Contents
In early November 2026, two stories broke almost at once. Indonesia's financial services authority, OJK, sanctioned seven listed companies. In one case, it found that material assets in the financial statements "could not be supported by underlying documents". The directors and the public accountant who audited those statements were sanctioned too12.
The same week, a case surfaced at a rural bank (BPR) in Malang. Its president director and one other director are now suspects. They allegedly opened credit facilities without the borrowers' knowledge, took gold collateral that was never recorded, and left dozens of deposit certificates off the books10.
Different industries. Different scale. But the pattern is similar: the transactions were in the system, the documents could be checked, and the problem only came to light after running for years.
- Fraud typically runs 12 months before it is found. The median loss is USD 104,000 per case, across 2,402 cases in 143 countries1.
- Fraud is high on Indonesian internal audit's radar. 41% of Indonesian respondents put fraud in their top five risks, the highest of the four Southeast Asian countries IIA breaks out3.
- How fraud is found decides how long it runs. Through automated data monitoring: 7 months. Through law enforcement: 24 months2.
- Proactive monitoring roughly halves losses, from USD 150,000 to 70,000, and shortens schemes from 16 to 9 months. Yet only 17% of small organisations have it2.
- Five patterns recur in 2026 cases, and all of them leave traces in the data. The 10-point checklist in section 05 is ready to use before year-end close.
01The 2026 numbers
The numbers you should know.
Fraud is rising on the radar of Indonesia's internal audit leaders. The IIA's Risk in Focus 2026/2027 Asia Pacific survey found that 41% of 86 Indonesian respondents rank fraud in their organisation's top five risks3. In Indonesia's column, fraud is fourth, after cybersecurity (70%), digital disruption including AI (52%), and geopolitics (50%).
Source: IIA, Risk in Focus 2026/2027 Asia Pacific, Exhibit 3.23. Survey run 2 March to 6 May 2026. IIA breaks out only these four Southeast Asian countries.
The CAE roundtable in the same report adds a useful note. In their view, fraud risk is probably higher than the rankings suggest, because much of it is counted under cybersecurity3.
Almost one in five suspicious-transaction reports points to fraud. In 2025, Indonesia's financial intelligence unit, PPATK, received 183,281 suspicious-transaction reports. The most common indicated predicate crimes were gambling (47.49%), fraud (18.71%), and corruption (5.73%)4.
In the public sector, the numbers remain large. The supreme audit board's first-half 2026 summary (BPK IHPS I 2026), submitted to parliament on 6 October 2026, covers 762 audit reports. They include problems worth Rp45.6 trillion in losses, potential losses, and revenue shortfalls56.
For small banks, fraud can mean closure. OJK revoked the licences of 20 rural banks (BPR/BPRS) in 2024 and 7 in 2025. According to OJK, the causes were fraud incidents and/or inadequate governance and prudence7. In 2026, up to 3 November, the count had reached 16 banks, most of them BPR/BPRS8.
02The problem is time
The problem isn't how much. It's how long.
ACFE's Occupational Fraud 2026: A Report to the Nations analyses 2,402 fraud cases in 143 countries1. Several findings apply directly to internal audit teams in Indonesia.
Fraud runs a long time before it is found. The median case lasts 12 months. The median loss is USD 104,000. Fraud examiners estimate organisations lose about 5% of revenue a year to fraud1.
How fraud is found decides how long it runs.
Source: ACFE, Occupational Fraud 2026: A Report to the Nations, Fig. 152. Median duration in months, global data.
Asia Pacific leans heavily on tips. The region's median loss is higher, at USD 150,000. 52% of cases are found through a tip, and only 1% through automated data monitoring. Worse, 66% of Asia Pacific victims recover nothing. Indonesia contributes 20 of the 143 Asia Pacific cases in the report2. That is too few to support an Indonesia-only figure, so we don't quote one.
Proactive data monitoring cuts both losses and duration. Organisations with proactive data monitoring and analysis controls record smaller median losses and catch fraud sooner2.
Source: ACFE, Report to the Nations 2026, Figs. 28 and 292. 58% of large organisations have this control, but only 17% of small ones (fewer than 100 employees).
One point is often missed. Globally, automated monitoring is the first way only 3% of cases are found2. Yet its effect on losses and duration is large. So the value of monitoring isn't only about catching people. It narrows the room to move, makes fraud surface sooner, and provides evidence when a tip does come in.
Monitoring rarely finds fraud first. But it cuts its life almost in half.
03Five patterns
Five patterns that recur in 2026 cases.
Across the cases regulators and law enforcement disclosed in 2026, five patterns keep showing up. All of them leave traces in the data. For cases still in progress, everything described here is alleged, and the people involved are suspects until a court rules.
Pattern 1Recorded, but not real
In the Malang rural bank case, OJK alleges 71 credit facilities to 31 borrowers worth Rp15.6 billion were opened without the borrowers' knowledge1011. In the BPR Panca Dana case in Depok, fictitious loans were allegedly used to push down the non-performing loan ratio9. The data traces: borrower records that look alike, disbursements to accounts the borrower doesn't own, or restructuring just before a loan turns non-performing.
Pattern 2Real, but not recorded
In the same case, there is alleged gold collateral of about Rp600 million that was never recorded, about Rp5.8 billion of cash advances left off the books, and 78 deposit certificates worth Rp32.9 billion that were not recorded. The transactions under investigation total about Rp54.9 billion. That is a transaction value, not a proven loss. The case came to light through OJK supervision, and the file went to prosecutors at the end of October 202610.
Pattern 3Assets without supporting documents
In the case of listed company ASLI, OJK found that material assets in its first-half 2024, full-year 2024, and first-half 2025 statements could not be supported by underlying documents1213. The lesson is simple: every figure on the balance sheet should trace back to a source document, and that tracing shouldn't happen only once a year.
Pattern 4Affiliated transactions with incomplete procedures
Five other companies in the same OJK announcement were sanctioned over procedures for affiliated or material transactions12. These are procedural breaches, not allegations of embezzlement. But they are a reminder that related-party transactions need watching from the start, not after the report is published.
Pattern 5Controls bypassed from the top
ACFE finds the main causes of fraud are missing internal controls (33%), overridden controls (19%), and no management review (18%). When the perpetrator is an owner or executive, the median loss is USD 475,000 and the scheme runs 23 months2. The rural bank cases involving directors show the same thing: a control that relies only on a superior's approval isn't enough when the superior is the problem.
The transactions were in the system. Reading them came late.
04Red flags
Behaviour alone isn't enough.
According to ACFE, 84% of fraudsters show at least one behavioural red flag. The most common: living beyond their means (39%), financial difficulties (29%), and an unusually close relationship with a vendor or customer (17%)2.
But 16% show none at all. That is why behavioural red flags need to be paired with data red flags. Several behaviours cast a shadow in the system.
| Behavioural red flag | Its shadow in the data | Data source |
|---|---|---|
| Too close to a vendor | Vendor address, phone, or bank account matches an employee's. One approver almost always approves the same vendor | Vendor master, HR, approval log |
| Won't share duties | One user creates a new vendor and releases its payment | ERP user log |
| Never takes leave | No leave on record, yet active in transactions all year | HR, ERP user log |
| Living beyond means | Expense claims that keep rising, or sit just under the approval limit | Expenses, payroll |
| Financial difficulties | Repeated cash advances, extended staff loans, advances left unsettled for a long time | Cash, staff receivables |
Behavioural red flags from ACFE, Fig. 552. The data pairs are example analytic tests. Some, such as matching vendors to employees, are standard tests in UK public-sector data analytics guidance20. One red flag isn't proof. It is a reason to look closer.
05Checklist
A 10-point checklist before year-end close.
Use this as a starting point for the year-end review. You don't have to run everything at once. Answer honestly: "Done" means the test was run in the last three months and the result is documented, not "planned".
0 of 10 done
| # | Area | Question | Done / Not yet | If not yet, start here |
|---|---|---|---|---|
| 1 | Vendors vs employees | Have vendor addresses, phone numbers, and bank accounts been matched against employee data? | Export the vendor master and employee data, and match bank accounts first, since they are least likely to have typos | |
| 2 | Changed vendor accounts | Has every vendor bank account change in the last 90 days been checked, including the first payment after it? | Pull the vendor master change log, then confirm with each vendor on the number already on file | |
| 3 | Duplicate and dormant vendors | Are there vendors with similar names, the same tax ID, or the same bank account, and old vendors that suddenly came back to life? | Sort vendors by tax ID and bank account, then flag vendors with no activity for 12 months that reappear | |
| 4 | Duplicate payments | Have you searched for payments with the same or nearly the same invoice number, amount, and date? | Start with the last 12 months of payments: same vendor and amount within 30 days | |
| 5 | Just under the limit | Are there invoices or purchases just under the approval limit, or split to avoid a tender? | Count transactions between 90 and 100% of the approval limit, by requester and by vendor | |
| 6 | One user, two roles | Are there transactions created and approved by the same user, or purchase orders raised after the invoice arrived? | Compare the creator and approver columns in the ERP log, then PO dates against invoice dates | |
| 7 | Manual journals | Have manual journals posted outside working hours, at period end, or with large round amounts and no documents been checked? | First confirm the journal population is complete, then filter manual journals in the last five days of each period | |
| 8 | Payroll and expenses | Are there bank accounts shared by more than one employee, salaries paid after a leaving date, or claims made while the employee was on leave? | Match payroll bank accounts across employees, then HR leaving dates against the last salary transfer | |
| 9 | Assets, collateral, deposits | Can material assets, physical collateral, and deposit certificates be traced to source documents and do they match the system? | Take the ten largest assets or collateral items and match their contracts, certificates, and handover records | |
| 10 | Urgent transfers | Do large transfers and bank account changes always need two approvals and a call-back to a number on file, even when marked "very urgent"? | Write a one-paragraph rule: no two approvals and no call-back to a number on file, no transfer |
06Where to start
Start without a big project.
Many internal audit teams already know the list above but struggle to run it continuously. A few principles help.
- Step 1
Start with 5 to 10 tests. Prioritise the areas with the most risk. For many companies, that is the vendor master, payments, and payroll. For financial institutions, add credit.
- Step 2
Name an owner for every test. Process owners follow up exceptions. Risk or compliance sets the rules and thresholds. Internal audit tests whether the monitoring works, and doesn't run it day to day, so its independence stays intact18.
- Step 3
Avoid an alert flood. ISACA advises starting with a few alerts on key controls and using pattern-based thresholds rather than single events19. When there are too many alerts, teams stop reading them.
- Step 4
Record every follow-up. Who reviewed it, when, the conclusion (false positive, control gap, or needs investigation), and the supporting documents.
- Step 5
Report to the audit committee regularly. Not just the number of findings, but how long exceptions stay open and how many control gaps have been closed.
Don't forget the reporting channel. In Asia Pacific, more than half of cases come to light through tips. Make sure the whistleblowing channel is open to vendors and customers, not only employees, and that line managers know how to pass reports on. According to ACFE, 32% of whistleblowers first report to their direct supervisor, and organisations with both a hotline and anti-fraud training record 50% lower losses2.
Two regulatory developments make this more pressing. For financial institutions, OJK regulation POJK 12/2024 requires an anti-fraud strategy that includes a fraud detection system16. For companies audited under international standards, the revised ISA 240 applies to periods beginning on or after 15 December 2026, with more emphasis on the completeness of the journal entry population17. External auditors will ask in more detail. Better to have the answers ready from the inside.
07Questions
Common questions.
What is occupational fraud?
Fraud committed by people inside an organisation, such as employees, managers, or owners, who misuse their position. ACFE groups it into three types: asset misappropriation, corruption, and financial statement fraud. It is different from a scam, where outsiders defraud the public or a company.
How long does fraud usually run before it is found?
According to ACFE's 2026 Report to the Nations, the median is 12 months. Fraud found through automated data monitoring runs 7 months, while fraud first revealed by law enforcement runs 24 months.
What are examples of fraud red flags?
The most common behavioural red flags: living beyond one's means, financial difficulties, and an unusually close relationship with a vendor or customer. Data red flags include a vendor bank account that matches an employee's, a recently changed vendor account, duplicate payments, and transactions created and approved by the same user.
What is the difference between continuous auditing and continuous monitoring?
According to IIA guidance (GTAG, third edition, 2025), continuous monitoring is a management process to make sure controls are working. Continuous auditing is internal audit's own work: repeated, automated testing of data and controls. The two complement each other, but they have different owners.
Does POJK 12/2024 apply to non-financial companies?
No. POJK 12/2024 applies to financial services institutions. But its framework of prevention, detection, investigation, and monitoring is sensible practice for any company.
We're a small company. Do we need to go this far?
Small companies are the least likely to have data monitoring: 17%, against 58% of large organisations. You don't need all ten points. Start with the three closest to money going out: vendors vs employees, changed vendor accounts, and duplicate payments.
08Closing
Closing.
In the run-up to International Anti-Corruption Day on 9 December, talk about corruption and fraud will pick up again. Most of it will be about large public-sector cases.
For companies, the practical question is simpler: if there were fraud in your organisation today, how long until you would know? Twelve months? Or seven?
The answer rarely depends on how clever the fraudster is.
More often it depends on how quickly we read our own data.
Useful? Share it with an internal audit, finance, or risk team preparing for year-end close.
This piece draws on primary reports (ACFE Report to the Nations 2026, IIA Risk in Focus 2026/2027 Asia Pacific, PPATK Catatan Capaian Strategis 2025), OJK press releases and regulations, and news coverage as of 13 November 2026. Global and Asia Pacific figures from ACFE were read directly from the report PDF. We don't quote an ACFE figure for Indonesia alone because it rests on only 20 cases. The last national fraud survey (2019) is not used, because secondary sources quote its details inconsistently.
The cases mentioned are still in legal process. Everyone involved is a suspect or subject to an administrative sanction, not convicted, and the descriptions follow OJK statements and media reports. Transaction values under investigation are not proven losses. Sanctions over affiliated-transaction procedures are administrative breaches, not fraud findings. Akses · aksesgroup.tech
References
- ACFE. Occupational Fraud 2026: A Report to the Nations, press release. 12 May 2026. acfe.com/about-the-acfe/newsroom-for-media/press-releases/…
- ACFE. Occupational Fraud 2026: A Report to the Nations (PDF). 2,402 cases, 143 countries. acfe.com/-/media/files/acfe/pdfs/rttn/2026/2026-report-to-the-nations.pdf
- The IIA. Risk in Focus 2026/2027: Asia Pacific. 2026. theiia.org/globalassets/site/foundation/research/2026-2027-apac-report-en-riskinfocus.pdf
- PPATK. Catatan Capaian Strategis 2025 (2025 strategic achievements). 28 January 2026. ppatk.go.id/backend/assets/uploads/20260129025534.pdf
- DDTCNews. IHPS I/2026: BPK highlights state revenue and energy. October 2026. news.ddtc.co.id/berita/nasional/1823049/…
- Bloomberg Technoz. BPK records potential state losses of Rp45.6 trillion in the first half of 2026. October 2026. bloombergtechnoz.com/detail-news/123733/…
- CNBC Indonesia. OJK: many rural banks closed because of fraud. 9 January 2026. cnbcindonesia.com/market/20260109141516-17-701044/…
- Harian Jogja. OJK revokes BPR Agritrans Batumarta licence, 16 banks closed in 2026. 4 November 2026. ekbis.harianjogja.com/r-1277521/…
- Bisnis.com. OJK completes banking crime investigation at PT BPR Panca Dana; three suspects handed to Depok prosecutors. 23 February 2026. finansial.bisnis.com/read/20260223/90/1955081/…
- CNBC Indonesia. Banking crime case at a rural bank in Malang. 8 November 2026. cnbcindonesia.com/market/20261108062038-17-783240/…
- OJK. Press release SP 210: OJK completes investigation of banking crime case at BPR DCN Malang. 2026. ojk.go.id/id/berita-dan-kegiatan/siaran-pers/…
- CNBC Indonesia. OJK sanctions 7 listed companies and related parties. 7 November 2026. cnbcindonesia.com/market/20261107163500-17-783204/…
- Tempo. 7 listed companies sanctioned by OJK. November 2026. tempo.co/ekonomi/7-emiten-kena-sanksi-oleh-ojk-2304017
- Kontan. IASC received 737,686 scam reports up to October 2026. keuangan.kontan.co.id/news/iasc-terima-737686-laporan-kasus-penipuan-hingga-oktober-2026
- Kontan. As of June 2026, IASC received 1,379 scam reports indicating the use of AI. keuangan.kontan.co.id/news/per-juni-2026-iasc-terima-1379-laporan-…
- OJK. POJK 12 of 2024 on anti-fraud strategy for financial services institutions. ojk.go.id/id/regulasi/Pages/Penerapan-Strategi-Anti-Fraud-Bagi-Lembaga-Jasa-Keuangan.aspx
- IAASB. IAASB revises fraud standard to enhance public trust (ISA 240 Revised). July 2025. iaasb.org/news-events/2025-07/iaasb-revises-fraud-standard-enhance-public-trust
- The IIA. GTAG: Continuous Auditing and Monitoring, third edition. 25 September 2025. theiia.org/en/content/guidance/recommended/supplemental/gtags/continuous-auditing-and-monitoring
- ISACA Journal. Reducing barriers to adoption of CCM through alarm management. 2023, vol. 5. isaca.org/resources/isaca-journal/issues/2023/volume-5/…
- UK Government. Annex 11: Data analytics tests. assets.publishing.service.gov.uk/…/Annex_11_-_Data_analytics_tests.pdf
All links accessed 13 November 2026.