ALM WhitePaper

Document Sample
ALM WhitePaper Powered By Docstoc
					ALM Implementation in Banks – an approach paper




Surya Software Systems Pvt. Ltd.,
Bangalore
FUNDAMENTALS OF ALM...............................................................................................3
   OVERVIEW .............................................................................................................................3
   STRUCTURAL LIQUIDITY R ISK ..............................................................................................3
   INTEREST RATE R ISK ............................................................................................................3
   INADEQUACY OF BALANCE SHEET ........................................................................................4
   NON-TERM PRODUCTS ...........................................................................................................4
   PROBABILISTIC CASH FLOW PRODUCTS ................................................................................4
   OPTIONS, FUTURES AND DERIVATIVES .................................................................................4
   PROFITABILITY ......................................................................................................................5
BANK ALM POLICY ...........................................................................................................5
   PRODUCT ...............................................................................................................................5
   STRUCTURAL LIQUIDITY .......................................................................................................6
     Gap Measurement ............................................................................................................6
     Cost to close gap ..............................................................................................................6
     Maximum Cumulative Outflow........................................................................................6
     Scenario Analysis .............................................................................................................7
   INTEREST RATE RISK ..............................................................................................................7
     Interest rate Gap ..............................................................................................................7
     Duration Gap ...................................................................................................................7
A MODEL FOR IMPLEMENTATION .............................................................................9
ALM IMPLEMENTATION – PROBLEMS IN BANKS ..............................................11
CASE STUDY OF ALM IMPLEMENTATION IN A LARGE BANK IN
WESTERN AFRICA ...........................................................................................................13
Fundamentals of ALM


Overview

ALM techniques are used to manage assets and liabilities by timeframe.

Objectives of ALM are:

      Maximize profitability
      Minimize use of capital
      Ensure structural liquidity
      Ensure robustness in market risk management
      ‘Just in time’ money

Risks addressed by ALM are described below.


Structural Liquidity Risk

Liquidity has often been defined as ‘Ability to raise money that is not required’ in a
humorous vien, the argument being that the day you need this money, no one will be
willing to lend to you. Liquidity risk is simple mathematics but complex finance.

Structural Liquidity risk is measured in multiple ways. Gap between inflows and
outflows by timeframe is a measure. Cost to close gap may be used as a measure of
liquidity risk. 7-day Maximum cumulative outflow is another method. One may use all of
the above.

Interestingly, many trading strategies involving complex models make an assumption of
infinite liquidity. Reality is liquidity is finite and this can only be ignored at one’s own
peril.

Interest Rate Risk

Gap between interest rate sensitive assets and liabilities, spread over time is a
measurement of market risk. This is a simplistic measure, a minimum that is required.

Sensitivity of Net Interest Income (NII) to interest rate change is another view. Basel
recommends sensitivity to a 200 basis points parallel shift in yield curve as a standard
measure. 200 basis points is a Basel guideline. Central banks of individual countries may
impose different measures. This measure, when used over multiple scenarios can give a
reasonable picture of sensitivity of interest rate income to movements in interest rates.
Traditionally, banks have a trading book and a banking book. For purposes of
determining interest rate sensitivity, both books are mapped to zero coupon bonds,
preserving market risk. Combined book represents bank’s interest rate risk profile.
Traditional methods like modified or dollar duration gap and convexity gap analysis
enable risk measurement of interest rate sensitivity.


Inadequacy of balance sheet

ALM techniques are not meant to replace balance sheet techniques in any shape or form.
It is a completely different technique. First of all, ALM addresses time bucket concept,
whereas balance sheet ignores time element completely. Secondly, ALM demands
inclusion of off-balance sheet items that have potential impact on ALM sheet – for
example unutilized portion of cash credit i.e. options given to customers.


Non-term products

This is the complex finance part. Certain products, example savings bank have no
contracted terms. Thus, they present conceptual difficulty in being mapped to zero
coupon bonds as it is not possible to determine date when cash flows occur. Thus, such
products are generally split into two or more Zero coupon bonds, maturing on different
dates. These parts are termed volatile and core. Core is expected to mature in later time
buckets. Volatile portion is expected to be in first bucket. This may change by nature of
account and other dimensions. This is a study in itself.

Probabilistic cash flow products

Savings bank and current account are examples of banking book transactions of
probabilistic cash flow behaviour. Banks are net sellers of options, both explicit and
embedded. On trading book, probabilistic cash flows define instruments. Thus, complex
models based on sophistication are required to map derivative type of instruments into
the cash flow model.


Options, Futures and derivatives

Bank uses these instruments to hedge positions. To offer a customer a long position in
USD at a certain rate, bank has to hedge by taking a corresponding short position. Thus,
regardless of USD rate, bank is fully protected, offering customer protection as well.
Thus, options, futures and derivatives may be used to take positions, apart from hedging.
Difference needs to be identified as Basel II norms specify different treatments for the
same.
Profitability

Profitability by business units, as given out by simple balance sheet is distorted. A
business unit or a branch located in a residential area is by definition a deposit-taking
branch. Thus, its profitability should be measured by efficiency of deposit collection i.e.
weighted average interest rates of deposit collection by time buckets compared to a
standard yield curve. Thus, concept of funds transfer pricing has emerged strongly in the
past few years and is in itself a separate subject.


Bank ALM policy

ALM policy is drafted and updated by bank’s ALCO. ALM policy demands that board of
directors, Asset Liability committee follow a formal procedure. ALM Policy covers
bank’s position on all risks – credit risk, market risk, liquidity risk etc.

Banking keeps emerging as a practice and in times will change even further. Thus,
policies must be reviewed every now and then. This ensures that practices are current,
though business itself does not change. In India, for example, for a large number of years,
it was liability creation that was the prime driver. Once bank gathered enough funds,
there were multiple asset creation avenues. However, of late, it is asset creation that
comes first, followed by liability creation.


Product Section


Both assets and liabilities are considered products and parameters defined for both. For
example, deposits may have various characteristics and structures for interest rates. Even
plain vanilla deposits need to be priced and priced by timeframe. Competition may
introduce new products based on ALM positions. Banks may have to price based on their
ALM positions. Thus, this section of ALM policy defines products that bank may deal in
– both assets and liabilities.

Complexities are introduced by options – both explicit and embedded. Savings bank and
cash credit is a classic case of embedded options. Thus, ALCO needs to understand
impact of probabilistic cash flows before approving such products. Before being offered,
product creation needs to go through a proper introduction and approval mechanisms
through ALCO. Thus, policy should address parameters that should never be crossed.
Structural Liquidity

Structural liquidity policies must be defined for measurement and implementation of
liquidity controls in any financial institution. Individuals practice structural liquidity
measurement and control for personal portfolios. Thus, it stands to reason that these are
even more essential for banks.


Gap Measurement

Time buckets are defined as bands. 1-14 days, 15-days to 1 month, 1 month to 2 months
etc. is an example. This organisation is termed a maturity bucket scheme. All cash flows
are mapped to corresponding buckets. Thus, entire portfolio of cash flows is now reduced
to a bucket representation, thus making it easier to analyse.

Since all products are mapped, assets represent all inflows and liabilities represent all
outflows. Thus gaps in each time bucket is analysed. Regulators specify use of
percentage of tolerance for gaps. Practical bankers use an absolute amount.

Thus, as long as gap remains within tolerance, then it is deemed zero. Thus, the statement
in the beginning that zero gap is impractical and not desired either. Bank’s funding or
lending gaps may be very deliberate.


Cost to close gap

This is another measurement for structural liquidity. The last bucket is closed first using
market interest rate for that bucket. Some implementations divide all buckets to months
internally and calculate cost to close at month level. Cost to close of the last bucket is
them taken as an outflow in the previous bucket and that closed and so on all the way till
the first bucket is closed. That gives the total cost to close gap.

The other way is to simply calculate cost to close gap for each bucket based on interest
rate and assuming that all cash flows occur at the gap median.

Tolerance to limits of cost to close is defined as a measure of structural liquidity risk and
this is used for control.

Maximum Cumulative Outflow

Maximum cumulative outflow analysis is measured in number of days. 7 day or 15 day
MCO is used as a practical measure. This simply adds up all cumulative outflows. No
inflows are considered. Thus, this measure indicates a sum of all possible cash outflows
in seven day period.

Tolerance to MCO may be another measure to improve structural liquidity control.
Scenario Analysis

Liquidity analysis scenarios are generated. A typical measure would involve worst case
(MCO analysis), best case and likely. (This may be used to commit money in money
markets et. Al.). These scenarios are scrutinized and their impact approved by ALCO as a
matter of routine. All analysis referred to above provide measures enabled by these
scenarios.

Many banks, as a matter of routine, create scenarios on top of native cash flows. They
alter nature of native cash flows based on their prior knowledge. Derived cash flows are
indeed scenarios that have been pre-defined.


Interest rate risk

Interest rate risk is measured using traditional techniques for measurement of market risk.
Market risk exists due to volatility of interest rates. Financial institutions make money as
they take market risk. For example, if a bank provides a 10 year housing loan, then it is
has to locate 10 year assets to minimize market risk for that loan.

Both traded and non-traded assets and liabilities carry market risk and any technique
should address this. All products carry market risk and this needs to be addressed as well.

It must be understood that all statistical techniques are forecasting algorithms based on
history in some form or the other. There is no guarantee that history will repeat itself and
new paths and patterns may emerge. Statistical techniques help make judgments. They
are not replacements for flesh and blood managers.


Interest rate Gap

Interest rate gap of a bucket is calculated in a manner similar to liquidity gap. Tolerance
of gap in terms of percentage, absolute values is a risk control measure. Tolerance
provides control point as well. Buckets may be determined carefully. Making buckets too
small leaves a large number of buckets. Making them too large may hide some sensitivity
issues. Especially in volatile buckets (1-90 days), definitions must be small and in larger
buckets (3 years to 30 years), bucket size may be larger. As with liquidity, an absolute
value of gap may be used to measure this risk.


Duration Gap

Duration gap measures sensitivity of portfolio to interest rate changes. Dollar duration is
a measurement of drop in market value of (bank as) portfolio for a weighted average
change of interest of 1%. Modified duration is a measure perfected by bond traders is a
measure of sensitivity of market value of (bank as) portfolio to interest rate changes.
Thus, in this analysis, bank’s transactions are mapped as zero coupon bonds and their
market value determined using yield curves for discounting. Then duration is calculated
for the same. However, a point to note is that usage of a yield curve may not be correct
for some products. A more realistic discounting measure would be the product’s own
interest rate.



NII Sensitivity analysis

Sensitivity of Net Interest Income to movement in interest rates may be determined by
changing interest receivable and payable. It is assumed that 100% of assets and liabilities
will get re-priced. This may not be realistic and re-pricing % is a parameter that must be
determined by bank’s behaviour. Thus, sensitivity of NII to interest rate movement and
interest rate shocks are a interest rate risk measure that may be used. Unlike duration, this
is more simplistic and will not carry the concept of ‘time value of money’.



Scenario Analysis

Interest rate sensitivity analysis scenarios are generated. A typical measure would involve
worst case drop in NII - rate shock of x% on cost and y% on yield, best case and likely.
These scenarios are scrutinized and their impact approved by ALCO as a matter of
routine. All analysis referred to above may be measured for above scenarios.

Many banks, as a matter of routine, create scenarios on top of native cash flows. They
alter nature of native cash flows based on their prior knowledge. Derived cash flows are
indeed scenarios that have been pre-defined.
A model for implementation

For any risk measurement in finance, all complex instruments are reduced to simple
instrument in an artificial manner, preserving market value and market risk. Then, simple
instruments are analysed for risks.

The simplest financial instrument is a zero coupon bond.

BALM implements all above functions by reducing banking transactions to a simple zero
coupon bond. Mapping of term products to zero coupon bonds is very straightforward.
Maturity date is known and principal is expected back on the cash flow date. Interest
however, is put into another item called interest receivable as another zero coupon bond
as this is required for re-pricing effect on NII and interest bears zero interest rate.

However, mapping of non-term products is slightly more complex. They have to be
synthetically split into multiple zero coupon bonds. Splitting rules vary for each maturity
bucket. BALM supports multiple concurrent bucketing schemes. Thus, for example, in
scheme 1, savings bank may be split into 5% on 1-14 day bucket, 10% on 15-1 month,
and 85% in over 30 years. In scheme 2, however, split may be 10% in 1day – 1 month
bucket and 90% in over 30 years.

Derivation of these percentages to split by is outside the system. Models for determining
this may be quite complex as these contain options and cash flows of these options have
different valuation algorithms. These must be available to have evaluated options in the
first place. Thus, problem reduces to determing cash flows emanating from embedded
options, where no deliberate evaluation has been made, exemplified by savings bank.
Here, historical behaviour of rate of change is assumed to be in a log-normal distribution
and hence volatile portion determined. Non-volatile portion is determined by subtracting
volatile portion from the total.
BALM Architecture




     Transaction
       Interface       BALM                     BALM
        Objects        Model                 Applications


     Savings                            Liquidity      Interest
      Bank                                Gap            Rate
                                                         Gap
        Current
        Account
                                        Duration       Cost to
                                          Gap          Close
      Letters of
                          Cash
       Credit                              NII         Scenario
                    Flows in terms of
                        Amount,         sensitivity    Analysis
      Term          Cash flow Date,
      Loans            Currency,
                      Interest Rate
                                          VaR            FTP
         Term
        Deposits
                                         Ratio          Reg.
      ….                                Analysis       Reports
     ……..

                                        Trends         ALCO
                                                       Support
      SLR


      Trading
                                          …..
                                          …..
ALM implementation – problems in banks


Policy

Lack of a coherent, documented and practical policy is a big hindrance to ALM
implementation. Most often, ALCO membership itself may not be aware of implications
of risks being measured and impact. Policies should address all issues concerning the
bank, all policies should be clearly explained to all members of board, apart from ALCO
and these must be documented. Proper revisions to this document, a quarterly review
needs to be organized as well as parameters may be changing due to change in situations.


Understanding of complexities

Many people in a bank need to understand risk measurements and risk mitigation
procedures. Measurement of risk is a fairly simple phenomenon and does go on
regardless. Formalisation of understanding, especially at a top level is very essential.
Failures inevitably occur due to lack of understanding, coupled with a feeling that top
management knows all that there is in banking.



Organisation and culture

Risk organization in banks generally land up reporting to treasury, as they are people who
come closest to understanding complex financial instruments. The fact that they are a
business unit, in charge of ‘risk taking’ is overlooked. ‘Risk Taking’ and ‘Risk
management’ are generally two distinct parts of any organization and both must report to
a board completely independently.

Openness and transparency are essential to a proper risk organization. Most organizations
react badly to positions going wrong by taking more risks and enter a vicious cycle of
risks. Thus, it is required that banks follow policy in both letter and spirit. This policy
was derived when situation was not volatile and hence must have merit.

Most dramatic failures in the last decade have not been because of market risk or credit
risk but bad risk management organizations. This must be a big pointer to boards and
ALCOs on avoidance of such issues.
Data and models

Data may not be available at all times in requisite format. It must be remembered that
many data items are assumptions and gaps must be measured in perspective. There was a
case of a manual branch of a bank that was closed for 6 months in a year due to inclement
weather and was largely inaccessible. As data may not be obtained from this branch for 6
months, appropriate assumptions have to be made in any event. The argument is that for
all other purposes, assumptions are being made.

Sensible options need to be chosen and manual branch without computer was an
example. However, in modern banking, it is mapping of models to zero coupon bonds
that are an issue. Once again, arguments are that this should exist within the bank. Based
on sophistication required, multiple models may be used to validate this conversion. This
is strictly outside ALM framework but integrates into ALM framework.


Unrealistic goals

An ALCO secretary was seen desperately trying to tweak with parameters to ‘show’ less
gaps in liquidity reports. A zero gap is not practical. Returns are expected for taking risks.
Banks assume market and credit risk and hence they make returns.

ALCO’s job is to correctly determine positions and put in place appropriate remedial
measures using appropriate risks. It is not to show things as good when they are not. In
any event, market risks and credit risks are not the only causes for failure, as evidenced
by failures in the last decade.
Case study of ALM implementation in a large bank in
Western Africa

BALM implementation in the largest bank in Western Africa is being discussed.
Following diagram depicts deployment scenario of core banking system.

A DIAGRAM OF CORE BANKING



Specifics

165 branches were planned to be under core banking, 200+ were under BankMaster –
distributed TBA (Total Branch Automation system). Bank had decided to leave some
branches in manual mode and was not willing to computerize them as they were very
small and returns on investment for computerization was not justifiable.

Being the largest bank in the territory, its actions could move the market. Thus, any
measures on liquidity control etc.. had to be carefully exercised. Any panic in this bank
would plunge that country into chaos.

Bank had desired to do install ALM systems, without regulatory pressure from central
bank. Thus, top management were very keen. There was no problem in risk organization.
However, people were used to everyday transaction banking and needed time to get used
to centralized systems.

ALCO policy statement was in place and needed a review. Top management did not
relate to the policy, as it was an external document, created by a consultant. ALCO
meetings were not specific technical meetings. They were like EXCO meetings. All
issues except ALM were discussed. ALM statistics were produced in a very elementary
manner and ALCO did not raise many expected issues.

Treasury were not clear on the concept of ‘Held for trading’ and ‘Hold to Maturity’ type
of marking on transactions. Hedges were not clearly marked and it was not possible to
recognize hedges from positions. There were no logical explanations for some positions
or hedges.


Data organisation

Core banking was being rolled out during ALM implementation. Thus, data from
branches could be coming from TBA one day and be replaced by core banking the next
day. Data from manual branches came in only once a month rather than everyday. Thus,
on each data load, an option had to be provided to duplicate last known data.

The following diagram depicts data organization.

INTERFACE CONTROL



Hardware and system software

Representatives who sold ALM software were also representatives for hardware and
system software. Thus, timings were a non-issue. Procurement processes in many banks
may be long and after that vendors have to deliver requisite pieces. Thus, planning for
delivery of hardware and system software is a very important step and needs to be
planned up-front.

Implementation steps


Establishing requirements

ALM policy states company’s requirements of system implicitly. Interviews with senior
management to understand bank’s direction are used in addition to policy document
statements. In this particular instance, implementation manager was invited to an ALCO
meeting and understood existing ALCO proceedings.


System review and parameterisation

System is demonstrated to key people including ALCO secretariat. A second
demonstration the day after first is necessary to establish parameters for the system for
existing functions. Many banks for example want to create a set of derived cash flows as
a default. All such items are discussed here.


System study Report and Gap analysis

At the end of study, system study conclusions are reported. In addition, gaps recorded
and how to address these gaps is recorded as well. In this particular case, all policy
requirements were clearly addressed and there were no gaps. If there are gaps, then
implementation plan must address them through customization or change in expectations
from implementation.
Create ALM Sheet items

ALM sheet looks suspiciously like a balance sheet. However, it is not. This sheet is also
‘As-on’. There similarity ends. There is no need for multiple balance sheets at all. Time
value of money concept handled by ALM is not handled in balance sheet techniques.
ALM sheet is strictly cash flows as given to system.

A big difference is treatment of off-balance sheet items. As we are considering liquidity,
market and credit risks, all contingent items that do not find a place in balance sheet need
to be considered as well. This is a very important aspect to be considered. Interviews with
all aspects of banks dealing in cash flows of any sort are necessary.

If a subsidiary is given an option of taking funds from the bank, those need to be
considered as well. Employee stock and other options need to be taken into account.


Data sources

For each ALM sheet item, data source is identified. Data may be sourced from
transactions or warehouses where items refer to term products. Synthetic cash flow rules
are created where products are non-term products like savings bank. Any analysis to
determine volatility is in-built into this system at this point.


Data flows and verification

Data flow and integrity is established in this step. Data from all sources are gathered and
loaded into system. Data is verified by whatever means are available to establish
integrity, for example summing up with transaction systems or warehouses or G/L
systems at total level. Once verified, all systems are go!


Go Live

This signals sign-off on data and methods by ALCO and start of use of system. From this
point onwards, support and maintenance of roadmap are issues to consider.
Do’s and Don’t;s

     Do involve all ALCO members in decisions. Some functional heads may not be
      interested. It is best to have someone as a salesman for ALCO to sell ideas, how
      important these ideas are to implement central systems for better benefits for
      bank.
     Have a younger person, enthusiastic in nature as ALCO secretary. This person is
      responsible for all pre-ALCO analysis and distribution of ALM reports to relevant
      people.
     Do not deliberate a lot over non-term product distribution. It is anyway a
      probabilistic cash flow. Worry more about systems in place to constantly review
      this.
     ALM sheet item granularity depends on distribution for non-term products. For
      example, ‘savings bank’ may be one heading or ‘savings bank – salaries’ could be
      the level at which distribution of volatility differs. Thus, discuss these items
      beforehand.
     Do not be afraid to get it wrong. It may go wrong the first time. Thing to do is not
      panic, but start all over again quickly and not make same mistakes again
     Do not be tempted to use ALM as a balance sheet manager. For those who have
      not had central systems before, this may be the first time centralized data is
      visible. Do not use it as a central repository.
     Define functional objectives completely before starting this project. Do not keep
      tampering with it.
     Senior management may refer to well known books on this subject to get a quick
      revision.
     Do not over-engineer your ALM sheet. Let it evolve.
     Results of ALM are visible over a couple of years. Keep measuring what is
      required.

				
DOCUMENT INFO