INTRODUCTION
It is estimated that almost 100 trillion dollars will be needed for traditionally high-risk infrastructure projects by 2040 [i], including more than a trillion dollars in US infrastructure improvements[ii] and trillions of dollars a year to combat climate change.[iii]
The World Bank and other development organizations have expressed particular concern that the massive scale, urgency and importance of the Climate projects, many of which will involve critical infrastructure works, will create “unprecedented” risks” of loss and project failures, with potentially “existential” consequences.[iv]
Traditional controls, such as occasional, brief supervision missions, and ex post audits, will be wholly inadequate; greatly enhanced and innovative oversight procedures will be necessary, such as the digital Automated Fraud Detection and Prevention Programs described below.
THE ADVANTAGES OF AUTOMATED FRAUD DETECTION PROGRAMS
The Programs exploit the trend to digitized record keeping systems to offer major advantages over traditional “paper” control systems, including greatly enhanced transparency, efficiency and economy, and, most importantly, the unique ability to:
-
- Detect and prevent potential wrongdoing ex ante, before fraudulent transactions are approved and losses incurred.[v]
- Conduct continuous, real-time remote monitoring and supervision by independent oversight personnel, rather than having to rely on oversight by project officials who may themselves be involved in corruption.[vi]
EXAMPLE OF A DIGITAL AUTOMATED FRAUD DETECTION PROGRAM
All project participants – contractors, subcontractors, suppliers, inspectors, etc. – would submit relevant data on procurement, construction and payment transactions electronically from laptop or cell phone applications to a dedicated project website; this would permit the creation of a digital record of all relevant project transactions, even in “low IT” environments.
The website would be equipped with cloud-based software that would apply sophisticated rule-based algorithms and AI programs to instantly analyze the data as it is received to:
-
- Identify indicators or “red flags” of fraud, corruption, waste and abuse
-
- Match the indicators to the related scheme or schemes, including:
Collusive bidding (Cartel activity)
Bid rigging
Kickbacks
Conflicts of interest
Shell companies
Phantom vendors
Failure to meet contract specs
Product substitution
False, inflated and duplicate invoices
-
- Look for other indicators of the suspect scheme(s)
- Extend the analysis as necessary to reduce the risk of “false positives”
- Score the results to reflect the level and certainty of risk
- Immediately report the results to oversight personnel for appropriate follow up.
Most of the fraud tests could be standard rule-based algorithms, many based on proven forensic accounting tests; other, more sophisticated tests could be developed and tailored to the specific locations and circumstances. Most of the tests could be run on readily available project data, avoiding the delays and disruption caused by the need to access multiple, disparate databases as in many other fraud detection programs.
More advanced AI programs could be run to detect previously unknown or well-hidden indicators or reduce the risk of false positives. For example, the Swiss Competition Authority used AI to examine voluminous prior tender data to identify previously unknown indicators and patterns of collusive bidding.
The digital data and AI analysis would be far quicker, more comprehensive and accurate than traditional document analysis, and could cover data sets of virtually unlimited size, rather than being limited to smaller sample sets.
EXAMPLES OF THE ADVANTAGES OF AUTOMATED PROGRAMS
As noted above, the digital Programs offer significant advantages over traditional “paper” anti-fraud controls, including most importantly:
-
-
- The ability to provide ex-ante fraud detection and prevention
-
In the PROCUREMENT stage, the Programs could identify indicators of potential bid rigging and collusive bidding schemes (cartel activity), such as unusual bid patterns, instantly as bids are received, before they are evaluated and contracts awarded.
In the CONSTRUCTION stage, the Programs could detect and block the attempted use of substandard construction materials by, among other means, instantly matching the relevant supplier’s sales and delivery records to the contract specifications. The detection of such abuses in traditional control systems typically would occur, if at all, in ex post audits, months or even years after the frauds were completed and the damages sustained.
In the PAYMENT stage, the Programs could block false and inflated payment applications by automatically matching the claims in the applications to the relevant time and labor records, supplier records and inspection reports. Again, the detection of such misconduct in traditional systems typically would occur only long after the transactions were completed.
-
-
- The ability to conduct continuous, real-time remote monitoring by independent oversight personnel
-
The Programs could alert donor or other oversight personnel, perhaps thousands of miles away, to potential fraud indicators, as described above, instantly as they arise, permitting them to immediately block or suspend the suspect transactions pending further inquiry. Such independent oversight is critical in high-risk environments where most corruption and fraud losses are caused by local officials in charge of project implementation and supervision.
The increased transparency and prospect of detection provided by digital systems also should help deter attempted misconduct in the first place.
-
-
- Immediate access to voluminous project files in ex-post audits and investigations
-
The algorithms also could be applied ex post in audits or investigations involving digital record keeping systems. This would permit virtually immediate access to 100% of voluminous project files, yet another exercise that could take months or longer in traditional paper systems.[vii]
SEE A LIST OF SAMPLE AUTOMATED FRAUD REPORTS FOR AN INFRASTRUCTURE PROJECT
Initially the oversight should be conducted by the donor or related government agency but ideally would be conducted by a new an entirely independent international facility, yet to be determined or established. [viii] This is important because donors and government agencies are often reluctant to acknowledge or address fraud and corruption in their operations because of institutional embarrassment and the fear of negative impact on funding.
OTHER POSSIBLE ANTI-FRAUD TECHNOLOGY THAT COULD BE EMPLOYED
Other automated and AI powered programs that could be applied include:
-
- Comprehensive background checking programs to identify fictitious or unqualified suppliers and subcontractors or evidence of corrupt officials
-
- Investigative steps and procedures, e.g., whistleblower reports could be automatically linked to relevant project data to quickly evaluate complaints and expedite follow up steps; AI could quickly and comprehensively identify and analyze relevant documents, etc.
-
- Fraud prevention and anti-corruption programs, such as compliance with ISO 37001
- Satellite-based analysis of construction defects in road projects in otherwise inaccessible areas
-
- Hyperspectral analysis tools to verify that construction materials meet required quality standards
- Drones and geospatial monitoring programs
ENDNOTES
[i] E.g., “Global infrastructure investment needs to reach USD97 trillion by 2040;” GitHub (World Bank); https://www.gihub.org/media/global-infrastructure-investment-need-to-reach-usd97-trillion-by 2040/#:~:text=The%20report%2C%20Global%20Infrastructure%20Outlook,Goals%20(SDGs)%20for%20universal%20household.
[ii] $1.2 trillion US Infrastructure Investment and Jobs Act (IIJA); https://www.congress.gov/bill/117th-congress/house-bill/3684; https://www.brookings.edu/articles/fighting-fraud-waste-and-abuse-the-infrastructure-bill-and-lessons-for-the-future
[iii] “[C}limate finance needs for developing countries and emerging markets are put at 1 trillion dollars per year and rising;”https://baselgovernance.org/blog/strengthen-alliances-counter-environmental-corruption-new-practitioners-forum-transparency; Developed countries at COP 29 committed “$300 billion a year until 2035 to fight climate change, less than a quarter of the acknowledged $1.3 trillion needed annually to reduce emissions and build resilience in vulnerable countries;” https://www.voanews.com/a/nations-at-un-climate-talks-agree-on-300b-a-year-for-poor-countries-in-a-compromise-deal/7874947.html
[iv] “The risk of corruption in climate change financing is significant because massive investments need to be deployed on public infrastructure, an economic activity that—as discussed below—has traditionally been plagued by large-scale bribery and theft” https://www.piie.com/publications/policy-briefs/corruption-risks-loom-large-over-financing-green-infrastructure.
The risk of corruption in climate change financing is significant because massive investments need to be deployed on public infrastructure, an economic activity that—as discussed below—has traditionally been plagued by large-scale bribery and theft” https://www.piie.com/publications/policy-briefs/corruption-risks-loom-large-over-financing-green-infrastructure.
“The largest recipients of climate-related overseas development assistance are notorious for having significant systemic corruption, i.e. India, Bangladesh and Indonesia; The top recipients of climate finance rank high in corruption and are scheduled to receive almost 42% of all climate related overseas development assistance” https://www.u4.no/publications/corruption-and-climate-finance.pdf
Largest “upscale” of infrastructure financing into “low governance” countries in history, equal to four times annual GDP; “extremely high risk” of fraud and corruption with potentially “existential” consequences. https://www.worldbank.org/en/events/2023/06/27/deep-dive-sessions-the-green-transition-and-anticorruption.
“The $1.2 trillion US IIJA is not immune from fraud concerns. Despite the widely acknowledge fraud risks…IGs say inadequate attention paid to fraud: ‘the word fraud appeared only seven times in the 2000-page bill;’” https://www.brookings.edu/articles/fighting-fraud-waste-and-abuse-the-infrastructure-bill-and-lessons-for-the-future
[v] In addition to it’s obvious advantages, the World Bank has warned that ex ante fraud detection and prevention will be critical in climate projects because of the need for immediate remedial action, which does not permit the delays inherent in traditional ex post responses to fraud. https://www.worldbank.org/en/events/2023/06/27/deep-dive-sessions-the-green-transition-and-anticorruption
[vi] Many international development agencies delegate almost all project supervision and reporting to local project officials and oversight agencies, often the same officials and agencies involved in corrupt practices. Local inspectors and oversight officials also may be corrupt or pressured to ignore fraudulent and corrupt practices; even international civil engineering firms, serving as supervision consultants, may be removed if they fail to approve fraudulent works.
[vii] Section 5113 of the US Infrastructure Investment and Jobs Act (IIJA) provides for $100 million in grants to expedite the digitization of construction records to obtain its “business benefits.”
[viii] See, for example, the Pandemic Response Accountability Committee (PRAC), an independent group of U.S Inspectors General created to oversee and prevent fraud in COVID-19 spending.