You are on page 1of 32

Algo Suite

THE WORLDS FINEST


RISK MANAGEMENT SOLUTION FOR
T H E E N T E R P R I S E , I N T E G R AT I N G
MARKET RISK, CREDIT RISK,
LIQUIDITY RISK, ALM AND LIMITS

Algorithmics

KnowYourRisk.

In October 1999 Alan Greenspan said,


We are striving for a framework whose
underlying goals and broad strategies
can remain relatively fixed, but
within which changes in application
can be made as both bankers and
supervisors learn more... it is the
framework we must get right.*
We believe Mark-to-Future is
that framework.

Contents
The Changing Face of Risk Management

The Benefits of An Integrated Enterprise


Risk Management System

Algo Suite: Putting the E in ERM

Enter Mark-to-Future

Elements of the Algo Suite Solution

Algo Market

Algo ALM

11

Algo Credit

15

Algo Limits

21

Risk Architecture for the Enterprise

23

*Alan Greenspan, Chairman of the US Federal Reserve Board


Greenspan, A., 1999, The evolution of bank supervision, speech to American Bankers Association, Phoenix, Arizona, October 11, 1999.
(Available at www.bog.frb.fed.us/boarddocs/speeches/1999)

CLOSING SPEED:
3 HEARTBEATS.
Some days, just crossing the
street is risky; one second the
way is clear and the next...
youre history. Close analysis
of the data will provide you
with the stop, go, walk, run
information you need.
Thats Analytics.
Algorithmics. Know Your Risk.

The Changing Face


of Risk Management

Algorithmics spearheaded the concept of


enterprise-wide risk management at a time when
even the most advanced financial institutions
did not have an enterprise risk function. Today,
we are once again breaking new ground with
the only commercially available suite of products
that can truly integrate the market, credit,
liquidity risk and ALM functions under one
methodology with one architecture. With
Algo Suite you can simultaneously reduce costs,
improve the quality of risk information and
allocate capital more efficiently.

While financial institutions have taken steps


to measure risk in an effort to satisfy regulatory
requirements and avoid losses, their efforts have
focused primarily on the calculation of market
risk for a particular portfolio. This approach,
however, does not capture the interaction of
the different types of risk that a firm faces, the
marginal impact of a trade across the institution
or the effects of offsetting market moves.
The demise of Barings, the near-collapse
of Long Term Capital Management and the Asian
monetary crisis demonstrated all too clearly the
need to measure and manage all risksmarket,
credit, settlement, liquidity, ALM, legal and
operationalon an integrated, enterprise-wide
level. Only by calculating the optimal risk/reward
trade-off across the enterprise can you remain
competitive and maximize shareholder value.

The Benefits of an Integrated


Enterprise Risk Management System
Algo Suite: Putting
the Ein ERM

Despite the need for an integrated risk measurement framework, most banks still operate under
a silo approach with separate and inconsistent
analytics, data and assumptions. These silos
reflect an era when risks were not interrelated
and when no single system was capable of
managing market, credit and liquidity risk across
both the trading and banking books.

Algo Suite is the first enterprise-wide risk


management software that captures market,
credit and liquidity risk consistently across
both the banking and trading books.
With tools to collect, map and manage data, and
to measure, manage and control risks and report
results, Algo Suite fully covers your enterprise.
It integrates risks within a single framework.

An integrated system is ultimately less costly,


provides greater flexibility and ensures
consistent decision making across the enterprise.
As a result, capital is deployed where it earns
the best risk-adjusted return. Building on an
integrated risk management system, financial
institutions can implement industry best practice
and regulatory guidelines as they evolve.

Reduced Total Cost of Ownership. Under a


traditional silo approach, multiple interfaces are
required between position-keeping systems and
strategic risk management systems. For example,
a loan is mapped not only to the banks asset
liability management (ALM) system, but also to
the credit risk management system.
As a comprehensive, integrated risk management
system, Algo Suite reduces the number of interfaces that must be developed and maintained.
This reduces the total cost of ownership in terms
of integration and on-going maintenance costs,
as well as license, support and training fees.

Integrated
Enterprise Reporting

ALM
Reporting

Trading
Market Risk
Reporting

Trading
Credit
Reporting

Banking
Portfolio Credit
Reporting

INTEGRATED ANALYTICS
STRATEGIC RISK MANAGEMENT SYSTEMS

Integrated
ERM Analytics

Asset
Liability
Management

SINGLE GATEWAY

Trading
Market Risk
Management

Trading
Banking
Credit Exposure Portfolio Credit
Management Management

RiskMapper

OPERATIONAL SYSTEMS

Loans

Swaps

OPERATIONAL SYSTEMS

Credit
Derivatives

Loans

Swaps

Credit
Derivatives

THE BENEFITS OF AN INTEGRATED ENTERPRISE RISK MANAGEMENT SYSTEM: Algo Suite (left) is
an integrated and consistent solution with reduced cost of ownership, as opposed to the traditional
silo approach (right) which involves multiple extractions, mappings and reporting solutions that
increase system costs.

Enter Mark-to-Future

Consistent Capital Allocation and Performance


Measurement. Many banks use a Risk Adjusted
Return on Capital (RAROC) framework to measure
performance and allocate capital. The results
of a RAROC model are not reliable if the risks
of various portfolios or business units are
measured inconsistently. For example, the outputs
of a RAROC model would be suspect if you
measured the market risk capital for the
structural balance sheet using a 200 bp shock
to interest rates, while the economic capital for
the trading book was based on 95% VaR using
historical simulation. In Algo Suite you can
estimate the same risk and return measures
consistently across the organization.

At the heart of our software solution, Algo Suite,


lies our unique Mark-to-Future framework
(www. mark-to-future.com). At any point in time,
the levels of a collection of risk factors completely
determine the mark-to-market value of a portfolio.
Scenarios on these risk factors determine the
distribution of possible mark-to-market values.
Scenarios on the evolution of these risk factors
determine the distribution of possible mark-tofuture values (MtF values) through time. Because
scenarios capture future uncertainty as time
unfolds, and because Mark-to-Future has
scenarios as its core input, the framework
enables the calculation of future mark-to-market
values that capture future uncertainty across
scenarios and time steps.

Integration of Market, Credit and Liquidity Risk.


In crisis situations, adverse changes in market
conditions can trigger adverse changes in credit
quality and liquidity. In fact, many well-known
financial disasters occurred precisely because
of the high correlation of market, credit and
liquidity in stress periods. For example a devaluation of an emerging market currency can lead
to credit downgrades, reduced liquidity and
a flight to quality. Silo systems cant capture
this complex inter-relationship of risks.
In Algo Suite this linkage of market, credit
and liquidity risk is natural and occurs at the
scenario level.

Mark-to-Future is essentially about two things:


Mark-to-Future is a robust and forward-looking
framework that integrates disparate sources of
risk. By explicitly incorporating the passage of
time, the evolution of scenarios over time, and
the dynamics of portfolio holdings over time,
Mark-to-Future provides a flexible and unifying
platform for assessing future uncertainty.
Mark-to-Future enables a distributed extensible
risk architecture that can be leveraged within
a single organization and across several
organizations. Mark-to-Future allows for the
de-coupling of the computationally intensive
simulation stage from the post-processing
risk/reward assessment stage.

the simulation of the new MtF values, which


are then appended to the previously computed
results. Previously simulated MtF results need
not be recalculated; only the calculation of the
risk or reward measure need be repeated.
Mark-to-Future produces risk and reward
measures that explicitly capture the passage
of time. Challenging issues such as portfolio
path dependency, settlement and reinvestment,
dynamic re-balancing and thin market effects
can be effectively addressed.

Counterparty Exposure
Stress Testing
Value at Risk

Portfolio Credit
Risk/Reward
Assessment
(Algo RiskEngine)

Earnings at Risk

MtF Cube

Scenarios

Instruments

An architecture for risk and reward. As a risk


architecture for the efficient delivery of risk and
reward measures, Mark-to-Future provides the
following benefits:

Time

MtF values need only be computed once but


can support multiple applications and/or risk
clients. Thus, Mark-to-Future is efficient and
can be leveraged.
Mark-to-Future can be implemented in a fullydistributed scalable fashion globally within
an organization. The generation of MtF values
serves as the common input to many and
various risk/reward applications at different
levels of the organization, from desk-level trader
applications to enterprise-wide applications.
With the decoupling of the computationally
intensive simulation stage from the postprocessing stage, application service providers
can provide risk service over the Web using
thin-client applications.
Mark-to-Future is an open and extensible
framework that enables the incorporation of new
scenario-generation techniques, new pricing
algorithms and new post-processing applications. Thus the framework can be extended to
new lines of business and to evolving risk
management practice. In no manner is it linked
or committed to any one particular risk/reward,
pricing or scenario-generation methodology.

MtF Simulation
(RiskWatch)

THE MARK-TO-FUTURE FRAMEWORK: The


computationally intensive simulation stage
is de-coupled from risk/reward assessment.

It also accommodates differing views of what the


future will bring and different models for pricing
assets and positions, now and in the future.
A framework for risk and reward. As a risk
framework for producing simulation-based risk
and reward measures, Mark-to-Future provides
the following benefits:
Mark-to-Future is an intuitive framework.
Scenarios are the drivers of all future uncertainty; they are the language of risk. Individual
MtF values are produced as a direct function
of a given scenario whose realization can
be explained in a straightforward manner.
Individuals with different levels of sophistication
can contribute to risk-reward discussions
through the focus on plausible scenarios.
Scenarios of risk factors define future distributions of value. As individual risk factors can
be jointly (and arbitrarily) evolved over time,
Mark-to-Future can capture the implied relations
(including the correlation) among disparate
risk factors over multiple time steps. Thus
market, credit and liquidity risks may be
integrated within a common framework.
The realization of MtF values over individual
scenarios and time steps determines any risk
or reward measure. Most importantly, the MtF
results are additive across position, scenario
and time dimensions for all instruments, linear
and nonlinear. The addition of a new position,
scenario or time step requires only

Mark-to-Market

Mark-to-Future
Upside

Regret

multiple
scenarios

Value your aged portfolio at the horizon for each


future scenario, accounting for all future events
that affect value, such as credit state changes,
settlement, reinvestment, collateral calls, etc.
It is true forward valuation using the best available
models with no shortcuts.

Elements of the
Algo Suite Solution
Algo Suite is packaged into four completely
integrated solutions: Algo Market,
Algo ALM, Algo Credit and Algo Limits.
Each of the four solutions shares common data
transformation components, common scenario
generation, valuation and risk engines, common
web-based and historical reporting components
and common extensibility tools.

Algo
Market

Market Risk,
Liquidity Risk

Algo
ALM

Credit Risk

Algo
Limits

Algo
Credit

Trading Book
Banking Book

Flexibility of choice. We designed Algo Suite


with well-defined interfaces to facilitate
integration with custom-built and third-party
risk products (e.g., limit systems, risk databases
and risk reporting systems). You can select only
those products within Algo Suite that add value
to your business. It allows you to think big but
start small. You can purchase and implement
a product for a particular site and unify and
integrate your enterprise risk infrastructure
over time.

ALGO SUITE SOLUTION OFFERINGS

Algorithmics Mark-to-Future methodology


is evolving to provide a seamless, well-organized
and comprehensive framework for all my
risk needs: market, credit, liquidity, stress
and scenario-based risk management.
In designing our next generation risk
management on the Web, MtF will be at
the heart of that architecture.
Ariel Salama
Executive Vice President
HSBC USA

Algo Market
RiskWatch offers the most comprehensive
coverage of financial products in the industry.
Over 400 types of productsincluding bonds,
equities, commodities, foreign exchange and
a wide range of derivativesfrom more than
20 different markets are supported.

While some advocate one risk measure over


another, there is growing consensus that
effective risk management requires the
application of a wide range of risk measures
and methodologies. Algo Market allows you
to assess the market risks of multi-asset class
trading portfolios using a combination of
statistical risk measurement and stress
testing approaches.

Flexibility to Incorporate Additional Models.


In addition to our on-going model development,
we also offer seamless integration with
models provided by FEA, APT, RiskCare, Intex
and Andrew Davidson. We provide a robust
Application Programming Interface (API) that
allows you to easily add proprietary valuation
models to RiskWatch.

A Range of Market Risk Measures. Value-at-Risk


(VaR) has become an increasingly important
measure for strategic risk management. Recent
market events, including the Asian crisis and
market collapse in Russia, have underscored
the importance of complementing VaR analysis
with a comprehensive stress-testing program.

A Proven Solution to Reduce Regulatory Capital.


Given the regulatory rules concerning minimum
capital requirements, many banks have sought
regulatory approval for use of internal models.
Algo Market allows you to meet the quantitative
standards as outlined by BIS: daily calculation
of 99% VaR using a 10-day holding period,
complemented by a rigorous program of stress
testing. Extensive model verification and validation
documentation provided with RiskWatch

Algo Market supports all credible risk methodologies including parametric and scenario-based
VaR, stress testing and sensitivity analysis.
Broad Product Coverage. Covering 90% of the
banks financial products simply isnt enough.
To fully understand your risk profile and comply
with regulatory requirements, your system must
cover the range of financial instruments.

also allow you to meet stringent qualitative


standards. Many banks using our software have
received regulatory approval for their internal
models for general and specific market risk.
These clients can now free up capital reserves
and generate greater profits. In one mediumsized European bank, for example, over
200 million dollars of annual capital savings
was achieved using Algo Market, a significant
return on investment in our product.

Covariance Matrix Generation. Published


covariance matrices often lack the necessary
coverage of risk factors, for example, many
emerging markets. They also offer reduced
choice over key modeling inputs such as decay
factors and the historical estimation period.

A Complete Solution
for Scenario-Based VaR

Monte Carlo Scenario Generation. In addition


to standard Monte Carlo, HistoRisk supports two
advanced Monte Carlo sampling methodologies:
low discrepancy Sobol sequences and a
principal component-based stratified sampling
method to improve the performance of Monte
Carlo simulations.

With HistoRisk, you have complete control


over the decay factors, the historical estimation
period and the set of risk factors that are most
important to your business.

Scenario-based approaches, such as historical


or Monte Carlo simulations, explicitly calculate
changes in portfolio value under scenarios to
obtain a distribution of value changes. Scenariobased VaR directly addresses the limitations
of variance-covariance methodologies as they
pertain to non-linear products and non-normally
distributed risk factors.

However, these quasi-Monte Carlo methods


cannot be applied directly in Enterprise Risk
Management since they lose their effectiveness
as the number of risk factors exceeds forty.
Accordingly, HistoRisk applies quasi-Monte
Carlo methods in conjunction with Principal
Components Analysis to reduce the risk factor
space. The software supports two Principal
Component methods. The first method reduces
dimensionality by selecting those components
that explain most of the variance of the risk
factors, without regard for the structure of
the portfolio. A second method, unique to
Algorithmics, ranks the components obtained by
Principal Components Analysis with respect to
the size of the portfolio risk exposures, not the
variance of the original risk factors.

Algo Market offers all the components required


to compute scenario-based VaR using our
MtF framework. HistoRisk provides a complete
solution to store risk factors and generate single
and multi-step historical and Monte Carlo
scenarios. RiskWatch then revalues the
dynamically evolving portfolio across these
scenarios and through time to generate a
MtF cube from which market and credit risk
measures of interest are calculated.
Validated Time Series Database of Risk Factors.
Historical and Monte Carlo VaR calculations
require a historical time series of market risk
factors to estimate future distributions of profit
and loss. VaR outputs are extremely sensitive
to risk factor time series inputs. A single error
in the time series can result in vast errors in
volatility and correlation estimates. HistoRisk also
offers a practical solution to the problem of
missing market data. There are a variety of
options for filling in market data, including
simple fill forward and backward methods, linear
and geometric interpolation and a unique method
that fills risk factor data based on its most likely
occurrence, given its volatility and correlation to
other risk factors.

Moreover, the ultimate


standard for banks to achieve
over time remains unchanged,
namely the measurement of
non-linearity through a ten-day
price shock with full revaluation
of positions [ten days forward],
but with some flexibility as
to the specific methodology
to be used.
BIS 1996 1

1 Bank

for International Settlements, Amendment to the Capital Accord to Incorporate Market Risks,
Basle Committee on Banking Supervision, January 1996

of the cash generated or required by the position.


The distribution of changes to the cash account
is a measure of funding liquidity risk from which
time-dependent liquidity VaR can be computed.

Fat-Tailed Distributions. Standard Monte Carlo


VaR techniques assume that risk factor changes
are joint normal. However, many risk factors are
observed to have fatter tails than the normal
distribution would suggest.

A Flexible Solution
for Parametric VaR

HistoRisk allows you to model these fat-tailed


distributions using a variety of methods. As a
result, distributions with wider tails are produced,
which allows you to model extreme events.

The first step in any parametric VaR calculation


is to generate the cash flow map for each
financial product in the portfolio. Many systems
perform this calculation using analytical methods.
The problem with this approach is that specific
routines are required for every supported financial
product. Accordingly, the parametric VaR implementations of many vendors do not extend
beyond simple products.

Dynamic Portfolio Evolution. The risk of a


portfolio can vary greatly over time. To measure
risk over time, its important to model the change
in portfolio composition. These changes may
occur from portfolio aging, which leads to
settlement or re-investment, with associated
funding and collateral costs. Portfolio changes
may also be the result of dynamic strategies
designed to change the characteristics of the
portfolio in response to market events.

RiskWatch uses a flexible approach to create


the cash flow map called tweaking. This
allows you to calculate a parametric VaR for
all supported financial products, even complex
ones. This approach is extremely versatile
because it can be automatically applied to
all instruments.

Utilizing MtF, Algo Market allows you to model


the dynamics of a portfolio using regimes or
dynamic portfolio strategies. This allows you to
properly measure the risk of trading desks whose
positions change over time and under scenarios.
Further, for longer time horizonsthose appropriate for strategic decisionsthe dynamics
of a portfolio cannot be ignored. To measure
risk dynamically, simulation over multiple
periods is necessary.

The RiskWatch implementation of parametric


VaR has a number of other advantages:
VaR can be decomposed into its constituent
components: interest rate, equity, volatility
and foreign exchange VaR.
interest rate VaR can be decomposed into
direction, spread and curve risk.
VaR measures can be reported in any currency
of interest (e.g., a German financial institution
could report its global VaR in Euros and the VaR
of its U.S. investment bank in dollars).
the confidence interval and holding period
are user-defined.

Liquidity Risk
Conventional VaR models often exclude one very
important component of riskasset liquidity. They
assume no liquidity during the holding period and
infinite liquidity at fair market value at the end of
the holding period. In addition, the user is forced
to select a single holding periodeven for portfolios that contain products of varying liquidity.

Stress Testing

Taking advantage of Mark-to-Future as implemented in Algo Market, risk managers can


explicitly model the risk due to asset liquidity
and evaluate alternative liquidation strategies.
Each position in a portfolio has its own holding
period and the effects of increased bid/ask
spreads during periods of market turbulence,
for example, flight to quality scenarios, can
be modeled.

A rigorous stress testing program is an essential


component of any risk management program.
While many vendors provide only a partial
solution, Algo Market provides a broad range
of choices.
Conventional Stress Testing. Algo Market
supports the effects of an instantaneous shock
in todays risk factors on the value of todays
portfolio. Scenarios can be created on any risk
factors including interest rates, credit spreads,
volatilities, foreign exchange rates and indices.
Algo Market allows considerable choice in the

To measure funding liquidity risk, dynamic


positions in each scenario path create a
fluctuating cash account. At any time point,
therefore, there is a distribution of values and

From Risk Measurement


to Risk Management

definition of dependent and independent risk


factors. For example, the BB curve can be
bootstrapped and stress tested directly to assess
spread risk. Alternatively, it can be defined as
a spread on the base government curve so that
it moves when the base curve is stress tested.

Today, its not enough to simply measure VaR.


Risk is dynamic. You must manage risk by
identifying and reducing sources of risk.
With RiskWatch, you can:
identify the positions and risk factors in a
portfolio that have the most impact on risk;
identify trades to minimize the Value-at-Risk
of a portfolio;
restructure risk for a given level of
desired return.

Forward-looking Stress Testing. Conventional


stress testing assumes that the risk profile of
the portfolio is constant through time. However,
often the risk profile of a portfolio changes
because of many factors including hedge
expiry, dynamic hedging strategies, pull-to-par
effects for bonds, time decay of options
and re-investment of holdings into cash and
physical securities.

RiskWatch is unique in its ability to support


Hot Spots and Best Hedges analysis using
scenario based VaR methods in addition to the
more standard parametric methods.

Algo Market is unique in its ability to model risk


into the future. You can simulate forward in time
to assess the future hidden risks in a portfolio.
This allows you to detect risk in time for you to
take preventative action.

The additivity property of the MtF framework


also enables near-time marginal risk calculations
at the desk level. By pre-computing the MtF data
and releasing it across a global network,
consistent risk numbers are easily distributed
across the organization.

At a minimum, risk
managers need to stress test
the assumptions underlying
their models and set aside
somewhat higher contingency
resourcesreserves or
capitalto cover the losses
that will inevitably emerge
from time to time when
investors suffer a loss of
confidence.

Hot Spots Analysis. In large portfolios, risk


often shows up in unexpected places. To help
you identify the Hot Spots, those components
of a portfolio that generate risk, as well as those
that act as hedges, RiskWatch reports VaR
contributions at both the position and risk factor
levels. Although VaR is not additive, it can be
decomposed on a marginal basis. The resulting
decomposition can then be rescaled to obtain
the relative VaR contribution of each component.
Positive contributions identify sources of risk,
while negative contributions correspond to
components that reduce or hedge the portfolio
risk. VaR contributions help risk managers
attribute risk within a portfolio and highlight the
positions or risk factors that are most critical
from a risk management perspective.

Greenspan 1999 2

Conditional Stress Testing. In conventional


stress tests, the risk factors that are not explicitly
stressed are normally held at their nominal value.
An alternative and more realistic approach,
supported by Algo Market, automatically moves
the non-stressed risk factors so that they
maintain their historical correlation to the key
risk factors that are explicitly stressed.

Best Hedges Analysis. Once youve identified


the Hot Spots in the portfolio, it is then necessary
to determine how to alter those positions to
reduce risk. RiskWatch includes a technique that
displays how the VaR of the portfolio varies as
a function of the size of the positions within the
portfolio. The position size that minimizes VaR is
known as the Best Hedge. This analysis generates quick and cost-effective ways of reducing
positions when limits have been breached.

Greenspan, A., 1999, Measuring financial risk in the 21st century, before a conference sponsored by the Office of the Controller of the
Currency, Washington, D.C., October 14, 1999. Available at www.bog.frb.fed.us/boarddocs/speeches/1999/19991014.html
Hot SpotsTM is a trademark of Goldman Sachs

Providing Strategic Risk Information to the Front


Office. We recognize the importance of providing
these risk management tools to the front office.
Accordingly, we offer a CORBA API (an industrystandard distributed system architecture) to all
RiskWatch analytics, including Hot Spots, Best
Hedges and scenario-based optimization. This
allows traders to use these functions via their
trading system or custom-built applications
written in Excel, Visual Basic or Java.

Strategic Restructuring of Risk. Not only do you


need to understand the exposure of a portfolio
to multiple scenario shifts, but you also need
to be able to restructure risk in a stable fashion.
Dynamic hedging strategies based on BlackScholes replication are well suited to portfolio
replication in efficient, frictionless markets that
operate continuously. However, in practice, even
the most efficient liquid markets exhibit gaps that
render these strategies costly, if not impossible,
to execute. These gaps can include jumps
in volatility, price or correlation measures.
Also, unlike front-office hedging, strategic risk
management requires hedges that are valid
for longer periods of time.
Algo Market overcomes the limitations of these
traditional approaches through a patented
stochastic optimization methodology that
recommends appropriate hedging strategies.
The system attempts to rebalance a portfolio by
simultaneously considering weighted scenarios
over relevant risk factors to a designated roll-over
date. You can create various levels of protection,
trading-off the expected cost versus the quality
of the resulting hedge. This produces a more
flexible and effective hedge.

The Algorithmics solution enables


us to go beyond merely meeting
regulatory demands. We now
have the ability to look at future
risk and examine the ways in
which we can manage our capital
more effectively.
Renzo Avesani
Head of Risk Management and Research
Banca Intesa Group

10

Algo ALM
The Evolution of ALM

Asset Liability Management (ALM) is the process


of measuring and managing the market and
liquidity risk of the balance sheet. ALM is more
than simply risk protection. Its ultimate goal is
to enhance the net worth of the firm through
prudent leveraging of the balance sheet.

There are three primary factors that are driving


the development and implementation of the next
generation of ALM systems:
Market-Based Sources and Uses of Funds.
Increased weight of market-based sources
and uses of funds and the hedging possibilities
offered by derivatives have led to more
transparency in the market and an increased
capability to mark-to-market assets and liabilities.

ALM has undergone considerable change


in the last 20 years. This pace of change has
accelerated recently, resulting in a new
generation of systems and risk frameworks in
ALM. These changes borrow proven technical
methods from the capital markets and allow
you to manage ALM in an integrated enterprise
risk context.

Proliferation of New, Complex Instruments.


The past decade has brought a proliferation
of new complex instruments. Primarily
because of embedded optionality, these
new instruments cannot be evaluated using
the traditional approaches.

The drawbacks of the traditional approaches are


well known to market practitioners. Traditional
ALM methods overemphasize short-term net
interest income management at the expense of
the risk of changes to the market value of the
balance sheet. Because income simulation does
not capture risk beyond the horizon, the result is
a reward measurement without a corresponding
risk measurement.

Evolution of Accounting and Regulatory Rules


and Regimes. FASB and BIS are clearly steering
the banking community towards mark-to-market
accounting and the economic perspective of risk.

11

Gap

Duration

Parallel Shifts

Asset Liability
Management

Earnings
Simulation

Fair Value
Accounting

Parametric VaR

ScenarioBased VaR

Integrated Enterprise
Risk Management
Market Risk
Management

Stress Tests
Sensitivity
1970 - 1979

1980 - 1989

1990 - 1999

2000

CONVERGENCE OF ALM AND MARKET RISK MANAGEMENT

generation and valuation models. This allows


you to actively manage short-term profits and
economic value.

Banks should have interest


rate measurement systems
that assess the effects of rate
changes on both earnings
and economic value.

Optimizing the Trade-off Between Earnings


and Value. Hedging risk from an accounting
perspective does not guarantee that risk exposure
from an economic perspective will be reduced.
Conversely, hedging from an economic
perspective does not guarantee that risk from
an accounting perspective will be reduced.
Our patented scenario-based optimization allows
both perspectives to be analyzed simultaneously
to recommend hedging strategies that optimize
the trade-off between short-term profits and
long-term value of the firm.

BIS 19974

For example, BIS is now considering whether


to extend the use of internal models to interest
rate risk in the balance sheet (BIS 1999)5.
Currently, interest rate risk in the banking book
is not subject to any capital charge. These three
factors are causing a paradigm shift in the way
ALM is conducted (see table below).

Separation of Pricing and Scenarios for Risk


Management. Many ALM systems produce
earnings forecasts from statistical rate simulations and VaR distributions directly from the
term structure model thats used for pricing.
Since term structure models are based on a
small number of independent risk factors, the
system is limited in several ways. First, producing
earnings and VaR distributions using any method
other than Monte Carlo (e.g., historical simulation)
is problematic. Second, producing VaR measures
for portfolios of transactions that are valued off
of more than a single independent term structure
is difficult. And, third, accurate multi-currency
risk assessment is nearly impossible.

Benefits of Algo ALM


Consistent Measurement of Earnings and Value
at Risk. For ALM, its necessary to measure the
risk of the balance sheet from both an accounting
and an economic perspective. The accounting
perspective focuses on the change in earnings
resulting from a change in interest rates and
other risk factors. The economic perspective
focuses on the change in value resulting from
changes in interest rates and other risk factors.
Algo ALM supports both perspectives using
a single integrated analytical framework with
common scenarios, common growth and reinvestment assumptions and common cash flow

TRADITIONAL ALM

MODERN ALM

Focus on short-term earnings

Focus on earnings and value

Subjective scenarios based on parallel shifts

Objective scenarios based on statistical models


complemented by all types of subjective scenarios

Assumption of static portfolio (or at least dynamic


behaviour that is scenario independent)

Dynamic evolution of the balance sheet encompassing


growth forecasts, reinvestment, prepayment and
hedging strategies

12

Algo ALMs MtF architecture features a clear


separation of the scenarios used in risk
management and the pricing model. This allows
RiskWatch to produce realistic VaR/EaR
assessments across multi-currency balance
sheets from a single consistent scenario set
produced by HistoRisk.

Conditional Strategies

Dynamic Strategies

Pre-determined Strategies

Physical Settlement

Static

Dynamic Portfolio Evolution for Realistic Risk


Measurement. Interest rate risk in the balance
sheet is analyzed over a horizon of at least one
year, if not longer. Over such a long horizon,
assuming that the balance sheet is static severely
distorts risk measurement.

Cashflow Settlement

INCREASING SOPHISTICATION

INCREASING SOPHISTICATION
OF PORTFOLIO EVOLUTION

RiskWatch allows you to specify and evaluate


alternative, scenario-dependent growth forecasts and reinvestment strategies on the risk
profile of the balance sheet. The growth forecasts can include forecasts on spreads, volumes
and term preferences. These can be specified
directly by the user or can be driven from budget
or econometric models linked into RiskWatch.

Support for Traditional Analyses. RiskWatch


supports the traditional analyses used in ALM:
static and dynamic gap reports, duration and
convexity analysis and liquidity gap reports.
Comprehensive Instrument Coverage. RiskWatch
supports cash flow generation and valuation
models for a wide variety of balance and
off-balance sheet products. Valuation models
range from simple discounted cash flow and
closed form models to sophisticated lattice
and Monte Carlo-based models for complex
option-based products.

Stress Testing of Market Risk Factors. With


Algo ALM, you can assess the impact on both
earnings and value to changes in all risk factors
affecting the balance sheet including:
level and shape of the yield curve;
foreign exchange rates, volatilities;
spreads (e.g., changes in the spread between
LIBOR and the prime rate to measure basis risk).

You can extend the standard coverage with proprietary and third-party models. We also support
mortgage-backed and asset-backed securities
via a bridge to the Intex MBS database and the
Andrew Davidson prepayment models.

Stress Testing of Behavioural Assumptions.


Risk in the balance sheet can be highly sensitive
to behavioral assumptions such as withdrawal
and prepayment rates. In RiskWatch, these
assumptions can be model driven or stress
tested to evaluate the sensitivity of the outputs
to alternative assumptions.

Modeling of Non-Maturity Deposits. In the past,


ALM practitioners have treated non-maturity
deposits in widely different manners. The resulting implied interest rate risk varied from little
or none to very significant. More recently,
there has been a growing consensus that the
option-adjusted spread methodology should
be employed for valuation and hedging.

Analysis at Either the Transaction Level or at an


Aggregated Level. Traditional ALM systems
operate at either a transaction level or at an
aggregated level. RiskWatchs ability to consolidate transactions allows the balance sheet
manager to select the appropriate level of
aggregation for a particular analysis. A spectrum
of consolidation methods are supported,
including simple methods such as cash flow
aggregation and cash flow bucketing to
more advanced methods for products with
embedded options.

RiskWatch provides a flexible framework to


specify these models. The framework allows you
to easily specify custom models for deposit
volumes, account service costs and market rates.
The models that you can develop are quite
general: users can include a wide variety of
explanatory variables and incorporate such
features as partial adjustment, mean reversion,
asymmetric responses and path dependency.

4 Bank

for International Settlements, 1997, The evolution of bank supervision, Basle Committee on Banking Supervision, June 1997.
Available at www.bis.org/publ/.

5 Bank

for International Settlements, 1999, A New Capital Adequacy Framework, Basle Committee on Banking Supervision, June 1999.
Available at http://www.bis.org/publ/.

13

Flexible Funds Transfer Pricing. Funds transfer


pricing (FTP) is an analytical method to remove
the effects of interest rate risk from the performance of business units. RiskWatch supports
a variety of methods to calculate FTP including
strip funding approaches and option-adjusted
spread approaches that incorporate the cost
to hedge embedded options.
Complete Multi-Currency Support. RiskWatch is
a true multi-currency risk management system.
Each individual transaction in the system is
assigned a currency and the user can define
multiple discount curves for each currency.
Unlike other systems, RiskWatchs multi-currency
functionality extends to the realm of simulation
where future, scenario-dependent cash flows
are properly converted under the forward rate
in that scenario.

They do a better job than competitors


of balancing the financial engineering
and the software development and are
very receptive to client feedback.
Algorithmics culture is different: they tend
not to promise things that they cant deliver.
They understand the clients problems
and then use their creativity to custom
design a suitable approach.
Andr Horovitz
Executive Vice President and
Head of Group Risk Control
HypoVereinsbank

14

Algo Credit
Portfolio credit risk models are a powerful tool
to assess concentration risk. They can measure
potential changes in the risk of changes in
business lines, industry, credit quality, market
variables and the economic environment.
Accordingly, absolute and marginal risk
contributions can be estimated. Financial
institutions that use portfolio credit risk models
improve their overall ability to identify, measure
and manage credit risk by:
effective measurement of concentration risk;
effective and consistent measurement of
economic capital; and
active management and pricing of credit risk to
improve the portfolio's risk and return profiles.

Most major banking problems continue to


result from weaknesses in credit risk management practices. In the fall of 1998 when LTCM
almost collapsed, many market participants
were reminded of how inadequate their risk
assessment was: Standards for assessing
borrowers credit risks were often loose;
measures of credit exposures were ineffective;
the management of portfolio risk was poor and
the potential impact of changes in economic
conditions was not well understood.
Events such as the near-collapse of LTCM have
underscored the need to manage credit risk for
the entire portfolio, as well as the exposure to
individual deals or counterparties. Since concentrations are undeniably the single most important
cause of major credit problems, credit risk
management systems must be able to report the
composition of the credit portfolio for all trading
and banking book items and identify concentrations of risk. While the oversight of individual
deals or borrowers is important, your system
must also monitor the overall composition and
quality of various credit portfolios.

Algo Credit provides a consistent, integrated


framework for measuring and managing credit
risk across the enterprise including:
counterparty credit exposure;
commercial credit exposure;
issuer specific risk; and
portfolio credit risk.

15

METHOD

DESCRIPTION

ADVANTAGES

DISADVANTAGES

Factor approach

Exposure = max
(0,MTM) + notional factor

Easy to calculate

Doesnt incorporate
portfolio correlations

Consistent with 1988


Capital Accord

Ad-hoc rules required


for netting of add-ons

Theoretically consistent

Computationally intensive

Simulation approach

Value the aged portfolio over


its life across a distribution
of risk factor paths. Then
convert value to exposure by
applying applicable netting
and collateral rules.

Incorporates portfolio
correlations, netting, time
dimension
Works in multi-factor,
multi-currency situations
Can incorporate contingent
events such as mark-tomarket caps and posting
of collateral
Consistent with recently
proposed principles for the
measurement of credit risk

COMPARISON OF FACTOR AND SIMULATION-BASED APPROACHES

current exposurethe replacement cost

In addition to the many business benefits that


accrue from Algo Credit, you can also use
the software to reduce regulatory capital.
For example, in many jurisdictions, banks can
incorporate close-out netting when calculating
regulatory credit capital for off-balance sheet
positions. Close-out netting can dramatically
reduce capital requirements.

of the transaction today;


potential exposurethe replacement cost
of the transaction in the future (which depends
on future movements in market risk factors).
Both factor and simulation-based approaches
are supported in Algo Credit.

Counterparty Credit
Exposure Measurement

Factor-Based Measurement of Potential


Exposure. The factor approach calculates credit
exposure as the current market value of the
transaction plus an add-on thats designed to
incorporate the potential increase in value
resulting from future changes in market conditions. All add-on categories as defined by BIS
are supported. In addition, refined methodologies
that assign add-ons based on criteria such as
product type, risk factor, counterparty, remaining
time to maturity or moneyness are supported.
Under this approach, credit exposure can be
calculated on a fully netted or gross basis.
Alternatively, the net to gross ratio, as defined
by BIS, can also be used.

Since financial institutions establish credit limits


by counterparty, the volume of transactions that
can be executed with that counterparty are
directly related to the method selected to quantify
credit exposures. An overstatement of credit
exposure reduces profit opportunities, while
an understatement of credit exposure poses
unexpected risks. As such, its critical to measure
credit exposure accurately and consistently.

Simulation-Based Measurement of Potential


Exposure. The MtF framework, as implemented
in Algo Credit, overcomes the limitations of the
factor-based approach outlined in the table
above. Specifically, simulation-based approaches
consider portfolio correlations, aging effects
and credit mitigation, including netting. There
are three components in a solution to measure
potential exposure for derivative contracts.

The credit exposure to a counterparty is defined


as the value that would be lost if the counterparty
didnt fulfill its obligations. Exposure consists of
two components:

Scenario Generation. Generation of a distribution


of multi-step scenarios (paths) through time for all
market risk factors that impact portfolio value.

To use this type of calculation, banks must:


calculate netted exposures;
monitor limits in terms of netted exposures;
monitor potential roll-off exposures. (These
relate to sudden future increases in exposure
when short-dated obligations that are netted
against longer-dated claims mature.)
Algo Credit allows you to meet these regulatory
requirements and improve your return on
regulatory capital.

16

Mark-to-Future Simulation. Forward valuation


of the aged portfolio under all scenarios.

Net, Gross,
Collateralized Exposures

Credit Mitigation. Transformation of future


values to future exposures by incorporating the
impact of netting agreements, collateral and
termination clauses.

Credit Mitigation
(Algo RiskEngine)

In the Algo Credit solution, these components


are independent, providing the user with
considerable choice over scenario generation
and valuation methods.
Scenario Generation. Multi-factor, multi-currency
models that produce realistic scenarios of future
foreign exchange rates, equity prices and
interest rates are the foundation for the correct
estimation of exposures. HistoRisk provides
a comprehensive library of scenario-generation
methods for any number of asset classes
including interest rate, foreign exchange, equity
and commodity.

MtF Cube

Scenarios

Instruments

Time

MtF Simulation
(RiskWatch)

Scenario Generation
(HistoRisk)

A key challenge in all these models is calibrating


a number of parameters, including mean reversion coefficients, target levels, instantaneous
volatilities and correlations. In many implementations there is no mechanism to calibrate these
parameters, forcing the users to randomly
choose them. In HistoRisk, these parameters
can be automatically calibrated to historical data
or modified by the user to reflect his experience
and expectations.

MARK-TO-FUTURE: A framework for capturing


netting agreements and collateral effects
consistently over time and across scenarios.

testing of counterparty credit exposures to


capture peak exposures at all future time steps.
Its an approach that simulates counterparty
exposure in a scenario thats constructed as
the worst of the worst possible scenarios.
The worst possible values of each of the risk
factors affecting the counterparty portfolio are
identified at every time point of interest and
constitute the extreme stress test scenario.

Stress Testing Exposures. In practice, market


and credit events are often highly correlated.
For example, the credit rating of a Mexican
corporate bond may be highly correlated to
Mexican interest rates and the U.S. Dollar/Peso
exchange rate. RiskWatch allows the risk manager to assess the impact of joint events such
as this on the value and exposure of portfolios.

Mark-to-Future Simulation. Users can simulate


under non-equal time steps, a solution with two
benefits. First it allows for a more detailed analysis
over the near term. For example, the simulation
could proceed in daily steps for the first week,
weekly for the first month, monthly for the first
year and yearly thereafter. Second, it allows
potential roll-off exposures to be measured. These
relate to sudden increases in exposure, which
can happen when short-dated obligations mature.

Since most statistical credit risk models are


based on forecasts of term structures and other
relevant risk factors over multi-year horizons, it
makes sense to complement these models with
a comprehensive stress testing program.
Algo Credit supports single- and multi-step stress
testing. Single-step stress scenarios are commonly used to assess market risk. They shock
the risk factors once, at some specified point
in timeusually today. Over the longer time
horizons common in credit risk, multi-step stress
scenarios can be used to provide additional
insight into the sources of credit risk in a portfolio.

The RiskWatch MtF framework addresses


two key issues of simulating forward in
timeportfolio aging and path dependency.
Portfolio Aging. RiskWatch explicitly models
the settlement and re-investment of expiring
positions. Consider the case of a swaption that
is in-the-money at expiry and is settled into the
underlying swap. In RiskWatch, the credit

RiskWatch also supports scenario banding.


Scenario banding is a tool for systematic stress

17

exposure to the counterparty would include


the exposure generated by the swap, in addition
to the original swaption exposure.

tions where the products covered under the


collateral agreement are more limited than the
products covered under the netting agreement.

Path Dependency. The future value of many


financial instruments at a given time in the future
does not just depend on the state of the market
at t F, but also on the state of the market between
today and t F. In RiskWatch, path dependency
is modeled, resulting in forward valuations and
exposures that are accurate. For example, the
value of an interest rate swap at t F will depend
upon the prevailing level of interest rates at the
most recent reset date, which may have
occurred at any time between now and t F.

In all cases, the number and types of agreements,


as well as the jurisdictions, can be user-defined.

If banks agree to two-way


collateral provisions, they
should make sure that the
resulting additional credit
risk exposure is integrated in
the overall risk management
process, including measurement
of the potential future exposure.
Simply applying higher
confidence intervals or longer
time horizons is not appropriate
to capture the exposure
dynamics under turbulent
market conditions, especially
as they relate to the interaction
between market, credit,
and liquidity risk. 6

Credit Mitigation. RiskWatch offers a complete


solution to measure the risk-reducing impact of a
variety of credit mitigation mechanisms including
netting, collateralization and termination clauses.
Modeling Netting Agreements. Netting and collateral agreements can be complex. Algo Credit
provides a flexible solution to model the range
of netting agreements, including any associated
credit support documentation that governs the
collateral provisions of the agreements:
multiple netting agreements with a particular
counterparty that each cover a specific class
of products;
master/master agreements that sit above a
number of product and jurisdiction-specific
agreements;
ring-fencing where transactions only net within
certain jurisdictions;
cherry picking for situations where netting is not
allowed (in these situations, the counterparty
may elect to default on one obligation and
hold other obligations until maturity or payment
is required);
separate netting and collateral grids for situa-

Dynamic Collateral Management. RiskWatch


provides a comprehensive solution for assessing
the effects of collateral on current and future
potential exposures and liquidity. The MtF
framework within RiskWatch captures simulated
collateral events to accurately measure the
potential exposure reduction from collateral
and the associated market risk of the collateral
portfolio. That is, in each scenario and time step
in the simulation, collateral requirements are
checked vis--vis thresholds, initial margins
and the value of the portfolio and the collateral
in that scenario. Then, collateral is posted or
received as required. Correlations are fully
considered since the collateral and underlying
counterparty portfolio are simulated under the
same scenarios.

Counterparty ABC

ISDA

EC

Japan/ Brazil
USA

IFEMA

EC

NO_DOC

USA

In many instances netting agreements specify


thresholds that increase as credit quality deteriorates. Thus a downgrade can lead to large,
immediate cash outflows that can destabilize a
financial institution. In RiskWatch, credit support
annex variables such as thresholds and minimum
transfer amounts are updated dynamically in

NETTING HIERARCHY: In this example, either

an ISDA agreement, IFEMA agreement or no


agreement governs transactions with counterparty ABC. Transactions within the ISDA
agreement that are booked in EC jurisdictions
can be netted, transactions booked in Japan
and the United States can be netted, but the
transactions in the EC cant be netted with
those in the United States, Japan or Brazil.
6 Bank

for International Settlements, 1999, Sound Practices for Banks Interactions with Highly Leveraged Institutions,
Basle Committee on Banking Supervision, (E) January 1999. Available at www.bis.org/publ/.

18

make the validation of credit risk models difficult,


comprehensive stress testing programs must be
developed to ensure that sufficient credit capital
is held in reserve.

simulation in response to changes in your credit


state and the credit states of your counterparties.
This unique capability allows you to assess and
manage the impact of credit downgrades on
your liquidity requirements.

Generating Scenarios for Risk Factors. Most


portfolio credit risk models assume the independence of market and credit risk factors. Credit
spreads and recovery rates are often held
constant or simulated independently. Changes in
default and migration frequencies from one year
to the next are modeled only infrequently. There
is little analysis supporting these assumptions.

Restructuring Credit Risk with Credit Derivatives.


Credit derivatives are instruments that allow for
the transfer of credit risk between two counterparties. As such, they can be used to modify the
risk and return of a given portfolio. By allowing
users to develop joint scenarios on credit states,
credit spreads and other risk factors such as
volatility, RiskWatch is an ideal tool to assess
alternative credit derivative hedging strategies.
When credit derivatives are incorporated into
a portfolio, RiskWatch measures the credit
exposure to both the counterparty and the
reference asset.

Treatment of Market-Driven Exposures. For many


financial products, exposures are not known with
certainty but depend on future market or credit
events. The potential exposure to the counterparty of a swap, for instance, depends on the
future level of interest rates. The exposure under
a committed line of credit depends on the future
credit state of the borrower. Dynamic credit
mitigation techniques such as netting or
collateral arrangements may reduce exposure
profiles dramatically but they also introduce
additional risks.

Credit Exposure Measurement for Banking


Products. Algo Credit allows you to measure the
credit exposure of a wide variety of commercial
products including loans, syndicated loans, loan
commitments, letters of credit and mortgages.
Exposure can be measured with respect to the
borrower and to secondary parties such as
guarantors and insurers. The software also
considers the effects of collateral and
guarantor relationships. To improve performance
you can treat banking products as non-market
driven instruments where the exposure is not
dependent on the level of market risk factors.
Alternatively, banking products can be treated
as fully market-driven.

When market and credit events interact, changes


in market risk factors can increase both the
exposure to a counterparty and the counterpartys
default probability. Many portfolio credit risk
models, nonetheless, assume that changes in the
exposure to a given transaction or counterparty
are constant and independent of changes in all
other transactions or counterparties. It also
assumes that exposure changes are independent
of changes in the counterpartys credit rating.
The models only consider potential exposures
over an arbitrary horizon of one year. These
assumptions can significantly skew the output
of the credit risk model.

Specific Risk For Bond Portfolios. Spread risk


can be captured in Algo Credit by simulating a
given credit portfolio under scenarios on the
spreads that are applicable to each rating class.
The spread scenarios may be generated from
history directly or using Monte Carlo simulation.
Spot spread curves are entered to RiskWatch
in a manner entirely consistent with the other
market factors affecting the portfolio.

Algo Portfolio
Credit Risk Engine
The Algo Portfolio Credit Risk Engine is the first
commercially available solution for the integrated
measurement of market and credit risk. It
estimates portfolio credit risk at the enterprise
level, consistently aggregating deterministic
banking and stochastic trading book exposures.
It includes absolute and marginal credit risk
reports that identify credit risk concentrations
by business line, industry, country, credit
category or large counterparty.

Challenges of
Portfolio Credit Risk
There are critical issues that must be addressed
before portfolio credit risk models can be
implemented successfully.
Calibrating and Validating the Model. Portfolio
credit risk models require extensive historical
data pertaining to credit migrations, defaults,
economic variables and market conditions for
calibration. As data and methodology issues

Using the Portfolio Credit Risk Engine, you can


measure concentration risk and credit risk

19

The Algorithmics
Portfolio Credit
Risk Framework
in Five Steps

The framework integrates techniques for


portfolio credit risk and exposure simulation
and supports a default and credit migration
mode. It offers three key improvements over
current credit risk models. First, it defines
the joint evolution of market and credit risk
factors explicitly. Second, it models stochastic
exposures directly through simulation, as in
counterparty credit exposure models. Third, it
extends the Merton model of default to multiple
time steps. The framework is computationally
effective because it combines a MtF representation of counterparty exposures and a conditional
credit migration or default modeling environment.

1. Define joint scenarios for market risk factors


and credit risk drivers. Market risk factors
and systemic credit drivers are evolved over
the horizon. The credit drivers account for the
correlation of counterparty credit events.
2. Estimate counterparty exposures in each
scenario, accounting for (stochastic)

The Algorithmics Portfolio Credit Risk Framework


fully captures the interaction of market and credit
risk because credit migration and default in
the model are conditional on joint scenarios for
market and credit risk factors. Conditional
recovery rates and stochastic spreads can be
added naturally. Credit risk factors may include
economic indices to capture dependencies
between defaults or migrations and the
economic cycle. Both Monte Carlo methods and
comprehensive stress testing are supported.

recovery rates and credit mitigation.


3. Apply the default or migration model.
Estimate credit migration and default
frequencies conditional on each scenario.
4. Compute the conditional portfolio loss in
each scenario. Conditional on each scenario,
credit events are not correlated.
Only the credit risk that is specific to each
counterparty is modeled at this stage. Hence,
standard numerical or analytical methods

Building on the RiskWatch MtF architecture, the


Algorithmics Portfolio Credit Risk Framework
easily incorporates credit mitigation techniques
such as netting or collateral calls over time.
It correctly measures the credit risk of marketdriven stochastic exposures, wrong-way
exposures, credit derivatives and facilities
with credit optionality.

can be used to aggregate portfolio losses in


each scenario.
5. Aggregate across all scenarios.
Conditional loss distributions are averaged
over all scenarios.

The underlying RiskWatch MtF architecture also


ensures that counterparty exposures are aggregated consistently across scenarios, recognizing
that peak exposures for different counterparties
may occur under different scenarios and times.

capital effectively, allocate economic capital


consistently and actively manage portfolio credit
risk. In addition, you will be much closer to satisfying regulatory best-practice requirements for
risk measurement and management.

Credit defaults and migrations are calculated


over one or many time steps. Credit risk capital
can be estimated over a standard planning
horizon (e.g., one year) or over the entire life
of the portfolio. Using the notion of portfolio
regimes in the MtF architecture, the Portfolio
Credit Risk Engine estimates credit risk capital
for on-going business activities with reinvestments, as well as for static portfolios that
mature over time.

The Portfolio Credit Risk Engine offers an


integrated credit risk solution. Automated
interfaces to RICOS and RiskWatch provide
credit hierarchies and banking and trading
book exposures as inputs. Through an interface
to Standard and Poors CreditPro application,
you gain access to extensive historical data
on credit migration, defaults and recoveries.
Market and credit risks are modeled consistently
because the Portfolio Credit Risk Engine is tightly
integrated with HistoRisk.

The Portfolio Credit Risk Engine establishes a


benchmark for production quality best-practice
credit risk models. Credit risk estimates capture
the combined effect of market and credit risk and
build on sophisticated and accurate exposure
measures in the MtF framework.

The Portfolio Credit Risk Engine implements


the Algorithmics Portfolio Credit Risk Framework.

20

Algo Limits (RICOS)


Limits allow you to align firm-wide strategy
with the daily activities of risk takers. They also
provide a methodology for interaction between
the ERM department and the desk-level, risktaking activities of disparate business units.
And to the extent that you allocate capital along
the same lines as limits, dynamic limit allocation
allows dynamic capital allocation. This can
optimize your firms ability to take advantage
of business opportunities.

At the same time the new


[credit] methodologies often
remain confined to a narrow
group of credit rocket
scientists in the bank, while
the organization at large
continues to lend and control
risks the old way. Our analysis
attempts to determine the
extent to which the new
methodologies add real value
in a banks credit risk
management or remain only
an intellectual discourse.

Controlling Risk Using Advanced Credit


Measurement Techniques. Many banks are
moving towards more advanced methods of
measuring credit exposure and capital accounting for netting, collateral and portfolio effects.
Unfortunately, the operational control of credit
risk in many banks is still mired in the old-world
techniques of nominal risks and mark-to-market
plus add-on. One reason for this is cultural. The
second is technological. Many existing credit
limits systems are fundamentally designed to
support risk measures that are additive and are
incapable of making the leap into the new world
of advanced, non-additive risk measures.
7

Moodys 1999 7

Moodys Investors Service, An Analytical Framework for Banks in Developed Markets, Bank Credit Risk Rating Methodology, April 1999.

21

With limits set and exposures calculated, you


can reallocate credit lines, reduce exposure or
restructure credit risk using credit derivatives,
securitization or natural offsets.

Algo Limits controls credit risk using all measurement methods supported in Algo Credit:
notional credit exposure
mark-to-market plus add-on
simulation-based credit exposure
gross exposure
netted exposure
netted and collateralized exposure
expected and unexpected credit losses.

A Flexible Solution for Commercial Credit


Consolidation and Limits Management.
Algo Limits is a flexible solution for commercial
credit risk consolidation and limits management.
It allows you to distinguish between committed
and uncommitted lines of credit and underlying
drawdowns and exposures. This offers flexibility
in handling guarantees and collateral using
various rating rules and order structures to
determine how best to handle transfer risks.
The Algo Limits data model allows separate
definition of calculation rules by product type
allowing the definition of loans, overdrafts,
financial or trade finance products to a financial
institutions reporting rules. You can describe
the product catalogue to support operational
handling of product entry and risk management,
in terms of credit line and global risk limits.

In addition, Algo Limits has open interfaces to


control risks measured in third-party and proprietary risk engines.
Consolidating Risk Across All Markets. For
effective limit management, Algo Limits captures
and consolidates deal information from
commercial and capital market systems
across three dimensions:
own bank internal structure;
customer counterparty structure (includes
associated industry and country structures); and
product set structure (types and groupings of
deals/contracts).

Integrated with Enterprise Operations.


Algo Limits supports integrated credit risk management at all levels of your organization. Using
Algo Limits, a banks global credit risk control
group and regional credit managers in New York,
London and Tokyo can access the same database for product entry, online enquiry, limits
management and reporting.

When you consolidate this data into a central


repository, you can better understand the
total risks associated with industries and
countries. Your firm is better prepared to react
quickly to events such as the market meltdowns
in Southeast Asia, Korea and Russia.
Complete Risk Control for Informed
Decision Making. Limit setting allows you to
control risk by monitoring the use of available
credit capacity and excess. To effectively set
and manage limits, a repository of limit information must be accessed and consolidated with the
capital and risk demands of your organization.

Pre-Deal Analysis and Intraday Updates.


Algo Limits supports the direct input of new
commercial products and adjustments to
existing credit lines through online interfaces.
The software also supports pre-deal enquiry and
reservation control for commercial products.
Treasury pre-deal limit checks are supported
where the trade and its corresponding markto-market are fed from the front-office system.
Algo Limits calculates the add-on, incorporates
the effects of netting and checks limits across
all relative netting structures to provide details
of availability and excess.

Since Algo Limits stores internal and external


hierarchies and products, you can set limits for
each with a proper audit trail. This allows you
to respond quickly to warnings before problems
occur. Access to the data is secure because
Algo Limits segregates deal making and risk
management tasks.
Algo Limits optimizes the use of available trading
capacity and excess by allowing you to define
limits for all credit risk types and currency
combinations over defined time periods,
including tenor limits to monitor changes
in remaining maturity of portfolios.

22

Risk Architecture
for the Enterprise
want it accessible not only by the risk control
department, but also by the front office, senior
managers and, increasingly, their customers as
well. For an international organisation, this means
global distribution, more often than not using
the Internet.

As the practice of risk management evolves, so


must the technology that underpins it. Traditionally,
a small, centralized department was responsible
for monitoring risk and typically looked at daily
or less frequent figures. Furthermore, risk control
was fragmented across market, credit, liquidity
and operational areas. Nowadays, however, firms
want to continuously monitor exposures across
their entire organization, with an integrated view
of all risks. To achieve this requires not only a
rethinking of risk management practice, but also
a radical review of the technology architecture
that supports it.
The first generation of risk management systems
comprised a data warehouse for collating and
centralising data and a risk engine designed for
batch-run analysis. This suited the early risk
control departments and the formulaic methodologies they used. These systems typically
produced end-of-day reports of market Valueat-Risk, separate credit measures and so on.

Algorithmics, which pioneered enterprise risk


management technology, already has a risk
engine and data management software that are
proven in the traditional approach in over 100
installations. Now we have created an innovative
architecture that builds on these core components, leveraging them in a MtF framework. The
key elements of this new architecture are the
comprehensive use of industry standards at all
levels, a component-based distributed computing
environment that can meet the highest performance demands and an infrastructure that makes
system functions available everywhere over the
network. The Algo Suite risk architecture satisfies
four broad objectives, as follows.

But this is no longer good enough. Today, firms


want their risk analysis more immediately available as near to real time as possible. And they

The first objective is to collect and manage the


data required for risk management. This includes
the financial instrument terms and conditions,

23

The final objective is to provide flexibility and


extensibility to the system so it can adapt to
rapidly changing business requirements.

positions in portfolios, credit data such as master


agreements and transition matrices, pricing
models and market data. This data can reside
on a single or distributed data environment to
support global operations.

Reducing
Implementation Costs
with RiskMapper

The second objective is to enable fast batch


and intra-day computation by distributing the
results of the analytical processing across
multiple processors, multiple servers and
potentially across multiple geographical regions.

RiskMapper is a robust, proven mapping tool


that provides fast, maintainable and configurable
mapping from source systems into Algorithmics
analytic and data management products.
RiskMapper reads raw data exported from
source systems, performs the required mapping
and transformations and creates output files
for direct loading into RiskWatch, RICOS or in
any custom format.

The third objective is to provide a flexible mechanism to distribute the results of the analytical
processing. This can occur through standard
reports, through ad-hoc requests for risk
information over the corporate intranet and
even through requests from your clients over
the Internet.

Algo Cube
Explorer

HTML
Reports

Historical
Reports

Enterprise Java Beans


Application Server
POST-CUBE

Algo
Portfolio Credit
Risk Engine

Algo
RiskEngine

RICOS
MtF Cube

RiskWatch

HistoRisk

INPUT

PRE-CUBE

Market Data
Transactions
Portfolios
Curves

DATABASE
Credit Data
Counterparty
Hierarchies
Netting Agreements
Limits

RiskMapper

Loans

Bonds

Equities

Interest Rate
Derivatives

THE ALGO SUITE SOLUTION: An n-tier, component-based risk architecture enables scalable processing
and thin-client applications that are isolated from server-side changes.

24

data; attributes; and scripts used. This provides


a useful reference tool for audit purposes.

The issues raised by non-standard data exports


from source systems, new or evolving products
stored on those systems, and the need to
document data mapping for future administrators
comprise a significant portion of the resources
and time required to implement an enterprisewide risk management system. RiskMapper
can significantly reduce the expense and time
required to interface source data.

Scalable Processing
using Algo Suite
A most critical aspect of providing a global enterprise risk solution is to ensure it can scale as the
number of users grows and as the number and
complexity of positions grows.

Mapping Flexibility. RiskMappers easy-to-use


GUI allows you to define the required data
mappings and transformations to convert data
into a common Algo Suite format. The input into
RiskMapper is usually a csv (comma separated
values) file thats been generated from a source
system. The source file is defined to RiskMapper
and the applicable source system fields are
mapped to the appropriate RiskWatch or
RICOS objects.

Scalable User Access. Algo Suite employs an


n-tier architecture that allows partitioning of
machine activity based on application logic,
server-load and numerous other parameters.
Adding additional distributed servers to meet
your growing business requirements is a simple
configuration exercise. The application server,
which manages the logic and workflow between
the server components, shields the thin-client
applications from any server-side changes.

Mapping Transparency. Without RiskMapper,


the data transformation process is likely to be
embedded in code or scriptsinvisible to all,
except the programmer who maintains it. A key
feature of RiskMapper is the ability to define and
store these RiskWatch attributes through the
GUI interface instead of a script or csv file.
With the GUI interface, the user doesnt have
to type attribute values. Theyre already built
into a drop-down table, ensuring consistency
across instruments.

Scalable Risk Analysis. Calculating portfolio risk


measures is computationally intensive, especially
in credit risk and balance sheet simulations. The
Algo Suite MtF architecture explicitly addresses
this computationally intensive problem by splitting
the processing into pre-cube and post-cube phases. Each phase is based on an n-tier architecture
enabling distributed and scalable processing.
The pre-cube phase simulates all instrument
holdings over the user-selected scenarios and
time points. This phase, which is the most
computationally intensive, creates the MtF cube.
The persisted MtF cube can support an unlimited
number of portfolio combinations supporting
hundreds, even thousands of risk analysis
requests. In conventional architectures, all the
portfolio information must be available before the
risk analysis can start. In a global operation this
means that much of your batch window is taken
up just waiting for the portfolio details to arrive.
In Algo Suite, we have addressed this problem
with a technique called pipelining. Simply put,
Algo Suite can start creating the MtF cubes on
the instruments before the final position details
are available. In addition, multiple RiskWatch
servers can create the MtF cube. Within each
RiskWatch session, the computation is further
distributed over multiple processors using
symmetric multiprocessing. This is achieved by
distributing the scenarios among the available
processors, resulting in linear scalability.

All mapping rules, including transformations,


are defined and stored in the Windows NT
environment in a MS-SQL server database and
referenced when the data is loaded. ODBC
connects RiskMapper to the MS-SQL database.
When mappings are complete, NT launches the
production process or the mapping rules are
uploaded into a proprietary database for Unix
production overnight batch runs.
The benefits of this approach are numerous.
For one, the mappings are stored in a central
location rather than spread out over numerous
scripts. Maintenance is simple, fast and is
performed in a controlled environment. Second,
the mapping rules can be defined and maintained
by a business analystprogramming skills
are not required. Finally, the mappings are
transparent so the user can easily verify them
to confirm their definition. RiskMapper generates
its own documentation in HTML format.
The documentation describes everything thats
been defined for a particular instrument: source

25

The MtF architecture uniquely enables the


Application Service Provider (ASP) model by
leveraging your in-house investments in market
data, qualified staff, and software technology.
This results in fast time-to-market, lowest project
cost and best service. Specifically:
The MtF framework provides for efficient
processing. The scenario analysis to create
the MtF cube is performed only once even
if hundreds or thousands of clients hold
the security.
The analytic complexity of generating the MtF
Cube is isolated from the application. This, along
with the MtF frameworks application server,
enables rapid development of thin-client
applications targeted to the needs of the
ASP customer.
The use of Enterprise Java Beans seamlessly
enables Internet-based applications such as
web pages, applets and applications, and
provides security, scalability and reliability.

The post-cube phase is responsible for satisfying a request for risk analysis on a portfolio.
Leveraging the results saved in the MtF cube,
the Algo RiskEngine phase assembles and scales
the MtF tables that correspond to an arbitrary
portfolio hierarchy and calculates the appropriate
risk/reward statistics.

Delivery of Risk
Information
over the Web
As risk management information expands from
the middle office to the trading community, and
even to customers of the financial institution,
the idea of one user interface satisfying everyones needs is unrealistic. The objective then
becomes how to easily create graphical
applications that suit the specific needs of
the particular user. Shielding the applications
from the various server components and the
workflow between these server components
becomes a critical feature to enable fast
development of thin-client applications.
Through the use of an application server and
Enterprise Java Beans technology, Algo Suite
provides a well documented and robust
application interface to the component servers.
This speeds up the development of the client
applications by enabling the application
developers to focus on business logic instead
of the complex logic and interaction of the
server components.

Enterprise Risk
Reporting with the Algo
Reporting Database
The Algo Reporting Database provides a common solution for credit and market enterprise risk
reporting across the trading and banking books.
It stores enterprise risk measures and allows
you to manage and access that information.
The measures are stored in a flexible relational
database day-by-day through time. Because
the database is fully integrated, there is no
need to construct data transfer routines into a
reporting system.

The thin-client risk applications are made possible by the MtF architecture which effectively
turns risk analytics into risk data, delivering fast
response times regardless of the complexity of
the risk analysis requested by the application.
Risk data is created by pre-computing and
persisting the MtF cube. Since this risk data
persists in the MtF cube, the applications can
retrieve the data without having to perform the
computationally-intensive scenario analysis.
The result is the MtF cube becomes the standard
data foundation for the user-specific applications.

An imbedded Online Analytic Processing (OLAP)


server enables quick ad-hoc access to the
data for analysis. The OLAP front-end or other
commercial third-party report writers allow you
to generate reports from the historical database.
Flexible Results Storage. The Algo Reporting
Database stores results based on your reporting
needs. You can:
store results at the portfolio level only or also
at the position level;
define dimension hierarchies to aggregate
results (includes standard aggregations based
on portfolio, risk factor, product and currency);
store selected input position and instrument
attributes (e.g., notional amount, counterparty);
store data through time to allow time-based
comparative reporting;
store the results of several different simulations.

Providing Risk Information to your Customers.


Many large banks are already providing online,
web-based financial services to other financial
institutions such as custody services, clearing,
online trading and prime brokerage services.
Adding risk information is a natural value-added
extension of these services.

26

A standard report model describes the


underlying RiskWarehouse tables and data
relationships to the server and contains a series
of report templates that you can use to customize
your own reports or generate a set of reports.
The report model is the best starting point to
easily customize and incorporate additional risk
measures and dimensions that are unique to
your business.

Easy Data Retrieval and Reporting. An OLAP


server, embedded within RiskWarehouse,
facilitates data retrieval and reporting and allows
you to make intelligent business decisions. The
OLAP server, provided by WhiteLight Systems,
Inc., provides the reporting cornerstone for a full
decision support system to manage market and
credit risk.
All groups across the enterprise can access
and share the information within the database,
regardless of geographic location. Whether via
a hardcopy report, GUI interface for ad-hoc
queries, corporate intranet or third-party report
writer, the information is ready for use across
the organization.

Full Security. The Algo Reporting Database


provides comprehensive security features to
allow managers to set data authorizations from
a user level to a discrete data element level.
Open Architecture. The data stored in the
Algo Reporting Database can also be easily
linked together via the OLAP server with data
from other systems such as back-office P&L,
trading systems or limit systems. Data can be
retrieved from any other ODBC-compliant
relational database, or any source that is
supported via a CORBA layer.

Risk information is available along standard


dimensions such as time, portfolio node, product,
scenario set, currency or user-defined dimension.
You can sort by dimension or pivot between
dimensions with a click of the mouse. For example, pivoting and sorting between dimensions
allows you to view results by categories such as
date, portfolio and currency without complex
programming. You can quickly isolate risk factors
and drill-down to the most detailed levels of data.

With RiskWatch, we have the ability


to not only fast-prototype new financial
products, but also to fully integrate
these into our risk management system.
This allows us to fully develop a new
financial product, including hedging
and control systems, in a matter of
days instead of weeks or months.
Michael Durland
Managing Director & Deputy Head
Capitals Market Group
Scotia Capital Markets

27

System Extensibility

RiskWatch-Engine Edition. RiskWatch is also


available as a Risk++ server that communicates
over an intranet/Internet network using CORBA.
Java, C++, Visual Basic or Excel programs can
communicate with the engine edition of
RiskWatch. You can tailor these programs for
specific tasks such as trading, ALM or securitiesrelated auditing applications.

An enterprise risk management solution is a


strategic, long-term investment that must not
only adapt to the rapid pace of new financial
product innovation, but also to the rapid changes
in technology. Algo Suite is designed to satisfy
your immediate needs as well as your future
requirements, by being extendible by both you
and Algorithmics.

Risk in Revolutions
Per Minute

Our clients cite flexibility and extensibility of


the Algo Suite solution as major factors in their
decision to choose Algorithmics products.

If a life could be measured in RPMs, Algorithmics


has invested the talent and ingenuity of its
industry leading research team at track speed
for over a decade. Hundreds of man-years only
describes the intensity of our pursuit in the area
of enterprise-wide risk. The real achievement
has been identifying the questions that must be
answeredand providing those answers.

RiskWatch Extensibility. RiskWatch is built


around a flexible object-oriented framework
that allows you to customize and extend its
functionality. This is critical for a risk
management environment in which methods
for hedging and valuing financial products
and methods evolve on a daily basis. Simple
extensions can be incorporated using the
Visual Basic compliant RiskScript product.
More complex extensions can be prototyped
in RiskScript and re-coded using Risk++
for performance.

Today, Algorithmics serves over 100 clients with


more than 140 installations in 18 countries around
the world. Fifty percent of the top 16 banks
use our software. Our client portfolio diversity
however, also provides Algo with a broad
perspective and tremendous base of experience.
Major banks, investment houses, insurance
companies, regulatory bodies, corporations and
asset managers around the globe benefit from
our risk management solutions.

Risk++. This set of core C++ libraries is a key


component of Algo Suite. Developers can create
proprietary valuation models for use within
RiskWatch and add:
third-party valuation models and libraries to
allow custom calculations and methodologies
to be used as though they are programmed
directly into RiskWatch or RiskEngine;
specialized scenario generation procedures;
sophisticated curve bootstrapping procedures;
new instruments and risk factors.

Algorithmics perceives the world of risk in


RPMs, revolutions per minute, and our
promise remains to be first with the solutions.
Algorithmics. Know Your Risk.

These extensions become part of RiskWatch


and are as fast and robust as other internal
parts of the product. RiskWatch loads and
automatically integrates the extensions,
so extra coding isnt required.
RiskScript. This Visual Basic language has been
embedded in RiskWatch and HistoRisk. It allows
you to easily extend RiskWatch by creating valuation models, aggregation functions, simulation
expressions and curve functions. If performance
is an issue, valuation models are typically
prototyped and tested in RiskScript. The logic is
then reprogrammed in Risk++ for performance.
You can also use RiskScript to automate a
variety of tasks including reporting and repetitive,
interactive tasks.

28

For over a decade Algorithmics


has conceived and executed the
worlds finest Risk Management
software. Today, Algo Suite
provides our clients around the
world with an integrated solution
for the entire enterprise.
With Mark-to-Future,
Algo is projecting risk well
into the third millennium.

Know
YourRisk
Worldwide

Toronto
185 Spadina Avenue
Toronto Ontario
M5T 2C6 Canada
phone 416-217-1500
fax 416-971-6100
e-mail marketing@algorithmics.com
web site www.algorithmics.com

London
phone 207-481-3434
fax 207-481-3130

Madrid
phone 91-418-69-09
fax 91-418-69-99

New York
phone 212-625-5260
fax 212-625-9604

Johannesburg
phone 11-327-7272
fax 11-327-7279

Tokyo
phone 3-3221-9722
fax 3-3221-9765

Singapore
phone 65-536-7737
fax 65-536-7730

Paris
phone 1-53-53-67-70
fax 1-53-53-67-00

Sydney
phone 61-2-9293-2928
fax 61-2-9293-2828

Frankfurt
phone 69-97503-220
fax 69-97503-325

Mexico City
phone 525-520-4293
fax 525-520-4292

Vienna
phone 1-31686-0
fax 1-31686-2000

Rio de Janeiro
phone 21-533-3021
fax 21-533-3021
Bridgetown
phone 246-436-3421
fax 246-228-3847

Algorithmics
www.mark-to-future.com

Incorporated

June 2000 Algorithmics Incorporated. All brand names and product names are trademarks or registered trademarks of their respective holders. All rights reserved.
Creative: Scott Thornley + Company Inc.

You might also like