Ventana software support forum

Posted by ""Swanson, John"" <John.Swanson@sdgworld.net>
Can anyone point me to published work on System Dynamics and fraud prevention?

I have the following situation in mind. Large public transport systems will usually have people travelling without valid tickets, and the revenue loss can be substantial. So the operator might employ inspectors to patrol the system checking tickets and fining people who haven't paid. If you use enough inspectors people may see that they won't get away with it and fraud levels fall. Then management says 'why are we employing all these inspectors, there is no fraud?' and reduce their number ... so fraud rises again. It looks like a nice problem about finding the right balance between the cost of the inspectors and the lost / recovered revenue.

I've checked the Society's list of references, and sources like Sterman but not found anything directly analogous.

John Swanson
Associate
Steer Davies Gleave
28-32 Upper Ground, London SE1 9PD
e. John.Swanson@sdgworld.net
w. www.steerdaviesgleave.com
Posted by ""Swanson, John"" <John.Swanson@sdgworld.net>
posting date Fri, 26 Aug 2005 14:10:01 +0100

Posted by ""Eliot Rich"" <e.rich@albany.edu>
Hello.

The work on the Dynamic Trigger Hypothesis, presented at the 2004 ISDC by Andersen et al., has some elements that you might find useful. It depicts shifts in security vigilance over time due to economic and social pressures within an organization. Our particular context was the problem of insider threat detection.

Regards,

Eliot Rich
Posted by ""Eliot Rich"" <e.rich@albany.edu>
posting date Sat, 27 Aug 2005 08:05:40 -0400 (EDT)

Posted by Joel Rahn <jrahn@sympatico.ca>
This situation sounds like the Fear & Greed archetype I presented at a poster session at the Boston conference. There is a very small model associated with the paper on the conference website. The paper itself was inspired by earlier work by Weaver, et al. and seems to be related to the Dynamic Trigger Hypothesis mentioned by Eliot Rich in SD5477.

In terms of the 'Fear & Greed-a political archetype' poster paper: The decision threshold could be 'required evidence for ticket inspection' or 'fraction of transit lines with no ticket inspection', the false positives could be 'the number of those inspected who have a ticket', the false negatives could be 'the number of those not inspected who don't have a ticket'.

Then, increasing the threshold means
a) fewer ticket-holders (innocent, non-fraudulent travelers) are inspected and
b) more non-ticket-holders (guilty, fraudulent travelers) are not inspected. In this way, pressure to increase the threshold (effectively decrease the
inspection) comes from those with a Fear of loss of 'liberty' (i.e., freedom to ride without intrusive inspection) and pressure to decrease the threshold (effectively increase inspection) comes from those Greedy for gain of control (i.e., greater assurance of non-fraudulent behavior) and for greater revenues from fines.

John Swanson's mental model adds a new element, not dealt with in the archetype, in which the proportion of fraudulent behavior could change (in the same direction, after a delay) in response to changes in the threshold. Going out on a limb here, this feature could generate a long-term sequence of 'Fixes that fail' each 'failure' being in the direction opposite the preceding: 'too much fraud', 'too much control (hassle)',... Posted by Joel Rahn <jrahn@sympatico.ca> posting date Sun, 28 Aug 2005 10:19:34 -0400

Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr> Hi John,

You can find on the Vensim forum system dynamics subject a simple model that I have starded on the subject. But by lack of time, I have not checked it completely. So it has probably bugs in it, or the equations are too simplistic, generally linear and when optimizing it reaches some values and it behaves strangely. When I started it yesterday, I thought it was simple, it is not. Vensim site is : http://www.ventanasystems.co.uk/forum/
Regards.
J.J. Laublé Allocar
Strasbourg France.
Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr> posting date Sun, 28 Aug 2005 13:17:55 +0200