You’re an accountant! You’re in a noble profession! The word “count” is part of your title!
Mel Brooks, screenplay for The Producers
There have been quite a few thinkers who emphasized the fundamental importance of counting things correctly, including James Clerk Maxwell, Bertrand Russell, William Edwards Deming and Peter Drucker. Whether scrutinizing science, engineering, business or social policy, there is little disagreement about the need for accurate measurements if we want to properly understand the world and whether our attempts to influence it have been successful. But because we live in an attention economy, I already know that the title of this post guaranteed it will be ignored by the kinds of people who prefer to read an article with the following headline.
Sim-swap fraud rises by 1,000% as criminals seek to exploit growth of two-factor authentication
That article correctly states the following about the UK’s National Fraud Database.
The National Fraud Database says it has seen a 1,055% increase in the number of reports of sim-swap fraud, rising to almost 3,000 cases in 2024 from just 289 in 2023.
I picked upon an article from ITV, one of the largest providers of news to the British population, to emphasize how widely this statistic has been repeated. The same percentage increase in SIM swap fraud has also been cited in communications sector publications, social media posts and many more places than I care to count. None that I saw advised the following.
We asked Action Fraud, the national reporting centre, to tell us how many reports of Sim-swap fraud it has received since 2020.
It said reporting volumes halved between 2020 and 2021 and remained stable the following year, but almost doubled from 558 in 2022 to 1,070 in 2023, reaching 2,037 at the end of November 2024.
That quote was taken from an article by Which?, the journal of the highly-respected Consumer’s Association. Over the years, Which? has been pretty consistent in relying upon figures from Action Fraud, the entity that collates fraud data on behalf of Britain’s police forces. Meanwhile, the National Fraud Database is maintained by CIFAS, a long-established nonprofit entity trusted to collate and disseminate fraud intelligence on behalf of the government and numerous businesses. The UK therefore has two completely separate and independently managed sources of statistics on the frequency of fraud and they produce figures which wildly disagree with each other. Both sources get reported as if they are reliable even though they differ so greatly.
To reiterate for those with an attention deficit disorder, the UK has something called a national fraud database that says SIM swap frauds increased by a factor of ten between 2023 and 2024, while the UK also has a separate national fraud reporting center that says the number of SIM swap frauds increased by a factor of two over the same period. 10 times versus 2 times. Evidently they agree that SIM swap fraud is a worse problem than before but there is a huge disparity between the figures produced by the ‘national’ database for fraud and the ‘national’ reporting center for fraud. At least one of these national bodies has failed to capture the truth about SIM swap fraud. Perhaps neither of them has.
The particular problem with these statistics is that there is not even a credible explanation of who might be undercounting, who might overcounting, and why they are at odds with each other. In 2023, Action Fraud produced a number for SIM swap frauds that was almost four times as large as the number per CIFAS. One year later, the Action Fraud statistic is roughly two-thirds of the figure quoted by CIFAS. If there is a systematic bias that leads one or other figure to under or overrepresent the number of SIM swaps then you would expect the ratio between the two sources to remain approximately the same.
Whatever is wrong with these numbers, it cannot be explained away by the old canard that ‘probably the number of crimes was underreported’. Even if it is generally true that fraud is underreported, no explanation is offered for why there might have been a dramatic change in the propensity to report frauds. It does not take a lot of expertise to remember there are two independently managed sources of statistical data on trends in fraud and then to compare them to see how well they agree with each other. Statistics are being cherrypicked when they should be critiqued.
It speaks volumes about the near total disinterest in seeking reliable measures of fraud that Professor Alan Woodward and Daniel Gardham, both of the University of Surrey, also chose to repeat the 1,055% statistic without any additional context when writing a recent article for The Conversation, an outlet that boasts about its academic rigor! Woodward’s track record is that of a shameless rent-a-quote who tirelessly promotes himself. Nevertheless, any academic should have applied basic skepticism to this particular statistic, not least because experts working in this domain within the UK should be aware of the conflict between the numbers from the national fraud reporting center and the national fraud database. When a statistic about crime leaps upwards from a very low base there should be immediate suspicion that the large proportionate change was due to previous underreporting. Instead of being cautious when repeating statistics like this, academics like Woodward jump on the bandwagon because they know it increases their chances of being quoted by low-quality sensation-seeking news providers like The Sun.
I do not wish to explore this topic any more because ultimately I am not interested in unreliable statistics even if that places me in a minority. The point is already made: people are seeking to influence policy despite not knowing if the figures they cite are reliable. Or maybe they only really care about clickbait. There are now too many organizations that compete with each other for the job of enabling collaboration through collating and exchanging intelligence. They are competing for the money that comes with that role. That means they compete for attention. Where does that leave the oft-stated ambition to ‘restore’ trust? It is unlikely that trust will be restored if we have no trustworthy way of determining if anti-fraud policies are working or not.
Restoring trust will remain only a vacuous slogan if there is no serious intention to count the number of consumer crimes correctly. The problem with counting is that it is difficult and it is boring. Nobody wants to pay for it, but cheap measures tend to be poor measures. Counting is the kind of thing that accountants specialize in, and they expect to get paid a fair reward for their valuable skills. Accounting is also the opposite of the kinds of things that generate quick and easy interest from audiences who want to feel like they know what is happening but do not care to question the quality of information provided to them. If you have a prejudice that fraud is getting worse then somebody will produce the statistic you wanted. And if you have the opposing prejudice, perhaps because you work for the government or a vendor of anti-fraud solutions, then somebody else will produce the statistic you sought. Everybody gets to feel the world conforms to their expectations and nobody is forced to tackle the world as it really is. These are ideal conditions for breeding cynicism, not trust.
We need more people to check facts and fewer people repeating fictions. The profession with the skills to check numerical facts is the accounting profession, and the anti-fraud profession is especially short of decent accountants. When it comes to tackling fraud, there are lawyers, and data scientists, and engineers, and even a few criminologists, but not enough accountants, as should be obvious to anyone who spends more than a second questioning the reliability of fraud statistics that get bandied around. I choose the word ‘accountant’ deliberately because the point is not that somebody counts — most people ultimately know how to count — but that only some professions can be trusted to count impartially, and to report any failure to count impartially. That nobody challenges the overconfident repetition of dubious statistics is proof that the anti-fraud domain lacks people with the integrity required to count things correctly and to apply professional skepticism when one statistic contradicts another statistic. Counting is a noble profession, and one of the most important skills that we lack.
