The use of the Internet as a teaching tool has been
documented in a number of places in the last twelve months. The
collected opinion is overwhelmingly in support of its value as
a teaching tool if used in a suitable environment and for suitable
purposes. This paper adds to this discussion by describing the
operation of a postgraduate level course in Banking Risk Management
using the Internet as a data provider for 'Value at risk' (VAR)
models. The course used the J. P. Morgan RiskMetrics package which
is freely accessible over the Internet to supply consistent correlation
and volatility measures across a range of markets and financial
instruments. The paper describes how the course was organised,
what problems were faced in providing Internet access for a taught
Masters course and how the use of the Internet was integrated
with conventional teaching approaches.
The fact that the Internet has begun to be used to support the education process has been documented and described in a number of places. There has been a particular increase in the interest in the network's potential in this domain over the last twelve months marked by a significant increase in the experiments reported upon in the literature. Three papers are dedicated to this subject at this conference alone. This paper describes the attempt to use the Internet as the medium through which data was obtained to complete a course in Banking Risk Management. This course was offered for the first time to MBA students at the University of Birmingham, U.K . in the second semester of the 1995/6 academic year. The paper begins by describing a number of other experiments that have been undertaken of a similar nature to the one reported here to set this paper in its literature context. It goes on to describe the course that was given and details of the RiskMetricsTM package used to supply the datasets for the practical elements of the course. It concludes with some discussion of the problems that were faced in running this course and describes how they were overcome. At the point of writing this paper the course has yet to complete its first full running and therefore limited student feedback is available at present.
A number of authors have written on the subject of the use of the Internet in education. Some of this material has been published in the conventional way, some is available on-line.
Lymer (1996a) lists a range of examples where the Internet has been used in various ways in Higher Education. He describes five areas in which the Internet has been used to deliver material of educational value :
The July 1995 issue of Active Learning (Number 2) was also dedicated to this subject. In particular it contains papers by Sangster (1995b), who describes the potential and pitfalls of using the World Wide Web (WWW) in education, by Pickering (1995), who argues that the Internet may offer the educational utopia as described by Illich in 'Deschooling Society' (1970), by Yeomans (1995) who examines the role of the Information superhighway (effectively analogous to the 'Internet' as referred to in this paper) in promoting wider access to education and by Whalley (1995) who compares the use of the Internet as a delivery tool to the CD-ROM.
Further details of the use of the Internet in an educational environment can be found in the special issue of Computers and Education (1995). This volume contains a wide range of papers that describe such subjects as using the Internet to support scholarly discussion groups (Berge and Collins 1995), collaborative distance learning (Watabe, Hamalainen and Whinston 1995) and the promotion of literacy using the network (Kuntz 1995).
Sangster and Mulligan (1996) and Ibbotson and Grant (1996) add to this discussion in their respective papers on the integration of the WWW into an Accounting Systems course, and the proposed collaborative development of WWW based Accounting Tuition software. The plenary session of the CTI-AFM 6th Annual conference was also given over to this subject in 1995. The two papers presented there detailed the growth in the technology over the previous twelve months as related to its potential use in an educational environment (Lymer 1995) and the challenges that that growth offered to education (Sangster 1995a).
Online guides are also a good source of material in this area. Two sites of particular value are the ISWorld and W3Lessonware project sites. ISWorld (http://www.isworld.org) is a collection of resources put together by volunteers members to provide a central resource of information systems information and pointers for the Internet. One element of the range of information they cover is the Teaching and Learning domain. This site provides pointers to Information Systems (IS) course materials online, IS teaching methods, tools and technology, technology supported learning (electronic classrooms, tele-learning, collaborative learning etc.) and pedagogical issues (measures of teaching effectiveness, research on pedagogical issues etc.) (http://www.cba.bgsu.edu/amis/smagal/teaching/index.html).
The W3Lessonware project (http://www.comp.it.brighton.ac.uk/w3lessonware/)
is constructing a tool kit which offers a graphical interface
and a set of tools dedicated to creating and maintaining computer
based learning material on a Web server (more details of this
project are given in Lymer 1996b).
Sangster (1995b) suggested that the impact of the web in education will be significant. He suggested that educators will have to take seriously the impact it will have on their courses in the very near future. He details two routes from which he suggests educators will have to choose in deciding the role the web will take in their courses :
Sangster suggests that, despite approach 1 being the most common response to the web's influence on teaching currently, the only viable option in the future will be the active approach as the resources available on the web will need to be harnessed alongside the more conventional support tools if we are to maintain quality in our teaching.
Deciding on the active route then offers to the educator other decisions as to the scope and approach used in incorporating web resources into their courses. Mothobi & Strom (1995) quote Ibrahim (1995) who suggested that the pedagogical use of the web will develop along two major axes. He states that, although browsing the web for information can be a learning experiment in its own right, there are two structured ways in which the web can be used to support education :
(words added as per Mothobi & Strom 1995)
These axes are not mutually exclusive and can be applied in a complementary way if managed correctly. Mothobi & Strom (1995) describe the possibility of a third axis within Ibrahim's structure to suit their particular use of the Web where the increase in the users' knowledge can be facilitated and measured.
The course described in this paper takes an active approach by explicitly incorporating relevant web-based resources to re-enforce the teaching of the subject matter. On Ibrahim's axes the course would be placed closer to axis 2 than axis 1 as it is making use of resources on the web that are not primarily created for the particular course being taught.
The University of Birmingham U.K. runs a number of MBA programmes, one of these programmes specialises in courses relating to Banking and Finance from an International perspective. Students accepted for this programme are required to have a relevant first degree (accounting, finance, banking, business studies or economics) and a minimum of two years work experience in the finance or banking industries. The programme offers a range of optional courses one of which is a course in Banking Risk Management. As a pre-requisite, this course requires the participants to have studied portfolio theory in detail and to come from a strong quantitative background (ie knowledge of calculus, statistical techniques and Monte Carlo Simulation). In 1995/6, the first year this option was offered, seventeen students opted to take this course.
The primary course objective is to expose the students to a number of modern Financial Instruments (such as derivatives, swaps, structured notes, collateralised mortgage obligations etc.) that could be used to create a Bank's portfolio of holdings and to give them the tools to assess how the value of this portfolio would change over time with alterations in the positions adopted by the Bank and movements in the underlying factors. This objective is broken into a number of elements as detailed to the students on signing up for the course. Successful completion should enable students to :
- discuss the types of risks faced by financial institutions
- discuss Value-at-Risk (VAR) models, stress testing and other methods of risk measurement
- discuss financial instruments, their pricing and risks
- understand J. P. Morgan's RiskmetricsTM and be able to use the datasets they provide
The course is delivered by 2 one hour lectures per week for the duration of a semester (12 weeks). This is supplemented by a number of training practicals (in advanced use of the Excel spreadsheet, in the use of the Internet and a number in the development of VAR models). Assessment is based on a three hour written examination at the end of the course, and on a single piece of course work due at the end of the course. The assessment is weighted 75:25 examination to course work. This balance is likely to change in the favour of the course work element after the first year. A copy of the programme is given in appendix one.
One of the central problems faced by a modern bank in handling increasingly complex financial instruments is the management of exposure risk. In the past banks measured the risks in the individual parts of their trading books separately, but now they are moving towards a 'whole trading book' approach. This provides the bank with the ability to assess the risk exposure of the organisation as a whole. One of the most popular range of methods of aggregating the variety of positions held by a bank are the so called Value-at-Risk (VAR) models.
'The aim of a VAR model is to calculate, on a consistent basis, the likely loss that a bank might experience on its whole trading book, allowing for the hedges that exist between - as well as within - different markets. VAR models assess likely price changes of instruments within individual markets and at the extent to which prices in one market vary with those in others.'
( Bank of England Quarterly Bulletin, May 1995)
The VAR model is particularly concerned with assessing 'market risk'; the maximum an organisation can expect to lose on a given position during a given period with a pre-defined probability. To calculate this potential loss the organisation needs to estimate the potential changes in the rates and prices affecting the positions they have adopted. The potential loss can then be calculated by either assessing the sensitivity of the position to rate and price changes (the delta valuation method) or by calculating the actual loss that would occur at various changed values (the full valuation method). Using Delta valuation for risk estimation is a parametric approach (the potential rate and price changes are described by the parameters of volatility and correlation) whereas the full valuation method is a non-parametric approach (actual possible scenarios are used to value the potential loss).
These two methods form the basis of the two main
VAR approaches: variance/covariance and simulation. The variance/covariance
approach uses summary statistics describing past price movements
and correlations between price movements to estimate likely potential
losses in the Bank's portfolio of positions. The simulation approach
involves using data on past price movements to estimate what losses
would have been incurred on the positions held at present had
they been held in the past. Whichever approach is used in practice,
the primary source of information that makes the model of any
use is historical price information from financial markets. The
consistent provision of the range of data that is required to
allow for the use of VAR models in practice requires a significant
effort. It is in fulfilling this need that information providers
have found a niche market. However, since October 1994 people
with access to the Internet have been able to use a J. P. Morgan
produced solution to this problem and are able to by-pass the
cost of the value-added information providers if they are prepared
to accept the 'J. P. Morgan' approach solving to this problem.
This solution is packaged as the RiskMetrics datasets and methodology.
RiskMetricsTM is :
Figure 1: Daily Volatility and Correlation data
breakdown as provided by RiskMetrics package.
By combining the methodology with the changing datasets to create a VAR model, RiskMetrics can be used to measure and monitor the changing risk exposure of a bank, or indeed any organisation, over a period of time.
The RiskMetricsTM methodology and datasets are made available, free of charge, over the Internet (from the World Wide Web or via anonymous ftp) on a daily basis (http://www.jpmorgan.com/ or ftp.jpmorgan.com).
J. P. Morgan suggest they offer free access to this package for three reasons (Longerstaey 95) :
In order to fulfill the course objective of giving the students the experience of measuring risk using VAR models the Internet was to be used to download the RiskMetrics datasets to make them available for manipulation. J. P. Morgan supply their estimates of volatilities and correlations on a daily basis. The initial intention was therefore to allow students to select and download their own data set at a time convenient to themselves at the appropriate point in their project work. The implications of this was that there would be 17 copies of similar datasets stored on the university computer system, some possibly being identical if two students downloaded the same day's datasets. However, there was a possible capacity issue that needed to be addressed if this was to be viable. The largest daily file is only 970KB in size to download, but it expands to two files of 20KB and 3MB of volatility and correlation data respectively. The total of approximately 55MB of information would therefore be created in this process (17 x 3.2MB). This was not considered to cause a real capacity problem given that this space would only be required for a few weeks before it could be released back to general usage. It was considered that the extra flexibility and experience having control over their own data would offer to the students was an important part of the educational value of the course. Around a 130MB of storage space was allocated temporarily to this course to allow the students to download two sets of data each on which to operate. In practice this meant giving each student access to file storage space on the School network of the order of 8MB in which to carry out their project.
With these decisions having been made at the course planning stage a lab session was organised for the students in which they could be introduced to the Internet (for the benefit of those who had not used it before) and could be shown how to download and use the RiskMetrics datasets. This lab session was planned to last an afternoon in length. Of the 17 members of the group only 5 had not used the Internet before. This small number is not representative of the general level of usage amongst the student population at The University of Birmingham. The average number of experienced users of the Internet amongst the wider student community is likely to be much lower. This MBA class consisted almost exclusively of students from overseas who described their primary reasons for learning about the use of the Internet was to read newspapers from home that were accessible via the web and to communicate with fellow students and family members via email.
The lab session was also to detail some of the history of the network to help the students understand why it is organised and operated in the way it is at present and how these structures are constrained.
At the time of writing the course described in this paper has yet to be completed by the first group of students however, there were three problems discovered in practice when planning and establishing this course that are worthy of note in this paper. First, the download time from J. P. Morgan's site to the University of Birmingham meant that it was impractical to allow the students to download their own datasets as had been planned and some support had to be offered. It was considered that the educational and efficiency benefits of allowing the students to continue to manage as much of their own data as possible was important and should be maintained if at all possible. Second, there was limited access for taught masters students to machines in labs that were connected to the Internet directly on campus. Third, the level of experience needed to handle the data manipulation between the datasets as downloaded and the spreadsheet environment in which the students wanted to make use of the data was not insignificant and required more specialist training to be organised than had been planned for. These three problems are briefly expanded upon below and details of how they were overcome are given.
In planning the lab sessions and attempting to organise the integration of the Internet with the Excel spreadsheet (the tool the students were to use to create their VAR models within) it was discovered that downloading the datasets directly from J. P. Morgan's own site could take a significant period of time. Over the period of a week daily access was attempted to the RiskMetrics site at various times of the day to assess what point was likely to offer the best response time that could be recommended to students. It was discovered that the early morning (UK time) was by far the best time however, the files could still take more than an hour to download. It was therefore decided that the best option would be to download the files centrally and make them available over the local network to the students who could select which one of the series available they would like to use for their project. This option has proved to be an acceptable way around the slow download times that were experienced but has increased the support that was necessary to enable the course to operate.
No detailed attempt was made to discover where the bottle-neck was forming in the download process that caused such lengthy delays for files that are not that large relative to the volumes the network regularly moves around. It may be this problem was short-term in nature and will have been resolved by the time the course is offered again next year.
The students were able to gain access to the datasets that were placed on the School network in order to gain access to the volatility measures and correlation details that they required for their particular VAR model at the point they wanted them when developing their models in the Excel spreadsheets. In this way the Internet itself was not used directly by the students to gain access to the data they needed. However, the principles of using remotely generated and stored datasets was maintained. It is hoped in the future that more direct use of the Internet will be possible. It was however, possible to reduce the storage capacity needed for the project that had been an initial concern as the data sets could be placed in a directory that could be shared by the entire group rather than each student having their own version. In this way more data could be provided for the students within similar capacity constraints.
The second problem faced in organising this course
relates to access to networked machines for taught Masters students.
At the University of Birmingham there are a large number of public
access labs of computers that are regularly updated with new technology.
However, the growth of the Internet and the increase in the power
of the machines that are required to make effective use of it
have meant that only the newest labs provide a particularly efficient
environment in which to operate in the ways required for this
course with the volumes of data that is involved. Because these
labs are the newest they also tend to be the most heavily occupied.
The combination of new machines and public Internet access tended
to make these labs very difficult to book and use on a casual
'drop-in' basis. It was therefore necessary to book formal time
slots for these labs in which some of the work could be done and
other labs had to be used for developing the VAR model and doing
the later data manipulation.
The RiskMetrics packages produces comma delimited data files of volatilities and correlations. Handling the link between these large data files and the Excel spreadsheets in which the students had constructed their models required them to make use of some of the more advanced Excel features for manipulating remote data. For some students this caused some difficulties at first and some extra help was given both formally, in a training session, and informally to overcome this problem. The advanced level of the students taking this course however, meant that they were capable of dealing with more advanced problems of this nature and this generally did not cause any significant problems. In the future, more training on the importing features of Excel will be given at the initial training sessions on Excel.
A number of authors have described the use of the Internet in support of their teaching practices. This paper adds to this discussion by describing the setting up of an MBA course in Risk Management at the University of Birmingham, UK, that used the Internet as the medium to access the required data sets in which to carry out prescribed project work. It provides details of the J. P. Morgan RiskMetrics package (methodology and data sets) that was used as the tool set to develop VAR models of Banking risk exposure. It describes the planning stages of the course as they related to the provision of Internet access to the students opting for the course and gives preliminary details of the problems that were faced in putting the plans into action when the course operated. It will be necessary to obtain feedback from the participating students on the completion of the course in order to access more fully the value of the Internet in their learning but as this course has yet to be completed in its first running, this information was unavailable at the time of writing this paper.
The practical limitations of organising web access
to a remote site that transferred information at an unusually
slow rate meant that it was not possible for students to make
wider use of the Internet to personally obtain the data they needed
for their projects. However, they have been given the opportunity
to access the nature of the information available over the Internet
and have had opportunities to consider how it could be used in
the future to solve their information needs.
Bank of England Quarterly Bulletin (1995) Banks'
risk measurement, May. pp180 -183.
Berge, Z. L and Collins, M. (1995) Computer-mediated
scholarly discussion groups, Computers & Education, Vol
24-25, Special Issue on Education and the Internet, eds Kellogg,
W and Viehland, D.
Ibbotson, P, and Grant, L. (1996) Development
of World Wide Web Accounting Tuition Software: A Collaborative
Venture. 7th Annual Conference of the CTI in Accounting Finance
and Management, Brighton, April 2-3.
Ibrahim, A. (1995) Pedagogical Value of the World
Wide Web, in Held & Kugemann, pp 401-407.
Illich, I. (1970) Deschooling Society, Calder
Kuntz, P.S. (1995), Swahili-L - the Internet promotes
literacy, Computers & Education, Vol 24-25, Special Issue
on Education and the Internet, eds Kellogg, W and Viehland, D.
Longerstaey, J. (1995) Introduction to RiskMetricsTM
, IBC Conference, New York, May 18.
Lymer, A. (1995) The Internet and the World Wide
Web in 1994: Key changes and Developments, 6th Annual Conference
of the CTI in Accounting Finance and Management, Glasgow, March
Lymer, A. (1996a) Educational Impacts of the World
Wide Web, ACCOUNT, Vol. 8, No. 1, Spring p 9-11.
Lymer, A (1996b) Creating and Managing Educational
Materials Using the Web, ACCOUNT, Vol8, No. 2. Summer (forthcoming)
Mothobi, C. and J. Strom (1995) Using the Web
as an Educational Support Tool. 3rd Annual Conference
on the Teaching of Computing, Dublin. Paper available on-line
Pickering, J. (1995) Teaching on the Internet
is learning, , Active Learning 2, Themed issue 'Using the
Internet for Teaching', July, p9- 12
Sangster, A. (1995a) World Wide Web - what can
it do for Accounting education ?, 6th Annual Conference of
the CTI in Accounting Finance and Management, Glasgow, March 28-29.
Sangster, A. (1995b) World Wide Web - what can
it do for education ?, Active Learning 2, Themed issue ,
'Using the Internet for Teaching', July, p3- 8
Sangster, A. , and Mulligan, C. (1996) The Integration
of the World Wide Web into an Accounting Systems Course, 7th
Annual Conference of the CTI in Accounting Finance and Management,
Brighton, April 2-3.
Watabe, K. , Hamalainen, M. , Whinston, A. B. (1995)
An Internet based Collaborative Distance Learning System, Computers
& Education, Vol 24-25, Special Issue on Education and the
Internet, eds Kellogg, W and Viehland, D.
Whalley, W, B. (1995) Teaching and Learning on
the Internet, Active Learning 2, Themed issue 'Using the Internet
for Teaching', July, p25-29
Yeomans, K. (1995) Higher Education on the Information
Superhighway: Negotiating on a ramp, , Active Learning 2,
Themed issue 'Using the Internet for Teaching', July, p19-21.