Newsletter Old site

Newsletter Dec 2003

News update 2003-12: December 2003

A monthly newsletter on risk management in financial services,
operational risk and user-developed software from Louise Pryor

Comments and feedback to Please tell me if
you don’t want to be quoted.

Subscribe by sending an email to
Unsubscribe by sending an email to
Newsletter archived at

In this issue:
1. Human error causes loss
2. Random problem in Excel 2003
3. FSA update
4. Seasonal risks
5. Newsletter information

Apologies if you didn’t receive last month’s newsletter; it was
filtered out as spam by some hosts, because it contained a four
letter word (pointless pen without alternative provides salacious
entertainment). The full uncensored version is available at

1. Human error causes loss

Last week two travel companies announced their results. First
Choice achieved a record profit of 47.8m pounds on 2.26bn pounds
turnover. MyTravel achieved a loss of 911m pounds on 4.2bn pounds
turnover. Quite a difference. A number of reasons were given by
MyTravel, but the biggest problem appears to have been that they
got their pricing wrong, so they were selling holidays at a
loss. The operating loss (before exceptional items) was 358m, much
of which can presumably be put down as an operational loss (in risk
management terms).

The large operational losses that reach the newspapers are usually
due to causes such as fraud or a rogue trader. It’s not often that
we see such large losses that are, in the words of Peter McHugh,
MyTravel’s chief executive, due to “human error.” He went on to

“The idea was that we should become more efficient, and people
decided – properly I guess – that in order to do that we needed
to upgrade the technology so we could add accounting and other

“Other than being old, the old systems worked fine. The new
systems were supposed to make things better but, when you do a
major upgrade, the systems need to talk to one another and have
interfaces and, as the new system was implemented, they turned
off the old system and lost the interface between them.

“The execution was poor and a lot of management information
that we used to get disappeared. We went through a period in
2002 when we had less information than we used to have.”

In summary, they thought that a new IT system would change their
lives for the better. When it was deployed, they found that it did
not replace the full functionality of the old system. They were
missing information that had previously been available, and that
was used in their pricing models. This made them get their prices
wrong. Ouch.

This story illustrates some well known truisms.

First, change is risky. It really is one of the major sources of
risk. People say this often, and they are right.

Second, no IT system is a silver bullet. One of the seminal essays
in software engineering is one by Fred Brooks, in his classic
collection “The mythical man-month.” The essay is entitled “No
silver bullet – essence and accident in software engineering” and
its summary reads: “There is no single development, in either
technology or management technique, which by itself promises even
one order-of-magnitude improvement within a decade in productivity,
in reliability, in simplicity.” Like many of his essays, the
overall message is not limited to software engineering.

Third, garbage in, garbage out (GIGO). If you haven’t got good
information going in to a model, you are most unlikely to get good
results out however much effort you put in. The most elaborate
algorithms can’t entirely compensate for poor or missing data.

Fourth, an IT system doesn’t exist in isolation. It has to connect
to other systems, and be used by people. These interfaces are often
more complex than is realised.

Finally, we can’t put all the blame on the specific IT system
involved in this mess. It was a new central reservations system,
the same system that First Choice uses.

2. Random problem in Excel 2003

Don’t use Excel 2003 if you use the RAND() or RANDBETWEEN()
functions in your spreadsheets. It appears that a bug crept into
the final release that wasn’t present in the betas. RAND() was,
apparently upgraded so that it produced a better distribution of
pseudo-random numbers between 0 and 1. The trouble is that the
numbers it produces may be more random, but aren’t always in the
correct range. Oops!

There has been some discussion of this on various mailing lists and
newsgroups. Apparently the problem is reasonably easy to replicate
if you have Office 2003 (I don’t, so I can’t confirm this).
Microsoft has not yet acknowledged that there is a problem, let
alone done anything to solve it.

Many people who use Excel don’t use the built in statistical and
random number functions anyway, as in the past they have been
inaccurate in some circumstances (see my November newsletter for
some discussion of this). The more sophisticated users take their
statistical functions from other sources, or write their own. It
seems that they are right to do so.

The original reports of the bug came from Woody’s Watch at

3. FSA update

Admittedly it’s less than a month since my last newsletter, but the
number of new consultation papers and feedback papers is still very
low. On the other hand, there have been a lot of Dear CEO letters,
final notices and speeches.

Among the latter was a speech by David Strachan emphasising that
capital adequacy is not a substitute for effective risk management
( He was talking
about insurance companies, but the point is valid across the whole
of the financial services sector (and indeed beyond).

New consultation and discussion papers out this month:

CP207 Treating with-profits policyholders fairly

DP25 Development of transaction monitoring systems
DP26 Developing our policy on fraud and dishonesty

Feedback published this month:

CP155 Tier 1 capital for banks: Update to IPRU(BANK)

Current consultations, with dates by which responses should be
received by the FSA, are listed at

4. Seasonal risks

If you’re planning to go down any chimneys this month, make sure
you’ve done the appropriate contingency planning. Is your model of
the relative circumferences of you and the chimneys using up to
date inputs? Is the assumption of cylindricality appropriate? You
also have to consider what you would do in the event of a transport
failure (Rudolph may get breathalysed and done for being drunk in
harness). Global warming may affect your speed of delivery, as the
sleigh runs slower on sludge than snow. Your processes are
important too. I wouldn’t recommend either wrapping all the
presents before labelling any of them, for example, or labelling
any of them before wrapping. All in all, an operation full of
operational risk.

Even I couldn’t work in a reference to how important it is to test
your spreadsheets and financial models in the rest of the
newsletter, so I’m doing it here. XLSior, the Excel add-in I have
developed that provides automated testing, documentation and more
now has a new pricing structure: see for
details. There have been several releases over the last couple of
months to fix bugs and provide better functionality. Another one is
due towards the end of this month; it will include the ability to
use passwords when using XLSior to protect and unprotect multiple

Have a very happy Christmas, and all the best in 2004.

5. Newsletter information

This newsletter is issued approximately monthly by Louise Pryor
( Copyright (c) Louise Pryor 2003. All
rights reserved. You may distribute it in whole or in part as long
as this notice is included. To subscribe, email To unsubscribe, email All comments, feedback and other
queries to Archives at