commodity discount rate (alpha) expected growth rate (g)
(This is McDonald equation 6.8: it requires study)
4.5 (Helpful) What does contango imply on the commodity roll return?
Contango (normal backwardation) implies a loss (gain) on the futures contract roll return;
i.e., the futures price is dropping to converge with the spot price!
4.6 (To contemplate) Use 4.5 to explain impact of shift to contango on Metallgeschaft?
Metallgeschaft was hedging with short-term LONG futures oil contracts; the shift to contango
created roll return LOSSES which we immediately MARKED-TO-MARKET (although the
underlying long contracts gained, those gains could not be realized).

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 6

Question 5: Bonds DV01 [valuation]


Sarah is a risk manager responsible for the fixed income portfolio of a
large insurance company. The portfolio contains a 30-year zero coupon
bond issued by the US Treasury (STRIPS) with a 5% yield. What is the
bonds DV01?
a) 0.0161
b) 0.0665
c) 0.0692
d) 0.0694
5.2. The question says, The portfolio contains a 30-year zero coupon bond issued by the US
Treasury Does the US Treasury issue STRIPS?
5.3. This question assumes semi-annual compounding (this assumption should be stated!). What
is the price of this bond assuming semi-annual compounding?
5.4. What is the bond price assuming continuous compounding? (You must be able to do this
calculation with ease!)
5.5. Assuming semi-annual compounding, what is, respectively, the Macaulay duration and
Modified duration of this bond?
5.6. Assuming continuous compounding, what is, respectively, the Macaulay duration and
Modified duration of this bond?
5.7. Assuming semi-annual compounding, solve for the bonds DV01 as a function of the bonds
PRICE and MODIFIED DURATION. (Please focus on this question, important relationship!)
5.8 [Bonus, beyond L1] This question assumes a particular version of DV01 (by default). Can we
specifically name this DV01?

Answer: B
Explanation:
The DV01 of a zero-coupon is
DV01 = 30 / 100 (1 + y/2)2T+1 100 (1 + 5%/2)61 = 0.0665
Topic: Valuation and Risk Models
Subtopic: DV01, duration and convexit
Reference: Tuckman, Chapter 5
(the source explain appears to have typos but gets to correct answer). As follows:
1/10,000 * 1/(1+y/2) * [T * 100/(1+y/2)^(2T)]
= 1/10,000 [T * 100/(1+y/2)^(2T+1)]

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 7

= T /[100*(1+y/2)^(2T+1)]
= 30/[100*(1+5%/2)^(60+1) = $0.06662
5.2. The question says, The portfolio contains a 30-year zero coupon bond issued by the US
Treasury Does the US Treasury issue STRIPS?
No. The government does not directly issue STRIPS; they are formed by investment
banks or brokerage firms, but the government does register STRIPS in its book-entry
system. They cannot be bought through TreasuryDirect, but only through a broker.
http://en.wikipedia.org/wiki/United_States_Treasury_security
5.3. This question assumes semi-annual compounding (this assumption should be stated!). What
is the price of this bond assuming semi-annual compounding?
N= 30*2 = 60
I/Y = 5/2 = 2.5
PMT = 0
100 = FV
CPT PV = $22.73
5.4. What is the bond price assuming continuous compounding? (You must be able to do this
calculation with ease!)
PV = 100 * EXP(-5%*30) = $22.31
notice the CC bond price is necessarily lower than the s.a. bond price
5.5. Assuming semi-annual compounding, what is, respectively, the Macaulay duration and
Modified duration of this bond?
The Mac duration is 30 years (Mac duration = T for a zero coupon bond).
Modified duration = 30/(1+5%/2) = 29.2683
5.6. Assuming continuous compounding, what is, respectively, the Macaulay duration and
Modified duration of this bond?
The Mac duration is 30 years (Mac duration = T for a zero coupon bond).
Under the special case only of continuous compounding, Modified duration = Macaulay
duration; so, in this case only, Mod duration = 10, also! Can you show why this is the case?
5.7. Assuming semi-annual compounding, solve for the bonds DV01 as a function of the bonds
PRICE and MODIFIED DURATION. (Please focus on this question, important relationship!)
DV01 = Price * Modified Duration / 10,000; in this case,
DV01 = $22.73 * 29.26829 / 10,000 = 0.06652
5.8 [Bonus, beyond L1] This question assumes a particular version of DV01 (by default). Can we
specifically name this DV01?
This question assumes, as does the FRM typically, a YIELD-BASED DV01 which means that
the one basis point change is a change in the YIELD-TO-MATURITY (YTM). DV01
generalizes to include other 1 bps shifts; e.g., spot/zero curve, forward curve.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 8

Question 6: Mean & standard deviation [valuation]


Currently, shares of ABC Corp. trade at USD 100. The monthly risk neutral
probability of the price increasing by USD 10 is 30%, and the probability
of the price decreasing by USD 10 is 70%.What are the mean and standard
deviation of the price after 2 months if price changes on consecutive
months are independent?
a.
b
c.
d.

Mean
70
70
92
92

Standard Deviation
11.32
12.96
11.32
12.96

6.2 What is skew of price distribution after two months?


6.3. Is the final price distribution light- or heavy-tailed?
6.4. What is the value of a two-month option if strike price is $100 (i.e., strike = stock) and
riskless rate is 4%?
6.5 Solve for the implied (annualized) volatility

Answer: D
Explanation:
Develop a 2 step tree. Mean = 9% (120) + 42% (100) + 49% (80) = 92
Variance = 9% (120 - 92)2 + 42% (100 - 92)2 + 49% (80 - 92)2 = 168 Thus, standard deviation =
12.96
6.2 Skew = third moment about mean / cube of standard deviation (note: sample skew not
required). Skew = 0.62
6.3. Kurtosis = fourth moment about mean / standard deviation^4
Kurtosis = 2.38 or Excess kurtosis = -0.61
As a binomial distribution, this is a LIGHT-TAILED DISTRIBUTION
(i.e., as n increases, kurtosis approaches 3 and the binomial converges on normal)
6.4. What is the value of a two-month option if strike price is $100 (i.e., strike = stock) and
riskless rate is 4%?
Backward induction (@4% riskless rate) under two-step implies option price = $1.79
6.5 Solve for the implied (annualized) volatility.
We can approximate (albeit a shift from discrete to continuous) by employing Hull 11.13:
u = EXP(volatility*SQRT(delta time));
volatility = LN(u)/SQRT(delta time); and in this case
volatility = LN(1.1)/SQRT(1/12) = 33%

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 9

Question 7: Simple regression model [quantitative]


Which of the following statements about the ordinary least squares
regression model (or simple regression model) with one independent
variable are correct?
i.

In the ordinary least squares (OLS) model, the random error term is assumed to have
zero mean and constant variance.

ii.

In the OLS model, the variance of the independent variable is assumed to be positively
correlated with the variance of the error term.

iii.

In the OLS model, it is assumed that the correlation between the dependent variable and
the random error term is zero.

iv.

In the OLS model, the variance of the dependent variable is assumed to be constant.

a) i, ii, iii, and iv


b) ii and iv only
c) i and iv only
d) i, ii, and iii only
7.2. When we move from the regression with one independent variable to a multiple regression
model (i.e., two or more independent variables), the assumptions underlying the CLRM are
essentially the same except for the ADDITION of one key assumption. What is the additional
assumption and what do we call its violation?
7.3 Which answer given in the explanation for 7.1 (source) is technically incorrect?
7.4 If our regression meets the CLRM assumptions, what is the salient implication on regression
analysis?
7.5 How many one-word VIOLATIONS of the CLRM model assumptions can you name?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 10

Answer: C
Explanation:
i. Is correct. In Simple Linear Regression model, the random error term is assumed to be
stationary. It means that the variance of random error term must be constant, or by using
another term: it is assumed that there is no heteroskedasticity in linear regression model.
ii. Is incorrect. In Simple Linear Regression model, the independent variable and the error term
have constant variances.
iii. Is incorrect. The dependent variable is allowed to be correlated with the error term.
iv. Is correct. In Simple Linear Regression model, the variance of the dependent variable is
assumed to be constant. Thus, the correct option is option C.
Topic: Quantitative Analysis, Subtopic: Linear regression, Reference: Gujarati, Ch 7, pp. 140-145.
7.2. When we move from the regression with one independent variable to a multiple regression
model (i.e., two or more independent variables), the assumptions underlying the CLRM are
essentially the same except for the ADDITION of one key assumption. What is the additional
assumption and what do we call its violation?
Gujarati A8.6 is the additional assumption: No exact collinearity exists between any two
explanatory variables. The violation is called multicollinearity.
7.3 Which answer given in the explanation for 7.1 (source) is technically incorrect?
The Explain for (iii) says In Simple Linear Regression model, the independent variable and the
error term have constant variances but it should say either:
In Simple Linear Regression model, the assumption is that the error term has constant variance
In Simple Linear Regression model, the assumption is that the error term is uncorrelated with
the explanatory variable
but the requirement that the independent variable [must] have constant variance is
awkward at a minimum: the independent variables can be stochastic (as long as they are
uncorrelated with the error)
7.4 If our regression meets the CLRM assumptions, what is the salient implication on regression
analysis?
The key implication is that we produce DESIRABLE ESTIMATORS (slope, intercept). Specifically,
the estimators are BLUE: best linear unbiased estimator.
7.5 How many one-word VIOLATIONS of the CLRM model assumptions can you name?
If not Linear in PARAMETERS (not variables!) then: NONLINEAR violation!
If not constant variance, then: Heteroskedastic violation!
If covariance (u,u) is non-zero, then: Autocorrelation violation!
IF collinearity between X1, X2, then: imperfect (not deadly) or perfect (deadly!)
Multicollinearity violation!

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 11

Question 8: Null hypothesis [quantitative]


Bob tests the null hypothesis that the population mean is less than or
equal to 45. From a population size of 3,000,000 people, 81 observations
are randomly sampled. The corresponding sample mean is 46.3 and
sample standard deviation is 4.5. What is the value of the appropriate
test statistic for the test of the population mean, and what is the correct
decision at the 1 percent significance level?
a) z = 0.29, and fail to reject the null hypothesis
b) z = 2.60, and reject the null hypothesis
c) t = 0.29, and accept the null hypothesis
d) t = 2.60, and neither reject nor fail to reject the null hypothesis
8.2 Where is the technical mistake in the answer given?
8.3 What is the 99% confidence interval for the true population mean?
8.4. If we want to create a confidence interval for the population variance, what distribution is
employed?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 12

Answer: B
Explanation:
a. is incorrect. The denominator of the z-test statistic is standard error instead of standard
deviation. If the denominator takes the value of standard deviation 4.5, instead of standard error
4.5/sqrt(81), the z-test statistic computed will be z = 0.29, which is incorrect.
b. is correct. The population variance is known and the sample size is large (>30). The test
statistics is: z = (46.3-45)/(4.5/(sqrt(81)) = 2.60. Decision rule: reject Ho if zcomputed >
zcritical. Therefore, reject the null hypothesis because the computed test statistics of 2.60
exceeds the critical z-value of 2.33.
c. is incorrect because z-test (instead of t-test) should be used for sample size (81) >= 30
d. is incorrect because z-test (instead of t-test) should be used for sample size (81) >= 30
8.2 Where is the technical mistake in the answer given?
The population variance/standard deviation is not given; consequently the TEST STATISTIC
uses the sample standard deviation (4.5). Please note this itself JUSTIFIES the use of the
STUDENTS T DISTRIBUTION (technically, because a d.f. has been consumed). The Z-statistic is
justified if we use the population standard deviation.
Therefore, an ACCEPTABLE ANSWER is also: t = 2.6 and REJECT THE NULL (i.e., one-tailed
critical t @ 99% = 2.374).
Please note: technically the computed test statisticbecause it used the sample sigmais a
students t variable not a Z-variable.
However, this is a LARGE SAMPLE (n=81) and on that groundsi.e., that the students t with
large d.f. approximates (is asymptotic to) the normalwe can use the normal!
To recap the bottom line: by using the sample variance/sigma, the students T is
technically correct and justified (i.e., known population variance justifies the normal!),
but because the sample is large, we are okay to use the students T because as the sample
size increases the students t is approaching the normal.
8.3 What is the 99% confidence interval for the true population mean?
If students t, then 46.3 +/- (2.639)*(4.5/SQRT(81)) = 44.98 =< pop mean =< 47.62
If Z, then 46.3 +/- (2.576)*(4.5/SQRT(81)) = 45.01 =< pop mean =< 47.59
did you notice the change from a one-tailed null hypothesis to a TWO-TAILED confidence
interval?
8.4. If we want to create a confidence interval for the population variance, what distribution is
employed?
The chi-squared distribution

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 13

Question 9: Hypothesis testing [quantitative]


Which one of the following four statements about hypothesis testing holds
true if the level of significance decreases from 5% to 1%?
a) It becomes more difficult to reject a null hypothesis when it is actually true.
b) The probability of making a type I error increases.
c) The probability of making a type II error decreases.
d) The failure to reject the null hypothesis when it is actually false decreases to 1%.
9.2 If the level of significance decreases from 5% to 1%, the probability of a Type II error
increases from 95% to 99%. True or false?
9.3 Is there any way to simultaneously decrease the probability of both a Type I and Type II
error?
9.4 What are the normal deviates, respectively, for a 99% VaR (1% significance) and a 95% VaR
(5%) significance? Can you use these deviates in a sentence that expresses them in terms of
Type I/II error
9.5. (bonus) In the Basel II IMA back-test framework (i.e., green/yellow/red traffic light), what is
a Type I error?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 14

Answer: A
Explanation:
Type I error: The rejection of the null hypothesis when it is actually true.
Type II error: The failure to reject the null hypothesis when it is actually false. The significance
level is the probability of making a type I error.
a. is correct. Decreasing the probability level makes it more difficult to reject the null when it is
true.
b. is incorrect. Decreases the probability of making a type I error.
c. is incorrect. All else being equal, the decrease in the probability of making a Type I error
comes at the cost of increasing the probability of making a Type II error.
d. is incorrect. Increases the probability of making a Type II error, in other words, the
probability of failing to reject the null hypothesis when it is actually false decreased
9.2 If the level of significance decreases from 5% to 1%, the probability of a Type II error increases
from 95% to 99%. True or false?
False. If the significance level is 1%, then the probability of a Type I error is 1%. And, due to the
necessary trade-off, a low significance level does imply a increase in the probability of a Type II
error (i.e., to accept, or fail to reject, a false null) HOWEVER this probability is not equal to 1
significance.
9.3 Is there any way to simultaneously decrease the probability of both a Type I and Type II error?
Yes, but only by increasing the sample size. For a given sample size, the trade-off is unavoidable.
9.4 What are the normal deviates, respectively, for a 99% VaR (1% significance) and a 95% VaR
(5%) significance? Can you use these deviates in a sentence that expresses them in terms of Type
I/II error?
One-tailed normal deviate @ 99% = 2.33
One-tailed normal deviate @ 99% = 1.645
i.e., VaR is always one-tailed! Memorize these deviates, they are commonly tested!
In the case of a 99%, we could say:
If the VaR model is accurate, the probability of a loss in excess of 2.33 standard deviations is 1%
which is the probability of a Type I error; i.e., the left-tail rejection region is 1%, or even better
If the VaR model is accurate, we do expect a loss of AT LEAST 2.33 standard deviations in 1% of
instances, which is the probability of a Type I error; i.e., the left-tail rejection region is 1%,
notice this latter emphasizes the fact the VaR is not giving us information about the tail
(losses in excess of the VaR threshold)
9. 5. (bonus) In the Basel II IMA back-test framework (i.e., green/yellow/red traffic light), what is a
Type I error?
The null is the banks VaR model is accurate. A Type I error is to reject an accurate model (red
zone) and a Type II error is to accept an inaccurate model (green zone).

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 15

Question 10: Option delta [valuation]


Mr. Black has been asked by a client to write a large put option on the
S&P 500 index. The option has an exercise price and a maturity that are
not available for options traded on exchanges. He, therefore, has to
hedge the position dynamically. Which of the following statements about
the risk of his position are not correct?
a) He can make his portfolio delta neutral by shorting index futures contracts.
b) There is a short position in an S&P 500 futures contract that will make his portfolio
insensitive to both small and large moves in the S&P 500.
c) A long position in a traded option on the S&P 500 will help hedge the volatility risk of the
option he has written.
d) To make his hedged portfolio gamma neutral, he needs to take positions in options as
well as futures.

[my adds]
10.2. In regard to answer (d) above, explain why a futures contract is insufficient to neutralize
gamma.
10.3 Assume Mr. Black wrote 10,000 put options with strike price at the current index price
(ATM). Given an estimate for percentage delta and position delta; and use these estimates to
illustrate the impact of an index price decline on the position.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 16

Answer: B
Explanation:
The short index futures makes the portfolio delta neutral. It does not help with large moves,
though.
Topic: Valuation and Risk Models, Subtopic: Greeks, Reference: Hull, Chapter 17.
10.2. In regard to answer (d) above, explain why a futures contract is insufficient to neutralize
gamma.
The gamma of a futures contact is approximately equal to zero: the delta of a futures contract
equals EXP(rate*time), so the delta starts (at longer maturities) as 1.X and only slightly
converges to 1.0 as maturity decrease. Given that delta is almost a constant, the gamma is zero
and therefore, futures cannot neutralize gamma. Please note: neither a position in the
underlying (in this case, the index) nor an forward/futures contract on the same can neutralize
gamma (both are linear).
10.3 Assume Mr. Black wrote 10,000 put options with strike price at the current index price (ATM).
Given an estimate for percentage delta and position delta; and use these estimates to illustrate the
impact of an index price decline on the position.
A put option has a NEGATIVE percentage delta; if ATM, the put delta is generally in the
neighborhood of -0.4 (but it depends on the inputs!). Please note: delta of put= delta of call 1 =
N(d1) 1.
If we illustratively assume percentage delta = -0.4, then position delta of this SHORT position is
given by:
Position delta = -0.4 percentage delta * -10,000 written options = +4,000
... notice the short position is captured by a NEGATIVE in the QUANTITY. Here, the negatives
cancel to create a positive position delta; i.e., an increase in the underlying (index) produces a
gain for the short put
Now we might say,
Given a 1-unit drop (-1) in the index price, the position losses (-4000) or,
Given a 1-unit increase (+1) in the index price, the position gains (+4000) or,

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 17

Question 11: Estimate of invoice price [products]


On March 13, 2008, William Tell, a fund manager for the Rossini fund,
takes a short position in the March Treasury bond (T-bond) futures
contract. He plans to deliver the cheapest-to-deliver Treasury bond with
a coupon of 4.5% payable semiannually on May 15 and November 15 (182
days between), a conversion factor of 1.3256, and a face value of USD
100,000. The delivery date is Friday, March 15 (121 days after November
15 coupon payment date). The settlement price for the cheapest-todeliver Treasury bond on March 13 is 68 2/32. Which of the following is
the best estimate of the invoice price?
a) USD 90,118.87
b) USD 91,719.53
c) USD 92,367.75
d) USD 95,619.47
11.2 Which day count convention is assumed here?
11.3 Would the invoice price be higher or lower if the day count convention were instead,
respectively, 30/360 and Actual/360?
11.4 Assume the quoted price of the CTD bond above is $100.00 (i.e., the bond with a conversion
factor of 1.3256). Assume further that another eligible-for-delivery bond has a quoted price of
$110.00 with a conversion factor (CF) of 1.45. In this case, which bond is cheapest-to-deliver?
11.5 Why are there several bonds (a basket) that can be delivered in the CBOT Treasury bond
futures contact, as opposed to a single bond?
11.6 In regard to factors that determine the cheapest to deliver (CTD) bond, what bonds are
favored when bond yields are high (low)?
11.7 In regard to factors that determine the cheapest to deliver (CTD) bond, what bonds are
favored when the yield curve is upward-sloping (downward-sloping)?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 18

Answer: B
The invoice is based on a settlement price of 68 2/32 or 68.0625. The accrued interest is
calculated on the basis of the number of days since the last coupon payment date, November 15,
and the delivery date, March 15. That is 121. During the current six-month period between
coupon payment dates, November 15 to May 15, there are 182 days. Thus the accrued interest
on USD 100,000 face value of the bond is 121/182 * USD 100,000 * 0.045/2 = USD 1,495.88
Explanation:
The invoice price is USD 100,000 * 0.680625 * 1.3256 + USD 1,495.88 = 91,719.53
Topic: Financial Markets and Products. Subtopic: Cheapest to deliver bond, conversion factors
Reference: Bruce Tuckman, Fixed Income Securities, 2nd Edition. This is incorrect reference:
should be Hull Chapter 6
11.2 Which day count convention is assumed here?
Actual/Actual (121/182) is the convention for U.S. Treasury bonds.
11.3 Would the invoice price be higher or lower if the day count convention were instead,
respectively, 30/360 and Actual/360?
If 30/360, then AI = 120/180 * USD 100,000 * 0.045/2 = $1,500. HIGHER.
If Actual/360, then AI = 121/180 * USD 100,000 * 0.045/2 = $1,512. HIGHER.
11.4 Assume the quoted price of the CTD bond above is $100.00 (i.e., the bond with a conversion
factor of 1.3256). Assume further that another eligible-for-delivery bond has a quoted price of
$110.00 with a conversion factor (CF) of 1.45. In this case, which bond is cheapest-to-deliver?
The net cost = (quoted bond price) (settlement price * CF).
In the case of the original bond above, net cost =$ 100 - $68.0625*1.3256 = $9.78
In the case of the second bond, net cost = $110 - $68.0625*1.45 = $11.31
Therefore, the original remains the CTD.
Please note the logic of CTD:
The short must purchase the bond with cost = Quoted bond price + AI
Then delivers the bond in order to receive = (Settle price * CF) + AI
So net cost = cash paid cash received
= Quoted bond price + AI [(Settle price * CF) + AI] = Quoted bond price - (Settle price * CF)
11.5 Why are there several bonds (a basket) that can be delivered in the CBOT Treasury bond
futures contact, as opposed to a single bond?
Tuckman says, The design of bond futures contracts purposely avoids a single underlying
security. One reason for this is that if the single underlying bond should lose liquidity, perhaps
because it has been accumulated over time by buy-and-hold investors and institutions, then the
futures contract would lose its liquidity as well. Another reason for avoiding a single underlying
bond is the possibility of a squeeze. To illustrate this problem, assume for the moment that only
one bond were deliverable into a futures contract. Then a trader might be able to prot by
simultaneously purchasing a large fraction of that bond issue and a large number of contracts. As
parties with short positions in the contract scramble to buy that bond to deliver or scramble to

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 19

buy back the contracts they have sold, the trader can sell the holding of both bonds and
contracts at prices well above their fair values. But by making shorts hesitant to take positions,
the threat of a squeeze can prevent a contract from attracting volume and liquidity.
11.6 In regard to factors that determine the cheapest to deliver (CTD) bond, what bonds are
favored when bond yields are high (low)?
Hull: A number of factors determine the cheapest-to-deliver bond. When bond yields are in
excess of 6%, the conversion factor system tends to favor the delivery of low-coupon longmaturity bonds. When yields are less than 6%, the system tends to favor the delivery of highcoupon short-maturity bonds.
11.7 In regard to factors that determine the cheapest to deliver (CTD) bond, what bonds are
favored when the yield curve is upward-sloping (downward-sloping)?
Hull: when the yield curve is upward-sloping, there is a tendency for bonds with a long time to
maturity to be favored, whereas when it is downward-sloping, there is a tendency for bonds
with a short time to maturity to be delivered.

Question 12: Hedge [products]


The yield curve is upward sloping, and a portfolio manager has a long position in 10-year
Treasury Notes funded through overnight repurchase agreements. The risk manager is
concerned with the risk that market rates may increase further and reduce the market value of
the position. What hedge could be put on to reduce the positions exposure to rising rates?
a) Enter into a 10-year pay fixed and receive floating interest rate swap.
b) Enter into a 10-year receive fixed and pay floating interest rate swap.
c) Establish a long position in 10-year Treasury Note futures.
d) Buy a call option on 10-year Treasury Note futures.
12.2 What additional risk(s) in the total position (i.e., long note plus hedge trade) are not
necessarily hedged. Please think about this before looking at the answer
12.3 Assume the portfolio manager has $1 million invested in the long position with an expected
duration of 7.0 years. If he/she hedges with T-bond (T-note) futures with contract price of 90.00
and duration of 8.0 years, what is the duration-based hedge?
12.4 In the original question, assume the portfolio manager hedges with a 5-year swap where
the notional equals the principal invested. What is the problem with such a hedge?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 20

Answer: A
Explanation:
a. is correct. An increase in rates will increase the value of the hedge position and offset the loss
in value from the Bond position.
b. is incorrect. An increase in rates will decrease the value of the hedge position and add to the
loss in value from the Bond position.
c. is incorrect. An increase in rates will decrease the value of the futures position and add to the
loss in value from the Bond position.
d. is incorrect. An increase in rates (all else equal), will decrease the value of the call option and
add to the loss in value from the Bond position.
Topic: Financial Markets and Products. Subtopic: Futures, forwards, swaps and options
Reference: John Hull, Options, Futures, and Other Derivatives, 6th Edition (New York: Prentice
Hall, 2006) Chapter 7 - Swaps
12.2 What additional risk(s) in the total position (i.e., long note plus hedge trade) are not
necessarily hedged. Please think about this before looking at the answer
First, the long T-Note is funded by a (short-term) overnight repo; we can call this funding
liquidity risk.
Second, the swap involves a counterparty so the hedge transaction incurs counterparty risk.
(arguably, under an alternative taxonomy, we could argue the repo funding is subject to market
liquidity risk. And, as counterparty risk is a sub-class of credit risk, we could subsume both
under credit risk.)
Third, there remains BASIS RISK between the underlying T-note exposure and the swap hedge;
dont forget basis risk!
12.3 Assume the portfolio manager has $1 million invested in the long position with an expected
duration of 7.0 years. If he/she hedges with T-bond (T-note) futures with contract price of 90.00
and duration of 8.0 years, what is the duration-based hedge?
The number of contracts = $1 million / (90 * $100,000 face) * (7/8) = 9.72.
The PM would SHORT approximately 10 contracts.
Note answer (c) would be correct if it said establish a SHORT position in 10-year Treasury
note futures
12.4 In the original question, assume the portfolio manager hedges with a 5-year swap. What is the
problem with such a hedge?
The long Treasury note position likely has a duration > 7 but the swaps duration will be less
than 5; i.e., the swap can be treated as two bondsa fixed and a floaterand the duration of
floater is roughly zero (or time to next coupon) such that the swaps duration is approximately
equal to the fixed legs duration.
So, in this case, when interest rates increase, the underlying position will lose more value than
will be offset by the increase in the swaps value.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 21

Question 13: Market/Credit/Operational risk [foundation]


Jennifer Durrant is evaluating the existing risk management system of
Silverman Asset Management. She is asked to match the following events
to the corresponding type of risk. Identify each numbered event as a
market risk, credit risk, operational risk, or legal risk event.
Event
1) Insufficient training leads to misuse of order management system.
2) Credit spreads widen following recent bankruptcies.
3) Option writer does not have the resources required to honor a contract.
4) Credit swaps with counterparty cannot be netted because they originated in multiple
jurisdictions.
a) 1: legal risk, 2: credit risk, 3: operational risk, 4: credit risk
b) 1: operational risk, 2: credit risk, 3: operational risk, 4: legal risk
c) 1: operational risk, 2: market risk, 3: credit risk, 4: legal risk
d) 1: operational risk, 2: market risk, 3: operational risk, 4: legal risk
13.2 Where is the MISTAKE in the answer given?
13.3 Please try to identify THREE RISKS that are not included among market, credit or
operational risks.
13.4 Is legal risk an operational risk?
13.5 Which of the risks listed above are meant to be covered by Basel II capital requirements?
13.6 [this does not have a correct answer, but please think about it]. Where do we classify
LIQUIDITY RISK vis--vis Credit, Market & Operational risks?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 22

Answer: C
Explanation: a, b and d are incorrect. c is correct.
1. Insufficient training leads to misuse of order management system is an example of
operational risk.
2. Widening of credit spreads represents an increase in market risk. <<This is incorrect
3. An option writer not honoring the obligation in a contract is a credit risk event.
4. When a contract is originated in multiple jurisdictions leading to problems with
enforceability, there is legal risk.
Reference: Jorion, Value at Risk, Chapter 1
13.2 Where is the MISTAKE in the answer given?
Although market risk and credit risk do interact (e.g., market risk impacts credit risk), we
regard CREDIT SPREADS as primarily determined by CREDIT RISK not market risk.
13.3 Please try to identify THREE RISKS that are not included among market, credit or operational
risks.
Business risk, strategic risks and Reputational risks (although reputational risk is particularly
controversial in the taxonomy)
13.4 Is legal risk an operational risk?
Yes, according to Jorion, legal risk is a sub-class of operational risk. Therefore, it would also be
correct to identify even #4 as an Operational Risk
13.5 Which of the risks listed above are meant to be covered by Basel II capital requirements?
All of them:
Credit, Market and Operational risks constitute the big three risk buckets that are explicitly
charged at under either basic (standardized) and advanced/internal methods.
Legal risk would be captured by an EVENT TYPE (e.g., internal fraud, external fraud) in AMA.
In BIA or SA, legal risk would implicitly be implicitly included in the percentage of gross income
(GI) approach.
13.6 [this has not correct answer, but please think about it]. Where do we classify LIQUIDITY RISK
vis--vis Credit, Market & Operational risks?
In Chapter 1 of VaR 3rd Edition, Jorion argues that liquidity risk is a broad category along with
market, credit and operational risk.
More importantly, did you remember to parse liquidity risk into two distinct but interrelated
dynamics?

Market liquidity risk (aka, asset liquidity risk), and


Funding-liquidity risk (aka, cash flow liquidity risk)

There are various treatments of liquidity risks among different taxonomies. Some would classify
liquidity risk entirely under market risk. But see
http://www.bionicturtle.com/learn/article/where_is_liquidity_risk_is_the_risk_taxonomy/
where Kutler smartly (IMO) classifies market liquidity risk as a sub-class of credit risk and
funding-liquidity risk as a risk effect rather than a risk factor.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 23

Question 14: Models for estimating volatility [quantitative]


Which one of the following four statements on models for estimating
volatility is incorrect?
a) In the RiskMetrics EWMA model, some positive weight is assigned to the long-run
average variance rate.
b) In the RiskMetrics EWMA model, the weights assigned to observations decrease
exponentially as the observations become older.
c) In the GARCH (1, 1) model, a positive weight is estimated for the long-run average
variance rate.
d) In the GARCH (1, 1) model, the weights estimated for observations decrease
exponentially as the observations become older.
14.2 What is the most likely weight of the most recent observation (observation = squared
return) in the RiskMetrics EMWA model?
14.3 True or false: EWMA and GARCH (1,1) are both parametric approaches to VaR. If true, what
is their advantage/disadvantage as parametric methods?
14.4 True or false: In the multivariate density estimation (MDE) approach, the weights estimated
for observations decrease exponentially as the observations become older.
14.5 True or false: In Lindas Allens hybrid approach, the weights estimated for observations
decrease exponentially as the observations become older.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 24

Answer: A
Explanation:
a. is incorrect. The RiskMetrics model does not involve the long-run average variance rate in
updating volatility, in other words, the weight assigned to the long-run average variance rate is
zero.
b. is correct. In the RiskMetrics model, the weights assigned to observations decrease
exponentially as the observations become older.
c. is correct. In the GARCH (1, 1) model, some positive weight is assigned to the long-run
average variance rate.
d. is correct. In the GARCH (1, 1) model, the weights assigned to observations decrease
exponentially as the observations become older.
Topic: Quantitative Analysis, Subtopic: EWMA, GARCH models, Reference: Hull, Chapter 21.
14.2 What is the most likely weight of the most recent observation (observation = squared return)
in the RiskMetrics EMWA model
In the infinite series, if we are estimating todays volatility (n) the weight applied to period (n-t)
= (1-lambda)*lambda^(t-1). The most recent weight (at t=1), assuming a lambda of 0.94, is given
by: (1-94%)*94%^(1-1) = 6%*1 = 6%.
14.3 True or false: EWMA and GARCH(1,1) are both parametric approaches to VaR. If true, what is
their advantage/disadvantage as parametric methods?
True: EWMA and GARCH are both parametric: historical (empirical) data is used to compute a
variance/volatility. VaR is then computed per the distributional assumption (the historical data
is no longer needed).
Advantages of parametric methods include parsimony (no dataset) and analytical tractability
(e.g., we can scale standard deviation per the square root rule)
The salient disadvantage, for risk purposes, is arguably the inability of any distribution to
adequately characterize the tail (and extreme tail). More broadly, the distributional assumption
is unlikely to fully characterize the actual behavior of returns (e.g., skewed, fat-tailed, unstable).
Note this problem extends BEYOND NORMAL distribution; of course the normal does not
characterize returns.
14.4 True or false: In the multivariate density estimation (MDE) approach, the weights estimated
for observations decrease exponentially as the observations become older.
False (except if by coincidence): in MDE a state vector (current versus historical) determines the
weights
14.5 True or false: In Lindas Allens hybrid approach, the weights estimated for observations
decrease exponentially as the observations become older.
TRUE! The hybrid is essential a simulation method, but unlike simple historical simulation (HS),
the historical returns are weighted in declining fashion per EWMA. (the hybrid refers to a
hybrid between HS and EWMA, but ultimately this is a NON-PARAMETRIC SIMULATION
approach).

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 25

Question 15: Duration of the bond [valuation]


The table below gives the closing prices and yields of a particular liquid
bond over the past few days.
Day
Monday
Tuesday
Wednesday

Price
106.3
105.8
106.1

Yield
4.25%
4.20%
4.23%

What is the approximate duration of the bond?


a) 18.8
b) 9.4
c) 4.7
d) 1.9
15.2 What is the mistake in the question (why is the data virtually impossible)?
15.3. What is the implied dollar duration of the bond?
15.4 What is the implied DV01 of the bond
15.5 What is the implied convexity of the bond [this can actually be observed without
calculations!]

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 26

Answer: B
Explanation:
The duration can be approximated from the price changes. (106.3 - 105.8)/106.3/.0005 = 9.4
(106.3 - 106.1)/106.3/.0002 = 9.4
Topic: Valuation and Risk Models. Subtopic: DV01, duration and convexity. Reference: Tuckman,
chapter 5
15.2 What is the mistake in the question (why is the data virtually impossible)?
Notice that the price of the bond increases as the yield increases!
This implies negative duration which is only possible for an interest-only (IO) security.
15.3. What is the implied dollar duration of the bond?
The dollar duration (DD) is the slope of the tangent line. In this case, DD
= (106.1 105.8) / (4.23% - 4.20%) = 1,000
(note: consistent with the original flaw, the DD is positive but typical bonds have negative dollar
durations as the slope of the tangent line is negative).
15.4 What is the implied DV01 of the bond
DV01 = P*D/10,000 = 10.6.1*-9.43/10000 = -$0.10
Please note the calculation is unnecessary: we can observe the bond price changes $0.20 when
the yield shifts by two basis points (2 bps) from 4.23% to 4.25%. So, we can observe that a ONE
BASIS POINT SHIFT implies ~ $0.10 price change.
15.5 What is the implied convexity of the bond [this can actually be observed without calculations!]
The dollar duration/DV01 is exactly the same from 4.20% to 4.23% as it is from 4.23% to
4.25%; i.e., the DVO1 is CONSTANT at $0.01! Convexity is the rate of change of the (dollar)
duration (the 2nd derivative of price with respect to yield = the 1st derivative of the 1st
derivative). Convexity is therefore ZERO.
Mathematically, we can confirm:
Dollar convexity = (1000 dollar duration @ 4.25% - 1000 dollar duration @ 4.23%)/(4.25% 4.23%)
= 0/0.02% = 0;
Convexity = dollar convexity / $106.1 = 0.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 27

Question 16: Diversification for a VaR [valuation]


Bond
A
B

Yield

Maturity (Yrs)

STDEV of Yield

Annual Exposure

5%
3%

2
13

5%
12%

USD 25.00
USD 75.00

The correlation between the two returns is 0.25. From a risk


management perspective, what is the gain from diversification for a VaR
estimated at the 95% level for the next 10 days? Assume there are 250
trading days in a year.
a) 76,500
b) 283,000
c) 382,300
d) 1,413,000
16.2 The question incorrectly (or imprecisely at least) solves for the yield VaR (i.e., the worst
expected change in yield). Per Jorion, we would instead typically solve for a return- or pricebased VaR. What is the better answer to the question if we compute VaR in terms of price/value
at risk?
16.3. What is the individual VaR of Bond B?
16.4. If the correlation changes to 1.0, what is the difference between diversified and
undiversified VaR?
16.5. Define the marginal VaR (Bonus: Calculate marginal VaRs)
16.6. Define component VaR (Bonus: Calculate component VaRs)

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 28

Answer: B
Explanation:
1. Calculate the undiversified VaR
VaRundiv = 1.645 * 5%*(10/250) * 25 + 1.645 * 12% 10/250 *75 = 0.4113 + 2.9610 = 3.3723
2. Calculate the diversified VaR
1.645 0.25 2 * 5% 2 + 0.75 2 * 12% + 2 * 0.25 * 0.75 * 5% * 12% * 0.25 * (10/250) * 100 =
1.645 * 0.0939 * 10/250 * 100 = 3.0893
3. Difference is 0.283
Topic: Valuation and Risk Models. Subtopic: VaR for fixed income securities. Reference: Allen,
Boudoukh, Saunders, Understanding market, Credit and Operational Risk: The Value at Risk
Approach, Chapters 2, 3
16.2 The question incorrectly (or imprecisely at least) solves for the yield VaR (i.e., the worst
expected change in yield). Per Jorion, we would instead typically solve for a return- or price-based
VaR. What is the better answer to the question if we compute VaR in terms of price/value at risk?
Volatility (dP/P) = Duration * volatility (yield); i.e., Jorion 8.14
If we assume semi-annual compounding, then duration of bonds is:
Duration (bond A) = 2/(1+5%/2) = 1.95,
Duration (bond B) = 2/(1+3%/2) =12.81,
Price volatility (bond A) = 5% yield volatility * 1.95 = 9.76%, and
Price volatility (bond B) = 12% yield volatility * 12.81 = 153.7%.
Per the formula above (portfolio variance), the DIVERSIFIED 10-day VaR = $38.1293, and
the UNDIVERSIFIED 10-day VaR = $38.723.
The difference is 0.59386 ($594,000)
16.3. What is the individual VaR of Bond B?
= 153.7% * SQRT(10/250) * 1.645 * $75 = $37.92
16.4. If the correlation changes to 1.0, what is the difference between diversified and undiversified
VaR?
Zero. At perfect correlation, benefit of diversification reduces to zero.
16.5. Define the marginal VaR (Bonus: Calculate marginal VaRs)
Marginal VaR = deviate * Covariance (Asset, Portfolio)/Volatility (portfolio);
Marginal VaR (Bond A) = 1.645 * 0.0305/$115.9 = 0.0433
Marginal VaR (Bond B) = 1.645 * 0.1.7810/$115.9 = 2.5275
16.6. Define component VaR (Bonus: Calculate component VaRs)
Component VaR = marginal VaR * position. In this case,
Component VaR (Bond A, time scaled) = 0.0433 marginal VaR * $25 * SQRT(10/250) = $0.22
Component VaR (Bond A, time scaled) = 2.5275 marginal VaR * $75 * SQRT(10/250) = $37.91
And note that the SUM of COMPONENT VaRs equals the diversified VaR:
$0.22 + $37.91 = $38.1293

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 29

Question 17: Probability [quantitative]


Assume that a random variable follows a normal distribution with a mean
of 100 and a standard deviation of 17.5.What is the probability that this
random variable is between 82.5 and 135?
a) 68.0%
b) 81.9%
c) 82.8%
d) 95.0%
17.2. Which of Gujaratis sampling distributions (FRM assigned Chapter 4) converge on a normal
distribution?
17.3. In significance tests of the regression coefficients like slope (i) what justifies the
assumption of normal distribution and (ii) what fact anyhow often precludes its use?
17.4. (please dont peek, give this some thought) Assume we are given NO INFORMATION about
the distribution. Now, what is the probability that the random variable is between 65 and 135?
17.5 What is meant by the normals property of LOCATION-SCALE INVARIANCE (source: FRM
assigned Rachev)?
17.6 What is meant by the normals property of SUMMATION STABILITY (source: FRM assigned
Rachev)?
17.7 What is meant by the normals property of DOMAIN OF ATTRACTION (source: FRM
assigned Rachev)?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 30

Answer: B
Explanation: Prob (-1* < X < 2*) = (1 - 0.0228) - 0.1587 = 0.8185
a. is incorrect. Almost 68% of the observations will be within the interval from one standard
deviations below the mean to one standard deviations above the mean, which is within the
interval [100 - 17.5; 100 + 17.5].
b. is correct. 82.5 = 100 - 17.5 and 135 = 100 + 2 * 17.5. So, the percentage is 34% on the left
hand side of the mean, plus 95%/2 on the right hand side of the mean.
c. is incorrect. Almost 95% of the items will lie within the interval from two standard deviations
below the means to two standard deviations above the mean, that is within the interval [100 - 2
*17.5;100 + 2 * 17.5].
d. is incorrect. Assumes wrongly that 97.5% of the observations will be within [100 - 2 *
17.5;100 + 2 * 17.5].
Topic: Quantitative analysis, Subtopic: Probability Distributions, Reference: Damodar N Gujarati,
Essentials of Econometrics, 3rd Edition (New York: McGraw-Hill, 2006), chapter 4, pp. 80-84
17.2. Which of Gujaratis sampling distributions (FRM assigned Chapter 4) converge on a normal
distribution?
All of them!
Students t is heavy-tailed but approximates normal as d.f. > 29 (n=30).
Chi-square is non-negative and positively skewed but, for large d.f., also converges to normal
(Per CLT, as chi-square is sum of independent random variables)
F-distribution is (also, like chi-square) non-negative and skewed, but similarly approaches
normal for large d.f.
(note the binomial and Poisson also approach normal as, respectively, N and lambda, increase
toward infinity)
17.3. In significance tests of the regression coefficients like slope (i) what justifies the assumption of
normal distribution and (ii) what fact anyhow often precludes its use?
(i) The central limit theorem justifies the normal assumption;
(ii) But in practice we rarely know the POPULATION VARIANCE, so we use the sample variance
which consumes a d.f. and requires the students t distribution
17.4. (please dont peek, give this some thought) Assume we are given NO INFORMATION about the
distribution. Now, what is the probability that the random variable is between 65 and 135?
Chebyshevs Inequality says that P[X is WITHIN +/- 2 sigma] is AT LEAST 1 - 1/(2^2) = 3/4;
similarly,
The P [X is WITHOUT +/- 2 sigma] is LESS THAN OR EQUAL TO 1/(2^2) = 1/4
So, we can say, the probability the random variable is between 82.5 and 135 GREATER THAN OR
EQUAL TO 75%. (at most, 25% lie outside this interval)

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 31

17.5 What is meant by the normals property of LOCATION-SCALE INVARIANCE (source: FRM
assigned Rachev)?
Cconsider the random variable Y, which is obtained as Y = aX + b. In general, the distribution of Y
might substantially differ from the distribution of X, but in the case where X is normally
distributed, the random variable Y is again normally distributed with parameters and . Thus, we
do not leave the class of normal distributions if we multiply the random variable by a factor or
shift the random variable.
17.6 What is meant by the normals property of SUMMATION STABILITY (source: FRM assigned
Rachev)?
If you take the sum of several independent random variables, which are all normally distributed
with mean (mu) and standard deviation (sigma), then the sum will be normally distributed
again. For reflection, why is this relevant for financial series?
17.7 What is meant by the normals property of DOMAIN OF ATTRACTION (source: FRM assigned
Rachev)?
Rachev: The last important property that is often misinterpreted to justify the nearly exclusive
use of normal distributions in financial modeling is the fact that the normal distribution
possesses a domain of attraction. A mathematical result called the central limit theorem states
thatunder certain technical conditionsthe distribution of a large sum of random variables
behaves necessarily like a normal distribution. In the eyes of many, the normal distribution is
the unique class of probability distributions having this property. This is wrong and actually it is
the class of stable distributions (containing the normal distributions), which is unique in the
sense that a large sum of random variables can only converge to a stable distribution.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 32

Question 18: Bonds [valuation]


The following table gives the prices of two out of three US Treasury notes
for settlement on August 30, 2008. All three notes will mature exactly
one year later on August 30, 2009. Assume annual coupon payments and
that all three bonds have the same coupon payment date.
Coupon
2 7/8
4 1/2
6

Price
98.40
?
101.30

Approximately what would be the price of the 4 1/2 US Treasury note?


a) 99.20
b) 99.40
c) 99.80
d) 100.20
18.2. If the actual (observed) price of the 4 1/2 is $100.00, is the bond trading cheap or rich, and
what might explain?
18.3 Does the law of one price allow for arbitrage opportunities?
18.4 Which of the bonds carries the highest yield (YTM; assume semi-annual compounding as
this refers to Tuckman!)?
18.5 (bonus) Can different yields be explained under the no-arbitrage assumption, or is than an
internal inconsistency (error) in the question?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 33

Answer: C
Explanation: 2.875% * x + 6.25% *(1 - x) = 4.5% X = 52%
The portfolio that has cash flows identical to the 4 1/2 bond consists of 52% of the 2 7/8 and
48% of the 6 1/4 bonds. As this portfolio has cash flows identical to the 4 1/2 bond, precluding
arbitrage, the price of the portfolio should equal to 52% * 98.4 + 48% * 101.30 or 99.80
Topic: Valuation and Risk Models
Subtopic: Bond prices, spot rates, forward rates
Reference: Tuckman, Chapter 1
18.2. If the actual (observed) price of the 4 1/2 is $100.00, is the bond trading cheap or rich, and
what might explain?
If the market price is greater than the predicted (model) price, the bond is trading rich; i.e., $100
> $99.80.
Technical factors (e.g., supply/demand) can explain be they represent omitted model factors. In
Tuckmans discussion of Treasuries that trade rich/cheap he tends to explain with LIQUIDITY;
i.e., rich bonds are more demanded (or less supplied) and cheap bonds are less demanded (or
more supplied).
18.3 Does the law of one price allow for arbitrage opportunities?
No.
The law of one price states that absent confounding factors (e.g., liquidity, special financing
rates, taxes, credit risk), two securities (or portfolios of securities) with exactly the same cash
flows should sell for the same price.
Arbitrage is technically an instantaneous risk-free profit. Therefore, the law of one price implies
no arbitrage.
18.4 Which of the bonds carries the highest yield (YTM; assume semi-annual compounding as this
refers to Tuckman!)?
The bond-equivalent yield of the first bond = 2 * RATE(2,2.875/2, -98.4,100) = 2 * 2.265 =
4.530%
In calculator terms:
N = 2,
PV = -98.4,
PMT = 2.875/2,
FV = 100
CPT I/Y = 2.2647% (semiannual)
Similarly, the yield of the second bond = 4.711% and the third bond = 4.902%.
So, the third bond offers the highest yield-to-maturity (YTM).
18.5 (bonus) Can different yields be explained under the no-arbitrage assumption, or is than an
internal inconsistency (error) in the question?
Yes, because the yield essentially impounds the rates into a single flat yield curve.
But the six month rate can be different from (lower than) the annual rate. Put another way, a
non-flat yield curve can explain different yields (YTM).

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 34

Question 19: Duration and convexity [valuation]


A newly issued non-callable, fixed-rate bond with 30-year maturity
carries a coupon rate of 5.5% and trades at par. Its duration is 15.33
years and its convexity is 321.03.Which of the following statements about
this bond is true?
a) If the bond were to start trading at a discount, its duration would decrease.
b) If the bond were to start trading at a premium, its duration would decrease.
c) If the bond were to start trading at a discount, its duration would not change.
d) If the bond were to remain at par, its duration would increase as the bond aged.
19.2. What is the bonds modified duration?
19.3. What is the bonds DV01? Use this DV01 in a sentence to give the DV01 a definition.
19.4. Use duration and convexity to approximate the change in bond price given a 1% drop in
yield.
19.5 [tough] If this bond were to start trading as a discount, would an increase in maturity imply
an increase in duration and DV01?

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 35

Answer: A
Explanation:
a. is correct. At higher interest rates, the bond/price relationship is closer to linear than it is
when rates are low. So, the new duration would be lower than 15. Alternatively, one can think of
duration as a weighted average of the times when cash flows are made, where the weights are
the percentage of the total value of the bond. When rates rise, the present values associated with
the later payments are relatively smaller and the duration falls.
b. is incorrect because it is the exact opposite of a, the correct answer.
c. is incorrect. It fails to recognize the logic stated in a.
d. is incorrect because duration is mainly a function of duration and, all else constant, duration
would decrease as the bonds maturity shortened.
Topic: Valuation and Risk Models. Subtopic: Duration and convexity. Reference: Bruce Tuckman,
Fixed Income Securities, 2nd edition, Chapter 5
19.2. What is the bonds modified duration?
Assuming semi-annual compounding, modified duration = Macaulay duration / (1 + yield/2).
Because the bond trades at par, the yield = coupon = 5.5%. Therefore, modified duration =
15.33/(1+5.5%/2) = 14.92
19.3. What is the bonds DV01? Use this DV01 in a sentence to give the DV01 a definition.
DV01 = Price * Mod Duration/10000. In this case,
DV01 = 100 * 14.92/10000 = $0.15 (i.e., $0.1492)
If the yield (YTM) drops by one basis point (bps), in a parallel shift of the yield curve, the
bond price will increase by approximately $0.15
...note we are generally referring to a yield-based DV01 as in the rate shocked is a YTM.
19.4. Use duration and convexity to approximate the change in bond price given a 1% drop in yield.
Estimate change = duration impact + convexity adjustment. In this case,
=(-) (14.92)(-1%) + (321.03)*(-1%)^2*(1/2) = +14.92% + 1.62% = +16.54%
please note the convexity adjustment is always a positive addition; i.e., duration alone will
understate the actual bond price in both up and down yield movements (the convexity gap, for a
bond without embedded options, is always positive)
19.5 [tough] If this bond were to start trading as a discount, would an increase in maturity imply
an increase in duration and DV01?
Duration (Mac or modified) is always an increasing function of maturity, so for the discounted
bond, an increase in maturity would indeed increase the duration.
However, DV01 is infected by a price effect in addition to a duration effect. For the original par
bond, maturity extension will increase duration; but as a discounted bond, extensions to
maturity DECREASE PRICE (think pull to par but in reverse!). Therefore, the impact on DV01 is
ambiguous.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 36

Question 20: Historical simulation [valuation]


Rational Investment Inc. is estimating a daily VaR for its fixed income
portfolio currently valued at USD 800 million. Using returns for the last
400 days (ordered in decreasing order, from highest daily return to
lowest daily return), the daily returns are the following: 1.99%, 1.89%,
1.88%, 1.87%,, -1.76%, -1.82%, -1.84%, -1.87%, -1.91%.
At the 99% confidence level, what is your estimate of the daily dollar VaR
using the historical simulation method?
a) USD 14.08mm
b) USD 14.56mm
c) USD 14.72mm
d) USD 15.04mm
20.2 Is the answer given the only correct answer? If not, give alternative(s).
20.3 We could also calculate the historical volatility (X%) of the 400 days series and compute a
99% VaR as given by 2.33*X%. What is the disadvantage of such a parametric approach?
20.4 As a risk manager, you perceive recent days (which have been more volatile) to be more
relevant to the VaR estimate than distant days (which were relatively calm). Propose a
modification to the approach.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 37

Answer: B
Explanation: VaR = 1.82% * 800 = 14.56 million
20.2 Is the answer given the only correct answer? If not, give alternative(s).
No, the answer given is not the only correct answer. Per Dowd, the 1% VaR HS is the
observation given by (1-confidence)*n + 1. In this case, the 99% VaR refers to the 5th highest
loss observation (i.e., if sorted where losses are positive). Dowds logic is simply that
observations in the tail constitute 1%; in this case, the four worst losses constitute the 1% tail.
please note this helps us state VaR in the following way: at least 1% of the time, we expect
losses to be worse than 1.76%
under Dowds approach, therefore, answer (a) of 14.08 MM is also correct!
Further, the answer given by =PERCENTILE (array, 1%) is also acceptable although difficult to
quickly calculate. PERCENTILE (this array, 1%) = 1.7606%
20.3 We could also calculate the historical volatility (X%) of the 400 days series and compute a
99% VaR as given by 2.33*X%. What is the disadvantage of such a parametric approach?
2.33 is the deviate implied by a normal distribution at under the one-tailed 99% VaR; i.e., 2.33 is
a normal deviate. The first problem, then, is that we have IMPOSED a NORMAL distribution on
data that is likely non-normal (e.g., may be kewed or heavy-tailed)
20.4 As a risk manager, you perceive recent days (which have been more volatile) to be more
relevant to the VaR estimate than distant days (which were relatively calm). Propose a
modification to the approach.
You can use the hybrid approach (Linda Allens term) which is the exact same as Dowds AGEWEIGHTED HISTORICAL SIMULATION. In this approach, the weight assigned to historical return
days will decline, per EWMA, in constant proportion so that more recent days receive more
weight. The 1% is then unlikely to be the 4th or 5th worst observation as it would be under
simple HS.

www.bionicturtle.com

FRM 2010 LEVEL 1 (QUESTIONS 1-20) 38

You might also like