Castle Hall Blog

Can AI Detect Fraud? Case 1: Woodford and Audited Financial Statement Reviews

Written by Chris Addy | 6/3/24 4:15 PM

Can AI find investment industry fraud? Well the short answer is of course no - AI is not (yet at least!) an omnipotent superpower able to uncover different sources of information or interpret data more effectively than an experienced due diligence analyst.

But, what AI can do is consistently review diligence information to screen different managers and funds. Then, AI can reliably highlight risks within an organised framework, provided it is given a detailed and well structured prompt (which must itself be written by an experienced ODD analyst).

So, in other words, AI is not infallible and it is not a genius - but it can review hundreds of managers and funds in very short order and flag issues which can be escalated and reviewed. And, AI for ODD seems to be surprisingly good - let's give it a mark of 8/10 overall.

To show some of Castle Hall's results, I'd like to give three examples of fund frauds and see what DiligenceAI delivers as a component of the overall ODD process, starting with an example of a review of a fund's audited financial statements.

Some of you may be familiar with the blow up of the Woodford Equity Income Fund - which eventually resulted in a GBP230 million settlement scheme.

The fund was in essence a UK mutual fund and hence its financial statements were publicly available. What happened here was an increasing concentration in illiquid assets while the fund faced rapid redemption requests as the star power of the firm's high profile manager, Neil Woodford, faded.

Castle Hall has taken those accounts and asked DiligenceAI to apply our standard audited financial review framework to the accounts. For every AFS review, our platform reviews the auditor and audit opinion; the balance sheet; the portfolio; the income statement; and footnote disclosures. The output is a consistent and well structured Castle Hall AFS review report which allows year on year comparison: this was input into the AI.

The goal of AI is not necessarily to uncover a smoking gun, but AI can be activated to review all, say, 186 funds in an institutional investor's portfolio to see which, if any, are identified to have "yellow" or "red" risk characteristics. And, in a day AI can deliver results that would take the analyst team several weeks (months?!)

This is what the AI reported:

DiligenceAI Review of LF Woodford Equity Income Fund

Based on the analysis of the LF Woodford Equity Income Fund's audited financial statements for the years ended December 31, 2018, 2017, and 2016, I would assign an overall rating of RED to this review.

Several material and significant issues have been identified that suggest urgent escalation and further operational due diligence:

Substantial decline in assets:

The fund's net assets have decreased by more than 50% over two years, from £9.5 billion in 2016 to £4.7 billion in 2018. This significant decline raises concerns about the fund's performance, sustainability, and the manager's ability to continue operations effectively.

High redemption outflows:

In 2018, redemptions amounted to £3.0 billion, representing 37% of the fund's opening net assets. This exceeds the 25% risk threshold and indicates potential liquidity risk. The fund's ability to meet redemption requests going forward should be closely examined.Increase in hard-to-value assets:

The proportion of Level 3 assets, which are inherently difficult to value, has doubled from 10% of NAV in 2017 to 20% in 2018. This shift elevates valuation risk and potential illiquidity concerns, warranting further investigation into the fund's valuation procedures and the appropriateness of these investments.
Significant net loss: The fund generated a substantial net loss of £1.3 billion in 2018, compared to smaller losses in 2017 and a modest profit in 2016. Persistent losses may strain the manager's ability to continue operations effectively.

Rising expense ratio:

The fund's expense ratio has steadily increased from 0.01 bps in 2016 to 0.13 bps in 2018, which may indicate additional costs being charged to the fund. This should be scrutinized to ensure that investors' interests are protected.

Subsequent events and illiquid asset movement:

The footnotes reveal that subsequent to the 2018 year-end, the fund has begun a strategy to shift its unquoted exposure from individual unquoted holdings into shares of Woodford Patient Capital Trust. In March 2019, the fund acquired over 81 million shares in Woodford Patient Capital Trust, with a significant portion of the transaction involving the exchange of a portfolio of unquoted assets. This move raises further concerns about the liquidity and valuation of the fund's holdings, as it appears to be concentrating its illiquid positions into another Woodford-managed entity. The appropriateness and motivations behind this strategy should be thoroughly investigated.

While the presence of a non-Big 4 auditor and the concentrated ownership by Hargreaves Lansdown are not necessarily red flags in isolation, they do merit further examination in light of the other risk factors identified.

Conclusion

Given the significant issues noted in the audited financial statements, particularly the substantial decline in assets, high redemption outflows, increased exposure to hard-to-value investments, and the subsequent transfer of illiquid holdings to Woodford Patient Capital Trust, I strongly recommend urgent escalation and a comprehensive operational due diligence review. This review should focus on protecting investors' interests, thoroughly assessing the fund's viability, and examining the appropriateness of the recent strategic shift in light of the fund's liquidity challenges.

Access our White Paper on how to integrate AI into the operational due diligence process here.