Accounting Variables, Deception, and a Bag of Words: Assessing the Tools of Fraud Detection
Bibliographic record
Abstract
Abstract We develop a data‐generated tool for distinguishing between fraudulent and truthful reports based on the language used in the management discussion and analysis section of annual and interim reports. Using this method, we are able to assign a probability of truth to each report which is then shown to be an effective indicator of fraud. Our work goes beyond the development of a tool alone, however, by conducting an extensive comparison of our probability‐of‐truth measure with eight alternative detection tools representing both quantitative and language‐based approaches. Comparisons are made across a variety of samples and show that our language‐based approach can be effective in both cross‐sectional and time‐series settings. It is useful both in distinguishing between fraudulent and truthful firms and in identifying fraudulent reports from a series of reports issued by a single firm. This second setting is one in which accounting‐based detection tools have frequently struggled. We establish that, not only is our probability‐of‐truth measure significantly associated with fraud, so too is the change in this measure from a firm's previous reports. Prior reports may serve an important benchmarking role in using language‐based tools to identify fraud.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.038 | 0.395 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.016 | 0.008 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.007 | 0.010 |
| Open science | 0.002 | 0.004 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".