Expense and Liability Manipulation

WorldCom's principal business cost was line costs: fees paid to local telephone companies and other carriers for access to their networks. These payments had the economic character of a rental or service fee, consumed in the period when the network was used. Under US GAAP, they should have been expensed in full as incurred.

Beginning in 2001, at the direction of CFO Scott Sullivan, WorldCom's accounting staff began transferring line-cost accruals from expense accounts to capital asset accounts. The justification offered internally was that these payments could be characterised as prepayments for future network access, and therefore had a future economic benefit that warranted capitalisation. This rationale was unsupported by the contracts and was never disclosed to auditors.

By the time the fraud was discovered by the internal audit team led by Cynthia Cooper in June 2002, the accumulated overstatement stood at $3.8 billion. The discovery was made without the cooperation of the external auditor, Arthur Andersen, which had audited WorldCom throughout the fraud period. Cooper's team found the capitalized entries by comparing general-ledger account codes against the descriptions of the underlying transactions.

Enron created over 3,000 special-purpose entities between 1993 and 2001. The ones that mattered for the fraud were a cluster of partnerships with names like LJM Cayman, LJM2 Co-Investment, and Raptor I through IV. These were controlled, despite appearances, by Enron's own CFO Andrew Fastow, who earned over $30 million in management fees from them.

The accounting rules at the time (FAS 125 and later FAS 140, now replaced by ASC 810) allowed an SPE to be kept off a sponsor's balance sheet if an independent third-party investor held at least 3% of the SPE's equity at risk. Fastow's partnerships were constructed to appear to meet this test. The 3% investors were often indirectly controlled by Enron itself or had their investment guaranteed by Enron stock, which meant the independence condition was never genuinely satisfied.

The practical result was that Enron could transfer poorly performing investments into these vehicles, recognise a gain on the transfer, and avoid consolidating the debt the vehicles then carried. When asset prices fell, the Raptor vehicles that had been used to hedge Enron's merchant investments became insolvent. Enron had guaranteed their debts, and those guarantees materialised all at once in 2001. The November 2001 restatement revealed $591 million in previously hidden losses and added $628 million to reported debt.

Provisions require judgment: a company must estimate the probability of an outflow and its likely amount. This judgment is genuine and necessary; it is also a door for manipulation. Setting a litigation reserve at the low end of the range when management knows the exposure is toward the high end, booking an asset impairment later than the evidence requires, or failing to accrue a probable warranty liability are all understatement techniques.

IAS 37 requires a provision when an obligation is probable (more likely than not) and the amount can be reliably estimated. GAAP (ASC 450) uses a similar threshold. The disclosure requirement for contingent liabilities that do not yet meet the recognition threshold is itself informative: a company with large disclosed contingencies but minimal provisions is indicating that management has set probability assessments just below the recognition threshold, which deserves scrutiny.

Cookie-jar reserves exploit the same provision rules as understatement fraud, but in reverse. Instead of setting provisions too low, the company sets them too high in a period when earnings are strong and the excess charge can be absorbed. The overstatement creates a balance-sheet liability that can be released into income in a later period when earnings miss the target.

The technique is subtle because both steps, creating and releasing the reserve, look like normal accounting adjustments. An auditor reviewing the release sees a reversal of a prior period provision. Without access to the original provision-setting documentation and the analytical basis for the estimate, the overstatement may never be questioned. The SEC has brought enforcement actions specifically targeting this technique, including against W.R. Grace in 1999 and Sunbeam in 2001.

Expense and liability manipulation leaves a different ratio footprint from revenue fraud. The primary signals are on the cash flow statement and the balance sheet rather than the top line.

• Capex / revenue ratio: rising abnormally against industry peers signals that operating costs may be being capitalised. For WorldCom, capex as a percentage of revenue was dramatically above industry norms in the fraud period.
• Operating cash flow / net income: capitalisation fraud shows as operating cash flow exceeding net income, because the capitalised amount reduces the income-statement charge but does not reduce operating cash flow.
• Depreciation rate: assets with implausibly long useful lives or a falling depreciation charge relative to asset base suggest inflated asset values or improper classification.
• Provision / revenue trend: provision balances that move independently of business volume or litigation outcomes warrant document-level examination.
• Debt-to-equity on a fully consolidated basis: for off-balance-sheet structures, computing gearing using guaranteed contingent obligations and disclosed SPE liabilities often reveals the true debt load.

Journal-entry testing focuses on entries that cross the expense-to-capital boundary: debits to fixed-asset or construction-in-progress accounts supported by descriptions that read as operating costs. In a digital accounting environment, these can be extracted by keyword-searching journal descriptions against a list of operating expense terms, then sorting by the account receiving the debit.