Fraud and Deception Detection: 5 Language Fingerprints
Final month, I described how computer-aided text-based evaluation may help uncover fraud and deception in firm communications. However what different insights can we glean from this analysis into scandal corporations?
We used Deception And Reality Evaluation (D.A.T.A.) to look at 10 of the biggest company scandals in current historical past and located that the typical lead time between our textual identification of deception and the general public recognition of potential scandal was greater than six years.
Company Scandals: Time between Textual Proof and Public Recognition
| Ticker | Firm | Dimension, in US Hundreds of thousands | Scandal 12 months | Common Alert Rating in Lead-Up | Common Alert Rating Pre-Scandal | Years Warning |
| ACC | Adelphia | $2,300 | 2002 | -46% | -44.8% | 2 |
| AIG | AIG | $3,900 | 2005 | -30.6% | -52.4% | 12 |
| CUC | Cendant | $640 | 1998 | -37.9% | -48.8% | 3 |
| ENRN | Enron | $74,000 | 2001 | -87.4% | -76.3% | 8 |
| HLS | HealthSouth | $1,400 | 2003 | -42.2 | -27.1% | 9 |
| LEH | Lehman Bros. | $50,000 | 2008 | -37.2% | -3.8% | 13 |
| SAY | Satyam | $1,400 | 2009 | -28.9% | -38.4% | 6 |
| TYC | Tyco Worldwide | $600 | 2002 | -77.1% | -81.7% | 7 |
| WCOM | WorldCom | $3,800 | 2001 | -33.9% | -47.9% | 4 |
| WM | Waste Administration | $6,000 | 1997 | -39.4% | -41.1% | 2 |
| Whole | $144,290 | Common | -40.3% | 6.6 |
The apparent query is why. Why does it take regulators and markets so lengthy to acknowledge these scandals? And a follow-up query: What insights from text-based evaluation can we use to raised establish these scandals earlier? Let’s take these in flip.
Principle: It’s the Conduct
Why does D.A.T.A. detect deception quicker than acutely traders and regulators? After eager about this for some time, we developed a concept, and it boils right down to 86.5%. That’s the proportion of economic data that’s expressed in textual content, not in numbers, in annual experiences. Textual content communications reveal the conduct of company administration groups, and that conduct results in the result that’s expressed in numerical efficiency.
In order that 6.6 years between the preliminary indication of deception and when the scandal breaks is the typical size of time {that a} poorly behaving agency can faux it, till they only can’t therapeutic massage the numbers any longer.
What’s fascinating is that the 2 scandals that took over a decade to acknowledge each concerned monetary corporations: AIG and Lehman Brothers. Their annual experiences ran within the lots of of pages, and the speed of cash biking by way of their steadiness sheets and revenue and money stream statements was very, very excessive. Thus, it took appreciable time for his or her poor behaviors and selections — the inputs — to finally present up within the numbers, or the outputs.
If this concept is a legitimate rationalization for that lead time, then scandal must have language fingerprints that traders can mud for as both an early warning system or as a second opinion on the conventional elementary work that funding analysis groups conduct.
Language that Reveals Doable Scandal
After inspecting the ten scandals above in addition to Wirecard and different newer controversies, we recognized 5 textual fingerprints that differ from these of extra truthful corporations by greater than 50%.
Scandal Phrases and Firm Communications
| Language Fingerprint | Incidence Relative to the Imply |
| Phrases Indicating Friendship | +56.1% |
| Phrases Indicating Threat | +55.9% |
| Impersonal Pronouns | +54.1% |
| Phrases That Point out Variations | -53.6% |
| Phrases That Negate a Assertion | +50.4% |
Along with text-based evaluation, we additionally performed one-on-one conversations to raised discern between deception and reality and to establish a number of the extra pan-cultural misleading behaviors individuals have interaction in. Our findings aligned with what earlier lie detection researchers had uncovered: that every of the 5 potential deception indicators that floor in text-based evaluation additionally happen in person-to-person interviews.
So let’s drill a bit deeper into every of them.
1. Phrases Indicating Friendship
Lie detection researchers have proven that deceivers typically make use of obfuscation to create confusion. A technique they do that is by utilizing phrases that suggest friendship extra typically than the norm in enterprise communications. Misleading corporations make use of such phrases 56.1% greater than the typical, based on our evaluation. So if an annual report consists of quite a few ingratiating phrases, it could be proof of obfuscation and deception.
However a distinction is essential right here: Phrases that point out friendship — “pal,” “pal,” “neighbor,” and “gang,” for instance — are totally different from pleasant phrases.
2. Dangerous Phrases
Scandal companies favor phrases that point out danger at a a lot greater proportion than the typical firm. These embrace such phrases as “averse,” “keep away from,” “concern,” “problem,” “forestall,” “stopped,” and so forth. A lot of these phrases already have a tendency to lift securities researchers’ hackles, and as we identified within the final piece, companies are proactively excising these sorts of “purple flag” phrases from their annual experiences.
3. Impersonal Pronouns
“One other,” “everyone,” “somebody,” and “whichever” are the form of impersonal pronouns that dishonest companies make use of to a a lot larger extent — 54.1% extra typically — than their truthful friends. Why do they like to be impersonal of their communications? Researchers theorize that they’re making an attempt to create emotional house between themselves and people they want to mislead.
4. Phrases That Point out Distinction
Mendacity is cognitively demanding. One manifestation of that is that throughout the act of deception, the liar is usually unable to make distinctions amongst competing factors of view of their communications and so are much less probably to attract comparisons. So using phrases that recommend distinction is definitely a sign of truthfulness. Constructions that current contrasting viewpoints — “as in contrast with different years . . .” — are examples of this.
Deceivers even have an agenda: to persuade their goal to imagine their most well-liked narrative. They’re unlikely to attract distinctions between different narratives and can are inclined to deal with their most well-liked one.
5. Phrases That Negate a Assertion
Analysis additionally signifies that liars typically make use of extra unfavorable phrases than reality tellers. Because of this we drew the excellence between phrases indicating friendship and phrases which might be pleasant.
However researchers don’t at all times discover that the deceivers are extra unfavorable than the truthful. Our evaluation of dishonest agency communications suggests, nevertheless, that they have an inclination to make use of such phrases as “not,” “by no means,” “shouldn’t,” “doesn’t,” and “should not” at a 50.4% larger proportion than the typical.
Bonus
So what’s by far the strongest indicator of deception? The variety of swear phrases in an annual report. Although they’re rarities, swear phrases happen in scandal firm annual experiences a whopping 277.1% extra continuously than the imply.
In case you favored this put up, don’t overlook to subscribe to the Enterprising Investor.
All posts are the opinion of the writer. As such, they shouldn’t be construed as funding recommendation, nor do the opinions expressed essentially mirror the views of CFA Institute or the writer’s employer.
Picture credit score: ©Getty Photos / Matthias Kulka
Skilled Studying for CFA Institute Members
CFA Institute members are empowered to self-determine and self-report skilled studying (PL) credit earned, together with content material on Enterprising Investor. Members can document credit simply utilizing their on-line PL tracker.
