Key Takeaways
- 52% of frauds were detected by tips, the most common detection method for embezzlement.
- Frauds lasted a median of 12 months before detection.
- 15% of frauds detected by internal audit.
- The global median loss from occupational fraud in 2022 was $117,000, with embezzlement being the most common type.
- Median loss from billing fraud was $100,000 in 2022.
- Cash on hand theft median loss $30,000.
- Organizations recovered a median of 0% of their losses from embezzlement schemes.
- Only 13% of cases resulted in full recovery of losses.
- 61% of cases led to prosecution.
- 42% of occupational fraud perpetrators had been with their employer for more than 5 years.
- Executives were responsible for 23% of frauds despite being only 9% of perpetrators.
- Median age of perpetrators was 41 years.
- In 2022, asset misappropriation schemes, including embezzlement, accounted for 86% of all occupational fraud cases reported.
- Billing schemes represented 22% of asset misappropriation cases in 2022.
- Check tampering accounted for 11% of schemes.
Embezzlement is often detected by tips within about a year, yet losses rarely get recovered.
Related reading
01 · Category
Detection Methods26 stats
Detection Methods Interpretation
02 · Category
Financial Impact27 stats
Financial Impact Interpretation
03 · Category
Legal and Recovery28 stats
Legal and Recovery Interpretation
More related reading
04 · Category
Perpetrator Profiles25 stats
Perpetrator Profiles Interpretation
05 · Category
Prevalence and Incidence29 stats
Prevalence and Incidence Interpretation
Cite This Report
This report is designed to be cited. We maintain stable URLs and versioned verification dates. Copy the format appropriate for your publication below.
Lukas Bauer. (2026, February 13). Embezzlement Statistics. Gitnux. https://gitnux.org/embezzlement-statistics
Lukas Bauer. "Embezzlement Statistics." Gitnux, 13 Feb 2026, https://gitnux.org/embezzlement-statistics.
Lukas Bauer. 2026. "Embezzlement Statistics." Gitnux. https://gitnux.org/embezzlement-statistics.
Sources & references
15 datasets cited across this report · attribution is report-level

