19-10-2016, 02:14 PM
1459858324-banking.pdf (Size: 682.51 KB / Downloads: 8)
Abstract: The banking sector is a competitive environment, where business process re-engineering is constantly
needed. Business process modelling and automation are effective tools towards this direction, improving the
performance of business activities and enabling enterprise-wide monitoring and coordination. In this paper, we
present a case study of modelling and automating business processes in the Loan Monitoring Department of a
medium-sized Bank. Loan monitoring is a typical banking activity, which includes business processes
concerning loan approval, collection of delinquent loans and initiation of appropriate legal claims. These
processes are often performed in cooperation with external business partners, such as legal firms and brokers,
have collaborative properties and are considered to be of dynamic nature. Their efficiency strongly depends on
human operator experience and subjective criteria. The loan monitoring policy employed is a significant factor
for determining profits. Thus, relevant business processes should always be monitored, evaluated and refined.
Business process modelling was conducted using the Modified Petri-Net (MPN) model, which allows the
description of ad-hoc and collaborative business activities. Business process automation was performed using
Lotus Domino/Notes groupware platform, since widely known workflow management systems do not provide
the means for the description of such activities. The direct mapping and support of MPN main entities within
Notes environment ensured the accurate and complete implementation of all business processes and reduced
significantly programming cost. Loan Management System is the integrated environment build to support loan
monitoring activities. Our experience and the potential of the business process modelling and automation
approach are also presented.
INTRODUCTION
Integrating and managing complex organisations and
their information systems requires understanding,
partitioning and simplification of their complexity.
Business modelling supports these requirements by
providing means for describing process-oriented
systems and decomposing them into manageable
parts. Business processes (BPs) are collections of
activities with a common objective, such as fulfilling
a business contract or satisfying a customer need.
Business process definition (i.e. a description of a
BP at a high conceptual level necessary for process
understanding, evaluation and redesign) requires a
well-defined model, that provides a set of concepts
appropriate to describe BPs [Zelm et al, 1995]. The
model should be rich enough and enable process
validation (e.g. by simulation or static analysis) to
decide whether the process definition accurately
represents the system under study. BP modelling is a
significant tool for re-engineering organisational procedures and it is usually followed by BP
automation, aiming at improving business process
performance and enabling organisation-wide
monitoring and coordination. An automated BP is
referred to as a workflow, while a Workflow
Management System (WMS) is software used for its
coordination and control [Mohan et al, 2000].
WMSs also provide a set of interfaces to users and
applications involved in the workflow progress. For
efficient workflow development, one should start
with defining and understanding business processes
(BP modelling), before specifying and implementing
the corresponding workflow applications (BP
automation). Provision of generic and flexible
modelling methods is thus required both at BP
modelling and BP automation levels. Several
methods have been suggested for BP modelling,
most of which are based on textual programmable
languages or graphical notations, such as dataflow
diagrams, state transition diagrams, Petri-Nets and
related notations. Combination of different BP modelling methods has also been examined to give
new, enhanced approaches [Abeysinghe and Phalp,
1996].
Four types of business processes are usually studied
in the literature [Alonso and Mohan, 1997]. These
are: production, administrative, ad-hoc and
collaborative. Administrative and production BPs
refer to bureaucratic procedures that include welldefined
steps and are controlled by a set of wellknown
rules. Such processes can be easily described
by conventional modelling tools and are usually
automated using a WMS [Hollingsworth, 1995]. Adhoc
processes are similar to administrative
processes, except for the fact that they deal with
unique or loosely defined conditions, which are not
easily modelled, and can not be efficiently supported
by current WMSs. Collaborative processes are
characterised by the number of participants involved
and the synchronisation needed, and are handled
more effectively using groupware technology.
In this paper, we present our experience on business
process modelling and automation of the Loan
Monitoring Department (LMD) of a medium sized,
private Bank. Loan Monitoring includes business
processes concerning loan approval, loan collection
and loan case assignment, which are considered as
collaborative and dynamic. Business process
modelling was conducted using the Modified PetriNet
(MPN) model [Tsalgatidou et al, 1996], which is
an extension of Coloured Petri-Nets and allows the
description of dynamic activities. To automate
business processes modelled with MPN, we decided
to extend the capabilities of Lotus Domino/Notes
groupware product [Lotus Co, 2000]. The MPN
modelling environment and the WFS developed for
LMD, named Loan Management System (LMS), use
the same repositories maintained within Lotus
Domino platform, enabling the direct mapping of all
entities defined during BP modelling within the
WFS.
The rest of the paper is organised as follows: Section
2 provides an overview of loan monitoring processes
and their specific characteristics. Section 3 describes
the MPN modelling approach and its integration
with the workflow support environment. Section 4
provides an analytical description of the MPN model
for the delinquent loan collection process. In section
5, we discuss process automation using Lotus
Domino platform. Our experiences concerning the
system deployment over the last two years and its
impact on LMD operation are presented in Section
6. Conclusions reside in Section 7.
2. BUSINESS PROCESS DESCRIPTION
Bank of Athens is a medium sized, private bank. The
assignment of corporate loans to healthy, small companies, with relatively loose terms was one of its
most profiTable activities. The bank decided to offer
other loan types as well and, after three years, it
became clear that, as business grew, profits deviated
significantly from the expected ones.
LMD is responsible for the approval and monitoring
of loans. The main objectives of the department
include approving loans, collecting delinquent loan
instalments and initiating legal claims against noncredible
customers. LMD is in constant
collaboration with the Credit Office, all local
branches of the Bank as well as external business
partners (e.g. legal firms). Department employees
handle three main loan types: corporate, housing and
consuming. Monitoring and approval of each loan
type is performed according to a different policy and
data maintained for each category differ.
LMD is divided into two sections: The first for
handling consuming and housing loans and the
second for handling corporate loans. Each section
has its own Manager and consists of groups, formed
by no more than ten employees including Group
Leaders. All employees, including Managers, handle
loan cases, while under specific circumstances a
consuming or housing loan case may be also
assigned to an employee working in the corporate
loan section.
Although the account management system produced
detailed daily reports for loan monitoring, there was
no information system support for LMD. Due to the
enormous amount of data, classification and
processing was far from efficient, resulting in poor
coordination and low productivity. Furthermore,
data of dynamic nature concerning the history of a
delinquent loan case, e.g. delayed payments and
payment enforcement actions, was not maintained.
Both the re-engineering and WMS support of the
LMD were thus approved.
The first step towards automating LMD operation
was the identification and complete specification of
the business processes supported. After carefully
reviewing the organisational scheme and
interviewing employees, we concluded that three
main business processes are supported:
Loan Approval
Delinquent Loan Collection
Loan Case Assignment (for both approval and
collection)
Loan approval process is initiated by the Credit
Office at local branches, which serve as the entry
point for any loan approval request. The approval or
rejection of the request must be announced within a
period of 24 hours for consuming loans and 3 to 7
days for housing and corporate loans. Without information system support, the approval of
consuming loans was performed by the LMD
employee at the specific branch, based on locally
available data. The requests for housing and
corporate loans and the relevant certificates were
photocopied and forwarded to LMD. While LMD is
responsible for loan approval, contract establishment
and customer account updates are performed by the
Credit Office.
Delinquent loan collection is the most important
activity of LMD, initiated whenever the payment of
a loan instalment is delayed. In practice, this was not
feasible due to the large number of delinquent loans,
especially consuming ones. LMD had no mechanism
to detect delinquent loans and assign them to a
collector in a daily basis. In addition, loan
assignment could not be performed dynamically
according to evolving criteria, concerning either loan
groups or a specific loan case. Moreover, section
and group managers were not in position to modify
the criteria for delinquent loan case assignment to
the collectors supervised by them.
The requirement to assign the loan to a collector that
had already handled it in the past was also not
satisfied, as delinquent loan case history was only
kept in paper files. If the collector was an external
partner, all files had to be copied and sent via mail.
This was one of the reasons external partners
handled only a small portion of permanently
delinquent loans.
In spite of the three loan categories, we concluded
that loan handling was performed according to a set
of common rules, which are then customised on the
basis of predefined parameters, such as loan
category, amount owed, etc. However, the employee
approving or monitoring the delinquent loan case
makes decisions according to subjective criteria.
Lack of knowledge of the loan case history and
customer information affected the collector’s
judgement.
Loan monitoring activities described above can be
characterised as collaborative processes, since many
participants are involved in the completion of a
single step. The delinquent loan collection procedure
is an ad-hoc BP, since the steps to be accomplished
are not well established and exceptions may occur.
Furthermore, each loan case may be reassigned to
another collector at any time and collection policies
may be changed while handling a case. Thus,
collection activities evolve in a dynamic way [Casati
et al, 1998].
3. BUSINESS PROCESS MODELLING
Numerous modelling methodologies, such as IDEF0
[Marca and McGowan, 1993] and RADs [Ould, 1992], provide the means to understand the
behaviour of static systems [Starke, 1994].
Production and administrative activities usually fall
in this category. BP modelling is also used to tackle
the problem of changing or evolving systems [Phalp,
1998]. Ad-hoc processes can be viewed as an
evolving system. The business model used to depict
such systems should be flexible enough to facilitate
the accurate description of business activities. BP
modelling approaches based on extending Petri-Net
functionality [Murata, 1989] can provide better
solutions for this kind of problem [Oberweis, 1996],
[Tsalgatidou et al, 1996], since they focus on
depicting the relationship between activities and
resources rather than the relationship between
activities. BP models based on Petri-Nets can be
simulated using discrete event simulation [Rajala
and Savolainen, 1996].
Advantages of simulation-based validation of
business processes have been well recognised in
[Nidumolu et al, 1998] and [Hlupic and Robinson,
1998]. Some of them are: modelling of cyclic and
stochastic behaviour and complex rule-based
interactions between activities and resources,
visualisation of process steps, development of
accurate models through comparing them with
mathematical data and enhanced confidence in
predictions regarding performance impacts of
changes to the process [Gladwin, 1994]. Although
the benefits of simulation are not yet widely
exploited by the business engineering community,
there is a great potential offered, as indicated in
[Hlupic and Robinson, 1998], especially when
dealing with dynamic BPs.
As described in the previous section, loan
monitoring processes are both ad-hoc and
collaborative, thus of dynamic nature. Figure 1
depicts the steps accomplished for implementing a
workflow system to support LMD operation. After
gathering and evaluating information concerning the
department operation and policies, a BP model was
constructed, evaluated and then mapped into the
workflow environment.
The BP model must fulfil the following
requirements:
Enable the accurate description of ad-hoc
processes
Facilitate the evaluation of BPs through
simulation
Support the direct mapping of entities into the
workflow environment to minimise
implementation cost.
The modelling formalism adopted is the Multi-level
Modified Petri-Net (MPN) [Tsalgatidou et al, 1996],
which is an extension of Coloured Petri-Nets
I