Everyone should be interested in the decline effect. It affects every scientific field, and has practical implications for every profession. But of course to really make it useful for each profession, we have to talk about specifics. Out of those specifics will come new tools.
Since the spring I have been preoccupied with one particular project, applying the maximum entropy principle to climate. Now with the climate paper up on the blog and gradually reaching its intended audience, I have a number of other topics I want to go back and cover. One is applying the maximum entropy principle to forensic auditing. Let me give the context.
My ‘day job’ for the past decade has been in computer aided auditing. For much of that time I was primarily a publisher of books on advanced forensic techniques, like Dave Coderre’s excellent work Fraud Detection. Dave is now published by Wiley, and my main activity in the past few years has been statistical consulting.
But when I first came into the field, I wrote technical manuals and newsletters for ACL Services Ltd., makers of the famous ACL auditing tool. I had an opportunity then to study their customer database (using their own analytical software!) and to form some ideas about the ‘early adopters’ of the still relatively new computer-aided audit technology (CAAT).
Over the past two decades, ACL has probably sold in excess of 100,000 copies of its product. It is a familiar name to auditors. But the number of people who use the tool to its full potential is small. Many auditors use it as a kind of ‘can-opener,’ to extract data from mainframe accounting systems and export it to Microsoft Excel. From that point their analysis of the data winds up being limited to whatever functions they are familiar with in Excel. The library of more powerful analytical functions in ACL goes largely unused.
I concluded back in 2001 that most of the success ACL had achieved was due to a few hundred power users, or ‘champions’ as I called them. The ‘champions’ did not need to be sold on the value of computer-aided audit. They were the ones who organized seminars, who lobbied management to buy ACL for the entire audit department, who wrote custom programs to increase productivity. There was a familiar-looking skewness to the distribution of sales. Companies with a ‘champion’ bought multiple copies of the product, kept paying support fees, sent staff to training. But far more numerous were companies with no ‘champion,’ who typically were talked into buying one copy, or perhaps two, and then after a year or so, let their support lapse.
Today, a decade later, the CAAT interest group on LinkedIn still sits at fewer than 1,000 members. Posts concerning CAAT topics on the Institute of Internal Auditors website are relatively rare, compared with posts about audit checklists, or regulatory requirements, or benefits packages for auditors.
Make no mistake here, the benefits of CAAT are huge and I remain very enthusiastic about them. A single skilled auditor can use CAAT to recover over-payments and other kinds of processing error equal to his or her entire departmental budget. But it seems there just aren’t that many people with the right mix of skills and interests. It is a niche that cries out for more attention.
If you are new to the subject of CAAT, there are already several excellent sources of information, starting with Auditsoftware.net, the website established by my colleague and longtime associate, Richard Lanza. Among our projects was the Buyer’s Guide for Audit Software, which is still available on Amazon.com. Another excellent site is AuditNet, established by Jim Kaplan. A good recent book covering CAAT is Forensic Analytics by Mark J. Nigrini. I won’t attempt to duplicate these resources on this blog.
Instead I plan to continue where my newsletter Frequencies (see the download page) left off, nine years ago, talking about the application of size laws and the maximum entropy principle, and sampling and trend testing. This is a niche within a niche, so to speak. Within the community of auditors skilled in CAAT there is a smaller community who are sufficiently math- and science-oriented to find size laws interesting, or to tackle advanced sampling theory.
I plan to post occasional pieces on how the decline effect can be used to do financial analysis, spotting trends, finding anomalies, adjusting sampling and significance testing, and generally getting a deeper understanding of financial data. I also plan to write about the growth of the CAAT community, and possibly to do some training videos.
The audience for this may not be especially large — the audience for my ‘Pot Lid’ climate paper is larger, and the potential audience for my book is much larger — but those who are interested tend to be intensely interested. There is a small but growing market for consulting services in this area. And there are spinoff benefits like adaptive sampling that may wind up mattering to a much wider audience.