ERM quantitative risk analysis methods and techniques applied to
a small commercial bank
Ricardo Matos Chaim
Murilo Castellano
Since modeling and risk management must be viewed as a tool to improve
business performance, data and modeling tools are required to support financial risk
quantification and capital allocation. Market developments and enhanced regulations
require new techniques in order to improve Asset and Liability Management..Despite
the importance of risk evaluation, lack of reliable public information, the singular
probabilistic behavior of the return (or the loss) of market and credit risks and the
underlying nature of the business leads to a complexity that is difficult to handle without
a combination of methods and techniques that could together give a systemic view of
the problem. Based on a research over a 10 year data base, a methodology will be
detailed to quantify financial risks based on the combination of methods and techniques
such as parametric v@r, historical and monte carlo simulations, Bayesian inference and
game theory. The aim of the paper is to put together the techniques and describe the
usefulness of each one in order to develop a SD policy model that can use many
insights and informations from them.
1 Introduction
The primary goal of a financial institution is to maximize the outcomes
considering the difference between the return obtained from the investments and the
cost in fund raising as well as the revenue from direct intermediations to take a good
liquidity and solvency over time. In order to achieve this purpose the institutions take
risks, singling out the Market Risks, Credit Risk and Operational Risk, which belong to
Pillar | of the Basel Accord Il.
If on one side competitive forces have pushed financial institutions towards more
aggressive investment strategies, on the other side, although there is a wide variety of
definitions, risk management is generally related to assuring good solvency and liquidity
behavior according to Asset-liability management (ALM) approach.
More frequently, ALM terminology is generally used by Banks and
Insurance Companies to prevent the exposure to fluctuations in the ratio beyond
prescribed limits whichever they are: GDP, interest rate, stock market index, inflation
rate and so on. These random variables require the modeling process to consider
stochastic methods and techniques to give them a systemic approach in order to better
manage inherent risks and uncertainties that financial institutions might face.
The assets of a company should be more than the liabilities in its book. The
liability could be current ones (payables to manage & run business) or just the exposure
that could follow due to an incident.
2 Background
Because financial institutions are interconnected to others worldwide, an
international standard can help protect the international financial system from the types
of problems that might arise if a major bank ora series of banks collapse.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
Issued by the Basel Committee on Banking Supervision (BCBS), the purpose of
Basel Il Accord was to create an international standard that banking regulators can use
when creating regulations concerning how much capital banks need to put aside to
guard themselves against the types of financial and operational risks they face.
Among the several reasons for the existence of banking regulations, we single
out the following:
* Financial mediation — the main role of banks operating in the connection
between the saver agents and the resource taker agents;
* Bank racing — lack in credibility that a certain bank will honor its
commitments, making the clients withdraw amounts greater than its
capacity to honor the short term withdrawals;
* Contagion effect — the regulation tries to avoid situations in which
difficulties in a certain institution pose a risk to the entire financial system,
the so-called systemic risk;
+ Protection to depositors - comes from the necessity of the regulation
organ ensuring protection to depositors by intervening in a bank that
shows difficulty in honoring its commitments.
* Soundness in the payment system — looks for reliability in the banking
system to ensure the liquidation of all operations.
The first coordinated step taken by the regulatory organs to impose a more strict
risk administration came up in 1974 with the creation of the Basel Committee, which
was made up of financial authorities from G-10 countries, where issues related to the
bank industry would be discussed with the aim of establishing behavior patterns,
improving the quality of banking supervision, ensuring stability, and tightening the
security of the international banking system.
In J uly/2008 after a long discussion process the Basel Accord was signed. The
Basel Accord established and standardized the use of risk factors for assets and the
requirement of minimum capital compatible with the credit risk taken by the banking
institutions of the signatories in their operations.
The main objective of the accord was to ensure the stability and the soundness
of the international banking system through the establishment of a minimum standard
for capital requirement, therefore creating homogeneous conditions of competition
among international banks by harmonizing the rules.
The concept of capital of a financial institution is set by the Accord of 1998, which
is known as Basel I, as follows:
Level 1 Capital or Main — Shareholders Capital + Reserves (retained profits)
Level 2 Capital or Supplementary -— Other Reserves (not published,
reevaluations, etc.) + general provisions + hybrid tools of capital and subordinate debts.
It cannot exceed 100% of the level 1 capital and the subordinate debts are limited to
50% of the level 1 capital, since hybrid tools and the subordinate debt are banking
products with both characteristics of debt and capital.
Subsequently, in 1996, an amendment to the Basel Accord additionally
determined the requirement for capital adequacy related to the level of exposure to
market risk taken by the banks from the members.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
Thus in J anuary/1996, an addendum to the Basel |, called Amendment to the
capital Accord to incorporate Market Risks was published. It has the following relevant
aspects:
* Increase the control over risks taken by banks;
+ Extend the requirements for the definition of minimum capital (or
regulatory), embodying market risk;
* possibility of using the internal models in the measurement of risks as long
as they are approved by the regulatory organ;
* creation of level 3 capital, which corresponds to the bonds of subordinate
debt with maturity under 2 years;
In 1994, the Central bank of Brazil published the Resolution 200/94, which
established the minimum values of capital as well as the obligation of maintenance by
the banking institutions of minimum assets compatible with the credit risk of the active
operations through the use of deliberation factors.
In 1997 was created the Credit Central and through the Resolution 2.399 of the
National Monetary Council, capital for covering the credit risk in swap operations was
required.
In 1998 was established the implementation of internal controls of the activities of
financial institutions.
In 1999, the requirement of capital for covering exchange and gold
(resolution 2606) was established. It was set the maximum level of usage of Reference
Asset (RA) related to the permanent fixed asset establishing that the financial
institutions should classify the credit operations in a decreasing order and also verify the
provision for credits with doubtful liquidation (Resolution 2682).
The Circular 2.972/00 established criteria and conditions for the
verification of the asset quota required for covering the interest rate risk because of the
exposure of remunerated operations based on pre-fixed rates.
The System of Credit Information was created, thus replacing the Central of
Credit Risk. The criterion to control liquidity risk was also established.
In 2001 the Resolution 2.837 defined the Reference Asset (RA) as the sum of
levels 1 and 2;
In J une/2004 the Committee published the New Basel Capital Accord —
Basel Il, which is structured under 3 pillars:
¢ Pillar 1: Capital — requirement of capital for the Market Risk, Credit Risk,
Market/Liquidity Risk and Operational Risk.
+ Pillar 2: Revision through Supervision
¢ Pillar 3: Market Discipline
Still in 2004, BACEN published the Bulletin 12.476, which established a
chronogram of implementation of Basel II in Brazil. It was later adjusted by the Bulletin
16.137/2007.
Despite of having registered an advance related to the settlement of a
capital requirement standard to Brazilian institutions, so far the demand of capital
included only the credit risk and the market risk associated with pre-fixed positions and
indexed to currency exchange or to gold. The positions coupled with inflation rates and
other indexers, notwithstanding implying in relevant risks were not considered. In order
to eliminate such deficiency, the Central Bank published a set of Resolutions and
Circulars, which came into effect in J uly 2008.
In addition to the improvement in the calculus of the minimum capital requirement
for market risk, the Central Bank of Brazil determined the implementation by the
financial institutions of management structure of market risk, adapting the Brazilian
Financial System to the recommendations approved by the New Basel Accord.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
According to FSA (2012),
the crisis in financial markets over 2008 and 2009 prompted a strengthening of
the Basel rules to address the deficiencies exposed in the previous set of rules. The
Basel Ill proposals sought to strengthen the regulatory regime applying to credit
institutions in the following areas.
e Enhancing the quality and quantity of capital.
e Strengthening capital requirements for
counterparty credit risk (and in CRD III for market
risk) resulting in higher Pillar | requirements for
both.
e Introducing a leverage ratio as a backstop to risk-
based capital.
e Introducing two new capital buffers: one on capital
conservation and one as a countercyclical capital
buffer.
e Implementing an enhanced liquidity regime
through the Net Stable Funding Ratio and
Liquidity Coverage Ratio.
The Basel Ill proposals are a long-term package of changes that are due to
commence on 1 January 2013 and, based on the Commission’s timetable, the transition
period is expected to run until 2021.(FSA, 2012).
Therefore, in Brazil, the financial institutions must comply with the rules and
procedures established by the normative guidelines as in appendix |.
3 ERM risks
Once risks are voluntarily taken by banks with a view to benefiting from the
exposures, ALM could be defined as the ongoing process of formulating, implementing,
monitoring, and revising strategies related to assets and liabilities in an attempt to
achieve financial objectives for a given set of risks, uncertainties and constraints.
Because of the wide range of risks and uncertainties they face, the lack of
reliable public information, the singular probabilistic behavior of the return (or the loss)
of market and credit risks and the underlying nature of the business convey a
complexity that is difficult to be handled by the analyst without a combination of
methods and techniques when he is observing many aspects of the sample.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
The traditional risk paradigm views risks in terms of frequency and severity; the
higher the severity and frequency, the more critical the risk. It tends to be a reactionary
model in which the risk manager handles the claim and evaluates its impact because it
denies the analyst the ability to fully analyze the event. This kind of model views risks as
being synonymous with loss while ERM considers risks not merely as a problem but
also as an opportunity.
An enterprise risk management program is frequently based on management
software and computer modeling programs that provide the opportunity to track and
analyze losses both on a historic basis and in a hypothetical setting.
Under an ERM model, risk is viewed in a holistic fashion where everything in your
risk universe is connected to everything else and the risk manager's activities would be
much more comprehensive, including evaluating the activity in question and the
potential risks associated with it; refining job descriptions and requirements based on
the anticipated risks of the job; and exploring how to modify activities and policies
providing the facility to avoid future injury.
The growing complexity of financial products and the greater promptness
of businesses in the markets made necessary the adoption of market risk management
practices by the financial institutions, aiming the improvement of results as well as the
reduction of losses.
It is important to highlight the recommendation made by the Basel
Committee of Banking Supervision regarding the determination of minimum capital
requirement compatible with the risks taken, including the market risk, thus making the
financial institutions calculate the level of risk exposure.
It is important to single the Market Risk out, which might happen due to the
possibility of losses resulting from the fluctuation in the market values of positions held
by a financial institution, oscillations which might happen in the interest rates, exchange
rates, stock prices, or in the price commodities.
Structured models for the calculation of minimum capital requirement for
market risk are based on VaR according to the recommendation of the Basel
Committee of Banking Supervision. Since it is easy to understand, VaR is used by most
financial institutions as the main tool for measuring and limiting the market risk.
Despite the easiness for interpretation and use, VaR is less adequate for
risk measurement under severe instability in the markets, as in crises for instance.
Thereto, stress tests are adopted aiming at inferring the possibility of losses from abrupt
oscillations in the prices of assets, allowing the adoption of preventive measures. Such
tests are based on the use of scenarios, which might be deterministic or based on
simulation techniques.
Despite the advances made concerning the modeling and the
measurement of market risks, it is necessary to pursue the improvement of techniques
so that the process of risk management minimizes the losses from adverse market
movements, increasing the probability of generating foreseen flows.
4 Financial modeling dynamics
In order to make rational decisions, it is necessary to quantify uncertainty through
a risk analysis model such as a factor model based in many variables that interact to
each other and with information that can be obtained by available data or by expert
opinion.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
To Pézier (2004), the financial models fall into three main categories:
(i) probabilistic/statistical models describing uncertainties about the future values
of market factors; (ii) pricing models relating the prices and sensitivities of instruments to
underlying market factors; and (iii) risk aggregation models evaluating the corresponding
uncertainties on the future values of portfolios of financial instruments.
In the first category are the stochastic processes commonly used to describe the
evolution of market factors: geometric Brownian motion, stochastic volatility models,
GARCH models etc. A prime example of the second type is the Black-Scholes option
pricing model which, for a given choice of dynamics for the underlying asset price, adds
some efficient market assumptions and a hedging argument to yield a risk-free option
price. The third type is exemplified by value-at-risk (VaR) models which, with the help of
a few simplifying assumptions, produce a probability distribution (or at least some
statistics) on the future value of a static portfolio at a chosen future time. (PEZIER,
JACQUES. in ALEXANDER & SHEEDY, 2004).
Anderson & J ohnson (1997, p. 6-10) state that variables could be analyzed on an
event, a pattern or a structure way. While events are pictures of a single moment in
time, patterns are trends, or changes in events over time, while “thinking at the
structural level means thinking in terms of causal connections. For the authors, actions
taken at the structure level are creative, because they help to shape a different future,
but they recommend that the best actions remain focused on the present at the event
level. For them, “the art of thinking at the systemic structure level comes with knowing
when to address a problem at the event, pattern, or structural level, and when to use an
approach that combines the three”.
ACTION | TIME WAY OF | QUESTIONS YOU
MODE ORIENTATION | PERCEIVING | WOULD ASK
EVENTS React! Present Witness “what’s the fastest way
event to react to this event
now?”
PATTERNS Adapt! 1 Measure or | “what kinds of trends
track patterns | or patterns of events
of events seem to be recurring?”
STRUCTURE | Create Future Causal loop | “what structures are in
Change! diagrams and | place that are causing
other systems | these patterns?”
thinking tools
Figure 1: Levels of understanding
Source: Anderson & J ohnson (1997, p. 9)
Observed data may come from many sources which may be combined: scientific
experiments, surveys, computer databases, literature searches, computer simulations. It
is assumed that the analyst reviews the available data and the characteristics of the
variable that is to be modeled.
5 The methodology
Empirically it is possible to define the fitted theoretical distributions that of
observed data by an intuitive approach or by using techniques such as the maximum
likelihood estimators, goodness-of-fit statistics and plots. Most risk analysis models are
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
a combination of adding (subtracting) and multiplying variables together and their
results generally seem to be somewhere between normally and lognormally distributed.
Quantitative risk analysis requires several techniques that could aid to explain
data, particularly Bayesian inference and Monte Carlo simulation.
In order to assimilate the various uncertainties of a problem or to produce a
realistic view of the problem’s total uncertainty, Bayesian statistics are employed to
determine posteriori distributions of a variable by combining a priori opinion with
observed data. The strong law of large numbers is the principle upon which Monte Carlo
simulation is built, e.g. the larger the sample size, the closer their distribution will be to
the theoretical distributions.
After computing the value of the t statistic from two samples, Monte Carlo
simulation is used when there is interest in learning the true significance level for the t
statistic when the populations don’t follow the standard assumptions of normality and
equal variances as the true significance level will depend on
the stated level of significance a
the shape of the populations (normal, skewed, heavy-tailed, etc.)
the spreads of the two populations as measured by the two standard deviations
the sample sizes m and n
To Albert (2007), given a particular choice of a, shape, spreads, and sample
sizes, we wish to estimate the true significance level given by
al = P(ff/2tn+m-2,a/2).
A simulation algorithm to compute aT could be:
* Simulate a random sample x1, ..., xm from the first population and y1, ..., yn
from the second population.
* Compute the t statistic T from the two samples.
+ Decide if |T| exceeds the critical point and HO is rejected.
One repeats steps 1-3 of the algorithm N times and estimates the true
significance level by the formula:
AP ws number of rejections of Ho
N
It can be used to explore the pattern of the true significance level aT for different
choices of sample sizes and populations.
The Process of Measurement of Market Risk proposed is based on the use of
widely spread tools established by the best practices of market risk management,
including all the Bank positions. The process includes the following procedures:
(a) Market to Market of the positions (MTM)
(b) Calculation of VaR (Value-at-Risk), standard deviation, portfolio quality
ratio, etc.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
(c) Measurement of values of the positions sensitivities concerning the
fluctuation of interest rates.
(d) Stress Tests;
(e) Verification of gaps
(f) Backtesting; and
(g) Calculation of capital alocation for market risk, that is, required reference
Asset.
Market to Market is the process through which a financial asset is evaluated at
market value, thatis, at the probable value of purchase/sale where the buyer and the
seller are independent, not being a compulsory trading or resulting from a liquidation
process.
Market is understood as the one where the asset is traded ordinarily and frequently.
It must be selected among the available markets where the greater trading volume of
the asset occurs.
The Market Risk is directly related to the fluctuations of trading values of financial
assets, which may be measured according to the verification of the volatility of their
prices.
Besides being a crucial element in the process of risk measurement, market to
market allows the verification of real exposure of a position of a specific factor of risk or
market.
The appropriate market to market method to be used depends on the intrinsic
characteristics to each asset, especially concerning the related risk factors and
optionality, if itis the case.
Financial tools with high liquidity must be market to market based on the trading
prices effectively seen in the market.
The following primary sources was used as reference of prices and rates:
- Federal Government Bonds — ANBIMA (www.anbima.com.br);
- Private Bonds — Debentures — ANBIMA;
- Private Bonds — Other — Own Methodology;
Stocks, options over net stocks, stock terms -— BOVESPA
(www.bmfbovespa.com.br);
- Future contracts, swaps, commodities (agribusiness) - BM&F
Financial tools with low liquidity must be market to market based on the use of
market parameters observed for tools with high liquidity, of at least nature, term, risk
and similar indexers.
For tools with no liquidity in the market, the market to market process will take place
according to the present value in the future cash flows to be obtained, adjusted based
on the expectations of valid market interest rates, on the day of the precification.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
The following sections thoroughly describe the methodologies used in the market to
market process.
(A) PRE-FIXED TOOLS: The unit price at market value of a pre-fixed financial tool on a
t date corresponds to the nominal value on the maturity date (redemption value),
deducted by the pre-fixed market interest rate defined according to trading performed in
the future exchanges DI of 1 day, from date t until maturity, as follows:
WNC, , VN
PUMTM, = )——1_ + —*
" den) (1+ 1,22
PUMTMt: Unit price market to market on date t,
VN: Nominal Value
T: Maturity date of the bond
T: Calculation date
Ci: Interest coupon paid periodically, if itis the case
du: Number of business days between the calculation date and the maturity
date
1: Pre-fixed interest rate applied to the term |;
(B) TOOLS INDEXED TO CDI/SELIC
The unit price concerning market to market on date t- PUMTM'- will be
defined by the projected redemption value, considering the CDI/CELIC rate of 1 day, on
annua’l basis, published by CETIP/BACEN (www.cetip.com.br) until the calculation date
— t-— and from the calculation date until the maturity of the asset, on date T, the
CDI/CELIC rate of 1 day, on annual basis is defined according to the trading in BMGF in
the future exchanges of DI interest rate of 1 day, calculated with the respective issuing
percentage, deducted by a new factor created from the trading date until the maturity
with the market percentage traded and the expectation of future interest rates.
On the trading date the percentage of DI/SELIC is defined based on market
price, which includes the quota of credit spread and the daily calculation of the asset
must reflect the variations of these percentages according to the market/liquidity and the
quality of credit of the issuer/asset.
Methodology for calculation
Definition of the PU of the asset curve on trading date t
PU, = FATOR, * VN,
Where
PU, Unit price market to market on date t,
‘acum!
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
VN — Nominal Value of issue (generally R$ 1,000.00)
FATOR, wm - Accumulated factor 1 — daily percentage factor of accumulated
CDI/SELIC from the issue date until the business day immediately before the
calculation date. The calculation of the accumulated FATOR 1 is made using
the following formula:
DI,
FATOR,
vam =T]A(14 3
Where
daily CDI/SELIC (rate expressed once a year published by CETIP), or daily
TMS (rate expressed once a year published by BACEN) — based on the asset
issue rate.
2 —1)*(X%)+1}
X % percentage defined for the Bond on issue;
t
Il the sum of the product of daily factors from date 1 (issue) until date t
1
(business day immediately before the calculation date t)
Definition of accumulated factor 2
FATOR, uno - accumulated factor 2 - daily percentage factor of accumulated
CDI/SELIC from date t, - of calculation — until date T — maturity of the asset
using the projected rate of CDI or SELIC based on traded rates in the futures
market of DI interest of 1 day of BM&F. The calculation of the accumulated
FATOR 2 is made using the following formula:
1
DI,
FATOR, am = Tet we) —1)*(X%)+
Where:
DI, - daily CDI/SELIC or daily TMS — based on the future expected interest
rate.
X % percentage defined for the bond on issue;
T
Tl the sum of the product of daily factors from date t (calculation) until date
t
T (business day immediately before the redemption date)
The unit price concerning the market to market effect on date t - PUMTM,, will
be defined by the redemption value projected for date T based on the nominal value of
issue — VN accrued to accumulated factor 1 - FATOR - and accumulated factor 2 -
‘acum
FATOR, wm. previously defined, deducted by the calculated accumulated factor from the
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
date of calculation t until the redemption date T, considering the new percentage of
DI/SELIC — Y% - of market, according to the following formula:
PUMTM
Where:
(ss * FATORcuy, *VN
=
FATOR oun, )
DI,
1
FATOR aang = Tas Fe) -1)* (%)+0}
100
(D) TOOLS INDEXED AT 100% OF CDI/SELIC +a % A YEAR
The unit price concerning market to market on date t- PUMTM, - will be defined
by the projected redemption value, considering the CDI published by CETIP or SELIC
published by BACEN, from issuing date until the calculation date — t — and from the
calculation date until the maturity date T, the expected CDI/SELIC according to
additional spread of issue a % a year, deducted by the new factor constructed from the
trading date until the maturity with the projected CDI/SELIC adding the spread B%a
year, percentage of trading in market on date t; ©
On the trading date the spread B % a year is defined according to the market price,
which includes the quota of credit spread and the daily calculation of the asset must
reflect the variations of these percentages based on the market/liquidity and the quality
of credit of the issuer/asset.
Methodology for calculation
Definition of the PU of the asset curve on trading date t
PU, = FATOR
Where
*VN,
‘acum!
PU, -Unit price on date t
VN — Nominal Value of issue (generally R$ 1,000.00)
FATOR,
acum]
- Accumulated factor 1 — daily percentage factor of accumulated
CDI/SELIC from the issue date until the business day immediately before the
calculation date
The calculation of the accumulated FATOR 1 is made using the following
formula:
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
1
FATOR an = “Tete me 1 (a% + 1)253}
Where
I,: daily CDI (rate expressed once a year published by CETIP), or daily TMS
(rate expressed once a year published by BACEN)
a%: spread in percentage a year defined on the issue of the bond
t
I : the sum of the product of daily factors of CDI/SELIC from date 1 (issue)
1
until date t (business day immediately before the calculation date t)
Accumulated factor from calculation date until the redemption date
FATOR, uno - accumulated factor 2 — given by the projected average rate of
CDI/SELICbased on tradings performed in the futures market of DI interest rate
of 1 day of BM&F accrued to the credit spread a% a year based on market
tradings
The calculation of the accumulated FATOR 2 is made using the following
formula:
1
FATOR um, = Tet mye 1 (a% +1)20}
Where
DI, - daily CDI/SELIC or daily TMS — based on the future expectations
X %: percentage defined for the bond on issue;
uy
Tl : the sum of the product of daily factors of CDI/SELIC from date t
t
(calculation) until date T (business day immediately before the redemption date)
The unit price concerning market to market on date t- PUMTM, - will be defined
by the projected redemption value for the date T based on the nominal value of issuing
— VN, accrued to accumulated factor 1 - FATOR,,, to accumulated factor 2 -
FATOR, wm, previously defined, deducted by the calculated accumulated factor from the
date of calculation t until the redemption date T, considering the new market spread B %
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
a year, given based on the trading in market on date t, according to the following
formula:
punt, =( FATORw, *FATOR oun, *VN
FATOR scum,
Where
1
T 252 1
FATOR ay = Tra(+ Fe) (8% +125}
t
(E) TOOLS INDEXED TO PRICE INDEXES (IPCA/INPC/IGP-M)
The market value of a tool indexed to price index corresponds to the issuing price
(Nominal Value), updated by the correction factor related to the variation of the index of
the issue date until the moment of evaluation, deducting the present value by the
discount rate regarding the inflation coupon related to the remaining term of the tool, as
follows:
-{worce, °°), (nee, *)
MTM => * qu F —
(Len) (1+ 4)
MTM: Market to Market
VN: Nominal Value;
t: Calculation date
T: Maturity date of the bond
C;: Interest coupon paid periodically, if itis the case
1: Reference rate for the inflation coupon for term t;
du: Number of business days between the calculation date and the maturity
date
INDICE w: Index rate on the issue date of the bond
INDICE ,, Index rate on the calculation date considering the projection for the
current month
(F) TOOLS INDEXED TO DOLLAR
The market value of a tool indexed to dollar corresponds to the issue price
(Nominal Value), updated by the correction factor corresponding to the
exchange fluctuation of the issue date until the moment of evaluation, being
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
deducted at present value by the discount rate related to the exchange coupon
to the remaining term of the tool, as follows:
PTR see PTAK, yy
1 | PTAX, ‘) (PTAK,,
MTM = +
a a
(+1) (+1)
i=0
MTM: Market to Market;
VN: Nominal Value;
PTAX 11: Dollar Ptax of the date immediately before the agreement date of the
bond;
PTAX ,, Dollar P tax of the date of calculation, for clean coupon ;
t: Calculation date;
T: Maturity date of the Bond;
C;: Interest coupon paid periodically, if itis the case;
rT: reference rate for the dollar coupon for term t;
du: number of business days between the calculation date and the maturity
date;
(G) TOOLS INDEXED TO TR
The market value of a tool indexed to TR corresponds to the issuing price
(Nominal Value), updated by the correction factor corresponding to the TR fluctuation of
the issuing date until the moment of evaluation, being deducted at present value by the
discount rate related to the TR coupon with the remaining term of the tool, as follows:
act
VNA- (1+i) 36°
dc :
(1+ 1, ) 360
VNA =VN - (l ae TR seas)” (l t TRE
MT™M =
PU: Market unit price;
VNA: updated nominal value;
TR: Reference Rate for the period between the last and the next bond
anniversary date;
TRacumulada: Accumulated reference rate from the issuing date until the last
bond anniversary date;
i: prime lending rate;
1: Reference rate coupon for term t;
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
dc: Number of consecutive days between the calculation date and the maturity
date;
dcT: Number of consecutive days between the issuing date and the maturity
date;
dca:Number of consecutive days between the last anniversary date and the
calculation date;
dcp: Number of consecutive days between the last and the next anniversary
date;
(H) TOOLS INDEXED TO Tj LP
The market value of a tool indexed to TJLP corresponds to the issuing price
(Nominal Value), updated by the correction factor corresponding to the Tj LP fluctuation
of the issuing date until the moment of evaluation, being deducted at present value by
the discount rate related to the T) LP coupon with the remaining term, as follows:
act
_ VNA-(1+i) 3
~ de
(144, ) 30
VNA =VN -(1+TJLPaouniada)
PU
MTM: Market to Market value;
VNA: updated nominal value;
TJLPacumulada: Accumulated TJ LP from the issue date until the verification
date;
i:prime lending rate;
rm: TJ LP coupon for term t;
dc: Number of consecutive days between the calculation date and the maturity
date;
dcT: Number of consecutive days between the issue date and the maturity date;
dca: Number of consecutive days between the last anniversary date and the
calculation date;
dcp: Number of consecutive days between the last and the next anniversary
date.
(H)STOCK OPTIONS
Concerning market to market the options are daily valued based on the
quotations published on the file “Daily Report” (BDIN), made available on
Bovespa website daily.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
If the option becomes illiquid or stops being published by BOVESPA, we will adopt
as an alternative pricing model the Black & Scholes model, as follows:
CALL PUT
eC =S-N(d,)—X-e"-N(d,) eP=X-e™-N(-d,)—S-N(-d,)
2
nf $)-(re
a d, =d,-ovt
Where:
C: Price of the buy option (call);
P: Price of the sell option (put);
S: Price of underlying stock;
X: exercise price of the option;
r: interest rate (risk free interest);
o: volatility of the asset object;
N(.):standard normal distribution function.
In the case of a CALL (call option), the model calculates the intrinsic value of the option,
when deducting the cash price (S) weighed by the probability ( N(d1)) of having this
value, the present value of the strike price X.e*', weighing the probability N(d, ) that
this value happens.
In the case of a PUT (put option), the model calculates the intrinsic value of the option,
deducting the present value of the strike price X.e™', weighing the probability N(d, )
that this value happens, from the cash price of the asset (S), weighed the probability (
N(d1)) of having this value.
In the calculations (d,) and (d,) the factors of volatility and time are taken into
consideration, that is, the premise of the model is that these values happen at maturity.
9. CURRENCY OPTIONS
The options are market valued based on the BM&F reference prices -
Option Premium, published on BM&F website daily.
When there is no disclosure of the quotation by BM&F, the value of the quotation
for this day will correspond to the same value of the day before.
If the market continues showing liquidity for a specific stock, they must be market
to market based on the use of the Garman-Kolhagen model, as follows:
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
eC =S-e*-N(d,)-X-e"-N(d,) #P =-S-e*-N(-d,) +X -e*-N(-d,)
In(S/X)+(r-c+ 50%) t
a d, =d,-ovt
Where:
C: Price of the buy option (call);
P: Price of the sell option (put);
S: Price of underlying stock;
X: exercise price of the option;
r: interest rate (risk free interest);
o: volatility of the asset object;
N(.):standard normal distribution function.
The model described by the formula above allows the pricing of exchange
options based on the behavior of the cash price of the currency and the exchange
coupon rate.
10. FUTURES OPTIONS
Concerning the market to market the options are valued daily based on the
BMG&F reference prices - Option Premium, published on BM&F website daily.
If the option becomes illiquid or stops being published by BM&F, we will adopt as an
alternative pricing model the Black -76 model, as follows:
eC =e"|S-e"-N(d,)—X -N(d,)] eP=e".[X -N(-d,)—-S-e"-N(-d,)|
rt 2
ieohesh
4 -—— d, =d,-ovt
Where:
C: Price of the buy option (call);
P: Price of the sell option (put);
S: Price of underlying stock;
X: exercise price of the option;
r: interest rate (risk free interest);
o: volatility of the asset object;
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
N(.):standard normal distribution function.
InitialPrice
Y exercise price
oS —r SOi&Price
Volatility cg Price > Interest rate
Parameter } Option Price
Drift Parameter
Once the paths typically diverge, moving randomly in their ‘laws of motion’ as given in
the above equations and there is a tendency for the stock prices to ‘drift’ upwards. The
degree of dispersion of the simulated stock prices - the extent to which they move away
from each other over time — is governed by the volatility
The bigger is it the more dispersed the stock prices will be at any point in the simulation.
5 the methodology applied to a small commercial bank
The methodology was applied to a small comercial bank that invests mostly in
rural credit. As an analitycal approach, VaR_ provide the simplest and most easily
implemented method to estimate based on market data histories that can be obtained
from commercial suppliers or gathered internally as part of the daily mark-to-market
process.
Figure 1 Var Consolidated diary
GROUP POSITION RISK %VAR/POSITION $MARGIN DURATION
Hybrid capital -77.326.209,33 162,35 0,00 -40,32 1,00
Credit Cards 35.093.961,78 1.791,50 0,01 753,95 15,96
CDI Rural -1.356.294.687,62 584.067,41 0,04 -244.263,70 108,72
Rural Product
Note(CPRF) 22.976.558,31 8.410,06 0,04 3.786,15 95,31
Rural credit 789.575.418,01 439.593,84 0,06 337.061,64 167,35
Abroad deposits 463.242,34 8.708,79 1,88 -854,43 0,00
Own emissions Pos -2.274.563.802,51 4.775,69 0,00 -1.186,15 170,12
Own emissions pré -89.559.240,99 —21.300,61 0,02 -9.628,51 63,96
Equalizable Assets 387.061.572,05 190.007,95 0,05 105.367,61 130,99
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
Receivable Investment
Funds (FIDCS) 22.947.394,96 17.025,29 0,07 2.911,67 0,00
Asset operation savings 127.103.980,28 106.253,02 0,08 50.283,51 137,27
OP active rp pos 43.399.211,69 60,14 0,00 22,63 368,37
Active Pre OP RP 176.807.060,89 = 250.530,72 0,14 246.872,56 397,24
OVER ASSETS 64.999.769,35 90,07 0,00 33,90 1,00
OVER LIABILITIES -242.110.185,05 508,34 0,00 -126,26 1,00
Rural Savings -597.377.232,04 17.962,53 0,00 -6.261,82 8,37
Transfers assets 907.117.266,76 659.824,45 0,07 618.556,62 534,84
Liabilities transfers -884.147.051,17 553.107,63 0,06 -520.981,79 494,43
Terms 557.205.565,28 50.026,66 0,01 19.443,91 28,57
Floating Rat. Treasuries 3.196.651.250,61 10.458,24 0,00 1.242,84 628,35
Fixed Rat. Treasuries 752.090.103,11 40.483,64 0,01 16.064,29 17,80
Securities liabilities pos -1.530.684.527,29 3.213,84 0,00 -798,23 220,94
Securities Liabilities pre -3.021.938,01 147,51 0,00 -49,15 13,60
Portfolio VaR
R$ 618.210,92
As we can see at figure 1, the bank made operations in rural credit (Rural credit
group) at fixed rates and with a term longer than their funding (CDI Rural group), and
also preset and with a shorter period. Its market risks reside mostly there and are
concentrating on the interest rate risk because of the mismatch between assets and
liabilities operated at fixed rates.
In the column RISK it is possible to see the major contributions of the financial
instruments used in the bank, highlighting the active and passive RURAL CREDIT and
CDI RURAL groups. Column $ MARG shows the marginal effect of adding the position
of the financial instrument to the portfolio.
VaR is an estimate of the loss from a fixed set of trading positions over a fixed
time horizon that would be equalled or exceeded with a specified probability. The value
of any VaR estimate will depend on the stochastic process that is assumed to drive the
random realizations of market data.
The use of VaR involves two arbitrarily chosen parameters — the holding period,
generally ten days or one month and the confidence level that depends mainly on the
purpose to which the risk measures are being put. Once the bank is using risk
measures to set capital requirements and wish to achieve a low probability of insolvency
or a high credit rating, the confidence levels required for these purposes can be higher
than those needed to meet regulatory capital requirements and is fixed at 99% of
confidence.
Figure 1 shows that the one day VaR with 99% of confidence is just R$ 618
thousand, denoting a risk aversion behavior. The bank uses the management limit of
1% of its net worth to this metric.
Currently, it is in the house of $ 300 million and the limit is therefore R$ 3 million.
It is noteworthy that at the date of the position shown in Figure 1 the rural portfolio was
not completely done, and has an asset (R$ 789.5 million) less than the corresponding
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
liability (R$ 1.356 billion). With more two months of new hires rural credit the bank will
reach full funding of the CDI RURAL group.
The bank determines the VaR separately to allow the treasury management
actions solely within that area. Figure 2 shows the credit Portfolio.
Figure 2 — Daily Var of the treasury management
GROUP
Hibrid capital
Rural Product
Note(CPRF)
Abroad deposits
Own emissions Pos
Own emissions pré
Receivabel Investment
Funds (FIDCS)
Transfers assets
OVER LIABILITIES
Liabilities transfers
Term
Actives post titles
Active pré titles
Securities liabilities pos
Securities Liabilities pré
POSITION
-77.326.209,33
22.976.558,31
463.242,34
RISK
162,35
8.410,06
8.708,79
2.274.563.802,51 4.775,69
-89.559.240,99
22.947.394,96
64.999.769,35
-242.110.185,05
-33.848.877,88
557.205.565,28
3.196.651.250,61
752.090.103,11
21.300,61
17.025,29
90,07
508,34
40,52
50.026,66
10.458,24
40.483,64
1.530.684.527,29 3.213,84
-3.021.938,01
147,51
%
VaR/POSITIO
0,00 -29,33
0,04 7.842,09
1,88 1.810,42
0,00 -862,64
0,02 19.480,81
0,07 3.510,87
0,00 24,65
0,00 -91,82
0,00 4,27
0,01 48.543,13
0,00 5.860,29
0,01 39.173,43
0,00 -580,52
0,00 -119,80
N $MARG. DURATION
1,00
95,31
0,00
170,12
63,96
0,00
1,00
1,00
1,00
28,57
628,35
17,80
220,94
13,60
The daily VaR calculated was R$ 85.595,70, a very modest cash reflecting the fact that
the bank keeps its assets and liabilities closely related in terms of maturities and
indices. The negative sign in column "$ MARG" indicates whether it is active or passive.
Note that the major assets and liabilities are of the post-fixed (indexed to the CDI) and
therefore do not generate much interest rate risk. The bank raises nearly 2.27 billion of
its network customer’s variable rates in CDI and invests about $ 3.19 billion in bonds
also post fixed in CDI. Figure 3 shows the credit portfolio.
Figura 3: Credit Portfolio
POSICAO RISCO a $MARG. DURATION
VAR/POSICAO
Credit Cards 35.093.961,78 1.791,50 0,01 627,64 15,96
CDI Rural -1.356.294.687,62 584.067,41 0,04 -185.865,62 108,72
Rural Credit 789.575.418,01 439.593,84 0,06 304.272,66 167,35
Equalizable Assets 387.061.572,05 190.007,95 0,05 87.574,70 130,99
Asset operation
songs” 127.103.980,28 106.253,02 0,08 44.681,05 137,27
OP active rp pos 41.364.335,56 57,32 0,00 20,22 362,10
Active Pre OP RP 196.114.004,45 263.753,12 0,13 256.237,95 380,65
Rural Savings -597.377.232,04 17.962,53 0,00 -5.514,96 8,37
Transfers assets 907.117.266,76 659.824,45 0,07 613.464,19 534,84
Liabilities transfers -850.298.173,29 553.105,62 0,07 -511.727,90 514,08
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
The calculated VaR was R$ 603.769,93, showing that the majority of market risk
(arising from the mismatch of term lending and borrowing positions pre-determined) is due to
credit portfolio. Note also that the funding rural savings rate which is operated post-fixed in TR
(TR + pre-fixed coupon) is distributed between the active and equalizable assets. The latter is
active in operating the fixed rate with farmers and receive supplemental government that
subsidizes rural credit. The final rate is obtained by specific formula which provides the bank
with a performance compatible with the CDI.
Figura 4 presents the parametric of ten working days, calculated at 99%
confidence for the positions being considered with banking services. BCB Resolution
3464 indicated how institutions should classify their financial positions between banking
and trading. In the bank, the vast majority is comprised of banking positions, which
carries up to maturity. The rules of the Bank recommend special care with positions
denominated trading, or that can be sold before maturity. BACEN issued specific rules
stating standard templates and formulas for calculating the interest rate risk positions
classified as trading. The formulas are a variant of the traditional method of calculating
parametric VaR where the Bank supervise parameters such as volatility and
correlations.
Figura 4: Banking Position — Parametric VaR of ten working days
%
GRUPO POSICAO RISCO VAR/POSICAO $MARG. DURATIO
Hybrid capital -77.326.209,33 856,85 0,00 220,66 1,00
Credit Cards 35.093.961,78 3.775,36 0,01 -907,38 15,96
CDI Rural -1.356.294.687,62 1.390.271,11 0,10 173.676,04 108,72
Rural Product
Note(CPRF) 22.976.558,31 17.881,24 0,08 -2.645,29 95,31
Rural credit 789.575.418,01 833.255,61 0,11 345.620,06 167,35
Abroad deposits 463.242,34 29.346,47 6,34 1.497,36 0,00
Own emissions Pos -2.274.563.802,51 25.204,32 0,00 6.490,61 170,12
Own emissions pré -89.559.240,99 52.118,37 0,06 3.382,93 63,96
Equalizable Assets 387.061.572,05 409.333,81 0,11 12.132,61 130,99
Asset operation
savings 127.103.980,28 166.734,24 0,13 -13.372,16 137,27
OP active rp pos 41.364.335,56 162,23 0,00 -118,04 362,10
Active Pre OP RP 196.114.004,45 531.531,36 0,27 517.166,52 380,65
OVER ASSETS 64.999.769,35 254,93 0,00 -185,48 1,00
OVER LIABILITIES -242.110.185,05 2.682,81 0,00 690,88 1,00
Rural Savings -597.377.232,04 65.462,11 0,01 8.520,96 8,37
Transfers assets 907.117.266,76 1.491.751,46 0,16 1.348.725,31 534,84
Liabilities transfers -884.147.051,17 1.306.081,32 0,15 1.105.133,09 494,43
Actives post titles 3.196.651.250,61 55.622,69 0,00 1.345,70 628,35
Active pré titles 752.090.103,11 81.038,04 0,01 -19.594,08 17,80
Securities liabilities pos -1.530.684.527,29 16.961,43 0,00 4.367,91 220,94
Securities Liabilities
pré -3.021.938,01 509,91 0,02 95,70 13,60
Portfolio VaR 10 days
R$ 1.281.977,73
In figure 5 below we calculate the VaR of trading positions. Central Bank of Brasil
(BACEN) called portions of interest rate risk (Pjur) for the VaR calculated by formula
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
and for positions classified as trading. The total market risk (which is all due to interest
rate) is around R $ 4.33 million, being the most results from the trading positions (U.S. $
3.05 million). In Figure 5 one can see the calculation for the vertices (salaries
standardized methodology contained in the Central Bank of Brazil.) Note that there is
enough leeway in relation to the total VaR limit allocated for market risk. In one case we
have permission to take a maximum risk of $ 11.9 million.
Figura 5 — Consolidated Risk - Banking ande Trading
PJUR 3.054.402,70 |
VERTICE POSITION RISK mute Volatility
21 231.885.812 74.180 8,66% 0,00000521
42 136.829.325 87.543 8,96% 0,00000521
63 70.098.304 67.273 9,19% 0,00000521
126 6.164.092 10.809 9,85% 0,00000476
252 444.939 1.560 10,86% 0,00000476
2520 22.947.395 967.131 12,28% 0,00000572
TOTAL 468.369.867 1.208.497
Composicao Total Risk Limit R$ % PR Freedom %PR
VaR Banking +
Trading 4.336.380 | 11.949.966| 4,60% 7.613.586 2,93%
PR February (R$) : 259.781.873
In figure 6 below we list the results of simulation of loss in condition of stress or a stress
scenario. The scenario is based on historical simulation for the past ten years, ie, the system
retums the portfolio over the past ten years, noting the worst loss the portfolio would have a day
to day based on happened in the last ten years
Figura 6: Stress Scenarios
Treasury Position Losses due to stress
Hybrid capital -77.326.209,33 1.435,24
Rural Product
Note(CPRF) 22.976.558,31 -227.201,06
Abroad deposits 463.242,34 9.151,79
Own emissions Pos -2.274.563.802,51 42.217,90
Own emissions pré -89.559.240,99 574.921,28
Receivable Investment
Funds (FIDCS) 22.947.394,96 -43.488,77
OVER ASSETS 64.999.769,35 -1.206,45
OVER LIABILITIES -242.110.185,05 4.493,78
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
Terms 557.205.565,28 -1.705.235,61
Actives post titles 3.196.651.250,61 -59.979,59
Liabilities titles pré -3.021.938,01 4,.383,11
Liabilities post titles -1.530.684.527,29 28.410,85
Active pré titles 752.090.103,11 -1.376.698,29
TOTAL 400.067.980,78 -2.748.795,82
‘ 22/03/2001 — Interest rate
CREDIT Posicao chock
Credit Cards 35.093.961,78 -57.969,42
CDI Rural -1.356.294.687,62 15.429.434,34
Rural credit 789.575.418,01 -13.012.011,16
Equalizable Assets 387.061.572,05 -5.126.946,62
Asset operation savings 127.103.980,28 -1.024.877,96
OP active rp pos 43.399.211,69 -805,53
Active Pre OP RP 176.807.060,89 -6.309.885,72
Rural Savings -597.377.232,04 498.764,85
Transfers assets 907.117.266,76 -15.170.315,19
Liabilities transfers -884.147.051,17 13.191.111,23
TOTAL -371.660.499,37 -11.583.501,18
Stress total loss -14.332.297,00
Figure 6 shows the simulation for each portfolio (cash and credit) and the result of
consolidated loss in the order of R $ 14.33 million. Note, again, that the credit portfolio is
that it would suffer more with a stress on the rate of interest (in this case, a surge of
about 6% interest rate, which occurred in mid-2002).
Finally, Figure 7 presents the graph of backtest the model in the last 252 business days,
the case for a lambda (the decay factor in the EWMA model) of 0.89.
Figure 7: backtest graf with lambda equal to 0,89.
= Variacao (Etetiva:
— Risco (Estimado:
-300.000
-1,000,000
-1500,000
-2.000,000
27.01.2009 28032009 27.05.2009 26.07 2009 24092009 23:11 2009
Data
Verified the existence of eight errors (situations where the loss was greater than
the VaR calculated by parametric model). The test of Kupiec (used to assess the
maximum number of errors allowed in the number of evaluation days in the case, 252)
indicates the limit of five errors. The test indicated the need to change the parameter
lambda to 0.94 giving more weight to past data.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
Figure 8 presents the results of the test chart showingthat five situations extrapolation of
the loss, ie it is within the limit of five errors defined in Kupiec test.
Figura 8 backtest graf with lambda equal to 0,94.
Vasiazéo
300.006
=1.000.008
-1.500.006
2.000.008
2.500.006
2 91 2009 28.93 209 27 95 2009 26.07 2009 24 99 2009 23 11 2009
Dara
7 SD Modelling
A computer model of a financial system is a way to better understand its behavior
in order to foresee its future performance and to get insights concerning the impact of
some variables over other variables that together might impact the system. The
combination of scientific methods and computing tools make it possible to get new
forms of analysis and even desirable results and thus support a trend towards more
sophisticated modeling techniques.
Risks must be quantified in order to manage it. To understand a risk model, risk
managers require knowledge of probability distributions, simulation methods and a host
of other mathematical and statistical techniques. Market model is assessed by mapping
portfolios to their risk factors and forecasting the volatilities and correlations of them. A
credit model can be assessed using firm-value models that are based on the theory of
options, or statistical and/or macro-econometric models. Probability distributions are
applied to operational risks that are particularly very difficult to quantify because the
data are sparse and unreliable and because generally estimations are made based on
expert opinions.
Financial systems are complex because it requires the modeling of the outcomes
of a money’s allocation process, the calculation of the liquidity, solvency and so forth in
order to discover emergent behaviors considering their systemic impacts upon one
variable over others.
Using computer simulation is primarily a way to modeling and understanding
social processes for experiment, proof or discovery of social phenomena. To be useful,
simulation must be related to one or more the following primary goals: to explain the
reality, to predict the reality, to develop new tools to substitute for human capabilities, to
training, to entertainment, or, to assist in discovery and formalization. A frequent used
intention of a model in insurance or finance institution is to simulate population changes
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
in order to predict or foresee the size and age structure of a populational set of
individuals or to foresee the impacts of using a particular populational table over the
costs of the pension fund or bank. See Gilbert & Troitzsch (2002, p. 1-13)
This models could aggregate statistics to be calculated or used as estimates of
the future characteristics of the population.
Based on the case described, the following SD Model can give For backtesting
and model validation, relatively lower confidence levels are
Holding Period
t
PricelndexConsumer Value at Risk Confidence level
{ AdaptTime
OS ERM Risk
PriceIndex
Capital ConsumerChange gee
Exposure
Insolvency Ratio
LiquidityEstimate
Sees asst Overalllndex
Ratio
PriceGoodAdapt
CapitalTotal Time#Markets
oO Solvency
Ch {a
Provisions
If we wish, we can estimate the VaR of the stock price by simulating a large
number of terminal stock prices S(T). We then read the VaR from the histogram of S(T)
values so generated. In addition, Monte Carlo simulation can easily handle problems
with more than one random risk Factor. Monte Carlo methods have many applications
in market risk measurement and would be the preferred method in almost any ‘complex’
risk problem.
We might be dealing with underling risk factors that are ‘badly behaved’ in some
way (e.g. because they jump or show heavy tails) or we might have a mixture of
heterogeneous risk factors. For example, we might have credit-related risk factors as
well as normal market risk factors, and the credit risk factors cannot be modelled as
normal. We might have a portfolio of options. In such cases, the value of the portfolio is
a nonlinear (or otherwise difficult) function of underlying risk factors, and might be
impossible to handle using analytical methods even if the risk factors are themselves
‘well behaved’.
We might be dealing with instruments with complicated risk factors, such as
mortgages, credit derivatives, and so forth. We might have a portfolio of heterogeneous
instruments, the heterogeneity of which prevents us from applying an analytical
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
approach. For example, our portfolio might be a collection of equities, bonds, foreign
exchange options, and so forth. Monte Carlo simulation has many advantages over
analytical approaches to calculating VaR, once it can capture a wider range of (| market
behaviour. Below, a general model to put all the thinghs together.
Accumulsted infistion 2 rs
Inflation =
Months of application
es
Interest rates
volatility Savings of
Deposit initisl value
oO
Nenth
Deposits corrected by inflation
Notige of the amount
bee 5
Dividends ae
Monetary Ass ets
Accumulation snd Maturity
7 Conclusions
For backtesting and model validation, relatively lower confidence levels are
desirable to get a reasonable proportion of excess-loss observations. For limit-setting,
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
most institutions prefer confidence levels low enough that actual losses exceed the
corresponding VaR estimate somewhere between two and twelve times per year
(implying a daily VaR confidence level of 95% to 99%). This forces policy committees to
take the size of the limit seriously, since losses over that limit can occur with a
reasonable likelihood.
For the above reasons, among others, the ‘best’ choice for these parameters
depends on the context. What is important is that the choices be clear in every context
and be thoroughly understood throughout the institution so that limit-setting and other
risk-related decisions are made in light of this common understanding.
Appendix |: Brazil normative guidelines
Brazilian financial institutions must comply with the rules and procedures established by
the following normative guidelines:
(a) | Resolution 3.464/07: Concerns the implementation of the structure of market risk
management.
(b) Circular 3.464/07: Establishes the minimum criteria to classify the operations in
Trading Book.;
(c) Resolution 3.490/07: Concerns the verification of the Required Reference Asset,
which must be calculated based on the following equation:
PRE =PEPR +PCAM +PJUR +PCOM +PACS +POPR|
e PEPR =Quota concerning the exposures weighed by the weight factor to risk
imputed to them
e PCAM =Quota concerning the risk of exposures in gold, foreign currency and
in operations subjected to exchange fluctuation.
N
Pjur,
e PJUR = #1 , quota concerning the risk of operations subjected to interest
rates fluctuation and classified in the Trading Book according to the
Resolution 3.464/07, in which n is the number of different quotas related to
the risk of operations subjected to interest rates fluctuation.
e PCOM = Quota concerning the risk of operations subjected to the fluctuation
of commodities prices.
e PACS =quota conceming the risk of operations subjected to the fluctuation of
the price of stocks classified in the Trading Book according to the Resolution
3.464/07.
e POPR =Quota concerning the operational risk
(d) Circular 3.361/07: establishes the procedures for the calculation of the quota of
the Required Reference Asset concerning the exposures subjected to the
fluctuation of pre-fixed interest rates denominated in real (PJ UR [1]), dealt in the
Resolution n. 3.490, 2007.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
(9)
Circular 3.362/07: establishes the procedures for the calculation of the quota of
the Required Reference Asset concerning the exposures subjected to the
fluctuation of the foreign currencies coupons (PJ UR [2]), dealt in the Resolution
n. 3.490, 2007.
Circular 3.363/07: establishes the procedures for the calculation of the quota of
the Required Reference Asset concerning the exposures subjected to the
fluctuation of the price index coupons (PJUR [3]), dealt in the Resolution n.
3.490, 2007.
Circular 3.364/07: establishes the procedures for the calculation of the quota of
the Required Reference Asset concerning the exposures subjected to the
fluctuation of the interest rates coupons (PJ UR [4]), dealt in the Resolution n.
3.490, 2007.
Circular 3.365/07: Concerns the measurement of interest rates risk in the
operations not classified in the Trading Book.
Circular 3.366/07: establishes the procedures for the calculation of the quota of
the Required Reference Asset concerning the risk of the exposures subjected to
the fluctuation of stock prices (PACS), dealt in the Resolution n. 3.490, 2007.
Circular 3.367/07: Consolidates the procedures for the calculation and the
formulation of the information concerning the follow-up and the control of
exposure in gold, foreign currencies and in assets and liabilities subjected to
exchange fluctuation, on consolidated bases.
Circular 3.368/07: establishes the procedures for the calculation of the quota of
the Required Reference Asset concerning the exposures subjected to the
fluctuation of commodities prices (PCOM), dealt in the Resolution n. 3.490,2007.
Appendix II: Equations
Accumulated_inflation(t) =Accumulated_inflation(t - dt) +(Readjustments) * dt
INIT Accumulated_inflation =1
INFLOWS:
Readjustments =PULSE (Inflation/100,12,12)*Accumulated_inflation
Accumulated_inflation_2(t) =Accumulated_inflation_2(t - dt) + (withdrawl) * dt
INIT Accumulated_inflation_2 =1
INFLOWS:
withdrawl = if time <Months_of_application then 0 else (pulse
(Inflation/100,(Months_of_application+12),12))*Accumulated_inflation_2
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
Liquid_Assets_Values
Liquid_Assets_Values
Total_Investments.
I[P oupancal(t) =
I[P oupancal(t - dt) + (Dividends[P oupanca] +
\[P oupanca] - Withdrawl_1[P oupanca)) * dt
I[Poupanca] =0
Liquid_Assets_Values \[F undo_Curto_P razo](t) =
Liquid_Assets_ Values \[Fundo_Curto_Prazo](t - dt) +
(Dividends[F undo_Curto_Prazo] +
Total_Investments. \[Fundo_Curto_P razo] -
Withdrawl_1[Fundo_Curto_Prazo]) * dt
INIT Liquid_Assets_Values \[Fundo_Curto_Prazo] =0
INFLOWS:
Dividends[P oupanga] =
Liquid_Assets_Values
Mensal_Inflation)
\[Poupanca}A(Interest_rates_2[P oupanca]-
Dividends[F undo_Curto_Prazo] =
Liquid_Assets_ Values
o_Prazo]-Mensal_Inflation))*0
\[Fundo_Curto_Prazo}*((Interest_rates_2[Fundo_Curt
Total_Investments. \[Poupanca] =IF TIME >Months_of_application
THEN 0 ELSE Adjustment__of_the_deposit[P oupanca]
Total_Investments. \[Fundo_Curto_Prazo] =IF
TIME>Months_of_application THEN 0 ELSE
Adjustment__of_the_deposit[F undo_Curto_Prazo}*0
OUTFLOWS:
Withdrawl_1[P oupanca] =IF TIME <Months_of_application THEN 0 ELSE
Withdraw_Value[P oupanca]
Withdrawl_1[Fundo_Curto_Prazo] =IF TIME<Months_of_application THEN 0 ELSE
Withdraw_Value[F undo_Curto_Prazo]*0
Actual_Age =40
Adjustment__of_the_deposit[P oupanca] = if Deposits_corrected_by_inflation=1 then
(Deposit_initial_value[P oupanca}*Accumulated_ inflation) else
Deposit_initial_value[P oupanca]
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
Adjustment__of_the_deposit[F undo_Curto_Prazo] = if
Deposits_corrected_by_inflation=1 then
Deposit_initial_value[F undo_Curto_P razo}*Accumulated_ inflation else
Deposit_initial_value[F undo_Curto_Prazo}*0
Age =Actual_Age+(TIME )/12
Alert =|F Total_Payments+1 <Intended_Withdraw AND TIME >Months_of_application
THEN 0 ELSE 1
Amount_alert = Monetary_Assets_Accumulation_and_Maturity-
Withdrawl_1[P oupanca]+Dividends[P oupanca]
Annual_inflation = 12.6825
Deposits_corrected_by_inflation =0
Deposit_initial_value[P oupancga] = 1000
Deposit_initial_value[F undo_Curto_Prazo] =0
Difference = DIVIDENDO_TOTAL-Total_P ayments
DIVIDENDO_TOTAL =ARRAY SUM (Dividends[*])
Inflation = NORMAL(Annual_inflation, Inflation_volatility/100*Annual_inflation)
Inflation_volatility = 10
Intended_Withdraw = 1000
Interest_rates[P oupanca] =0
Interest_rates[F undo_Curto_Prazo] =0
Interest_rates_2[P oupanca] =NORMAL (Interest_rates[P oupancaJ/100,
(Interest_rates_volatility____ Savings_of[Poupanca]/100)*(Interest_rates[P oupanca]/100
)
Interest_rates_2[Fundo_Curto_Prazo] =NORMAL
(Interest_rates[F undo_Curto_Prazo]/100,
Interest_rates_volatility____ Savings_of[Fundo_Curto_Prazo]/100*(Interest_rates[F undo
_Curto_Prazo]/100))
Interest_rates_volatility____ Savings_of[Poupanca] =0
Interest_rates_volatility____ Savings_of[Fundo_Curto_Prazo] =0
Mensal_Inflation = (((Inflation/100)+1){1/12))-1
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
Monetary_Assets_Accumulation_and_Maturity =
ARRAYSUM(Liquid_Assets_Values_ I-*))
Month =time
Months_of_application = 120
Notice_of_the_amount = MAX (1,
(Monetary_Assets_Accumulation_and_Maturity/(Difference+0.001)))
Savings_Withdraw = 100
Total_Payments =ARRAYSUM(Withdrawl_1[*])
withdrawl_updated_by_inflation =0
Withdraw_Value[P oupanga] = if Time<Months_of_application then 0 else if
withdrawl_updated_by_inflation=0 then Intended_Withdraw*(Savings_Withdraw/100)
else Intended_Withdraw*(S avings_Withdraw/100)*Accumulated_inflation_2
Withdraw_Value[F undo_Curto_Prazo] =if time<Months_of_application then 0 else if
withdrawl_updated_by_inflation=0 then Intended_Withdraw*(1-S avings_Withdraw/100)
else Intended_Withdraw*(1-S avings_Withdraw/100)*Accumulated_inflation_2
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
REFERENCES
Albert, Jim. Bayesian computation with R. Baltimore-USA, Springer Science, 2007
Alexander, Carol & Sheedy, Elizabeth. The professional risk managers’ handbook.
PRMIA (The professional risk Managers’ International Association), 2004.
Anderson, Virginia; J ohnson, Lauren. System Thinking Basics. Whaltam, MA, Pegasus
Communications ,Inc, 1997.
Boccara, Nino. Modeling complex systems. New York, Springer —Verlag, 2004.
Bohner, Shawn A; George, Boby; Gracanin, Denis; Hinchey, Michael G. Formalism
Challenges of the Cougaar Model Driven Architecture. M.G. Hinchey et al. (Eds.):
FAABS 2004, LNAI 3228, pp. 57-71, 2005
Castellano, Murilo. Gestao de Riscos por meio de derivatives. Sao Paulo, Editora Atlas,
2009.
FSA (Finantial Service Authority). Available at: htto://www.fsa.gov.uk/about/
what/international/basel.
Pézier, Jacques. In Alexander, Carol & Sheedy, Elizabeth. The professional risk
managers’ handbook. PRMIA (The professional risk Managers’ International
Association), 2004.
Gilbert, Nigel; Troitzsch, Klaus G. Simulation for the social scientist. Philadelphi, Open
University Press, 2002.
ANDIMA. Agribusiness Securities: CPR - Rural Product Note / National. Association of Financial
Market Institutions - Brazil, OTC Clearing House, Rio de Janeiro: ANDIMA: CETIP, 2009.
60 p.; (Special Report. Funding Products) available at
http://www.andima.com.br/english/publications/arqs/cpr-rural_product_note.pdf
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF
AUTHORS’ BIO
Ricardo Matos Chaim obtained a PhD degree in Information Science at the
University of Brasilia, Brazil. Nowadays, he is an Associate Professor at the
Software Engineering Department of the University of Brasilia. He also worked
for 19 years at a Brazilian Governamental Social Insurance Company. His
current scientific interests center upon information management, risk
management, Systems Engineering, information technologies applied to
Insurance and methods to model risks, uncertainty and imprecision in financial
organizations and Pension Funds.
Murilo Castellano, MsC, is an Associate Professor of the Getulio Vargas
Foundation, Brasilia, specialized in market and credit risks, worked as an
Auditor and Risk manager at Banco do Brasil. He is the risk manager of
Bancoob. His current scientific interests center upon derivatives market,
enterprise risk management, quantitative methods applied to finance,
mathematical analysis and other financial and stock market themes.
Work supported by Fundacao de Apoio ‘a Pesquisa do Distrito Federal - FAPDF