Quantitative Risk

Management
NGUYỄN THỊ LIÊN
HOÀNG THỊ THÙY LINH
KHOA TOÁN KINH TẾ
ĐẠI HỌC KINH TẾ QUỐC DÂN

06/01/2024
CHAPTER 1. ESTIMATING
MARKET RISK MEASURES
 Learning objectives

Estimate VaR Estimate VaR Estimate Define

using a using a coherent
the
risk
historical parametric expected measures
simulation approach for shortfall
approach both normal and given P/L or
lognormal return return data.
distributions.

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 Learning objectives

Estimate risk Evaluate Interpret Q Q plots

measures by estimators of risk to identify the
estimating measures by characteristics of a
quantiles. estimating their distribution.
standard errors.

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING MARKET RISK MEASURES
DATA

Profit/Loss Data: The P/L generated by an asset (or portfolio) over the period t:

P/L t = P t + D t − Pt+1
If data are in P/L form, positive values indicate profits and negative values indicate losses.

Loss/Profit Data::
L/P t = − P/L t

L/P observations assign a positive value to losses and a negative value to profits.

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING MARKET RISK MEASURES
DATA

Arithmetic Return Data: The arithmetic return rt is defined as:

Geometric Return Data: The geometric return rt is

The geometric return is often more economically meaningful than the arithmetic return,
because it ensures that the asset price (or portfolio value) can never become negative
regardless of how negative the returns might be.
Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING MARKET RISK MEASURES
DATA
Geometric Return Data: The geometric return rt is
The geometric return is also more convenient.
For example, if we are dealing with multiple periods, the geometric return over those
periods is the sum of the oneperiod geometric returns. Arithmetic returns have neither of
these convenient properties. Calculating the geometric return has the advantage that it can
be linearized, especially when calculating for many periods:
 St 
Ln    LnSt  LnSt 1
 St-1 
 St   St St -1 St - k 1   St   St -1   St - k 1 
Ln   Ln ...   Ln   Ln   ...  Ln 
 St - k   St -1 St - 2 St - k   St -1   St - 2   St - k 

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING HISTORICAL SIMULATION VaR

Suppose we have 1000 loss observations and are

interested in the VaR at the 95% confidence level.
The HS approach Since the confidence level implies a 5% tail, we know
estimates VaR by means of that there are 50 observations in the tail, and we can

ordered loss observations. take the VaR to be the 51st highest loss observation.
More generally, if we have n observations, and
our confidence level is α, we would want the
(1 − α) ∗ n + 1th highest observation.

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING MARKET RISK MEASURES
ESTIMATING PARAMETRIC VaR: we therefore need to take
account of both the statistical distribution and the type of data to
which it applies.

Estimating VaR with Normally Distributed Profits/Losses: .

VaR(α) = −µ + σ ∗ zα

where zα is the standard normal variate corresponding to α.

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING MARKET RISK MEASURES

ESTIMATING PARAMETRIC VaR: we therefore need to take

account of both the statistical distribution and the type of data
to which it applies.
Estimating VaR with Normally Distributed Arithmetic Returns

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING MARKET RISK MEASURES

ESTIMATING PARAMETRIC VaR: we therefore need to take

account of both the statistical distribution and the type of data to
which it applies.

Estimating Lognormal VaR

Now assume that geometric returns are normally distributed with
mean µr and standard deviation σr .

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING COHERENT RISK MEASURES
Estimating Expected Shortfall
The fact that the ES is a probability-weighted average of tail losses
suggests that we can estimate ES as an average of ’tail VaRs’.

The easiest way to implement this approach is to slice the tail into a
large number n of slices, each of which has the same probability
mass, estimate the VaR associated with each slice, and take the
ES as the average of these VaRs.

The ES gives all tail-loss quantiles an equal weight, and other

quantiles a weight of 0.

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING COHERENT RISK MEASURES
Estimating Coherent Risk Measures
A coherent risk measure is a weighted average of the quantiles
(denoted by qp) of our loss distribution:

where the weighting function or risk-aversion function (p) is specified by

the user.
The ES is a special case of

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 ESTIMATING THE STANDARD ERRORS OF RISK
MEASURE ESTIMATORS

Standard Errors of Quantile

Estimators: a 90% confidence interval Standard Errors in

for a quantile q is given by: Estimators of Coherent

Risk Measures

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
 THE CORE ISSUES: AN OVERVIEW
Which risk measure? Which level? Which method?
The first and most important is to The second issue is the Having chosen our risk
choose the type of risk measure: level of analysis. Do we measure and level of
do we want to estimate VaR, ES, analysis, we then choose a
wish to estimate our risk
etc.? This is logically the first issue, suitable estimation method.
measure at the level of
because we need to know what we
the portfolio as a whole
are trying to estimate before we
start thinking about how we are or at the level of the
going to estimate it. individual positions in it?

Nguyễn Thị Liên & Hoàng Thị Thùy Linh Quantitative Risk Management
Thanks for your
listening