How Autonomy Fooled Hewlett-Packard
In 2011, Hewlett-Packard paid $11.1 billion for UK software firm Autonomy – a 64% premium for a company with nearly $1 billion of 2010 revenues, and possessing “a consistent track record of double-digit revenue growth, with 87 percent gross margins and 43 percent operating margins.” So they thought. Little more than a year later, HP recorded an $8.8 billion impairment charge, citing Autonomy’s accounting improprieties as the reason. Investors wondered how could HP have gotten it so wrong before they plunked down $11.1 billion in cash? And they wondered what HP uncovered that led to the writedown.
Now we know. In mid-November, the SEC ordered the former CEO of Autonomy’s U.S. operations, Christopher Egan, to fork over $800,000 of compensation resulting from the takeover, in which HP relied on figures he had helped inflate. The facts of the case are now public. Although this case related to current IFRS revenue recognition rules, it can happen again, and to any company.
One fact really stands out: in each of the 10 quarters preceding the acquisition, Autonomy’s revenues were within 4% of analyst expectations. That’s a level of precision that should arouse suspicion. In hindsight, achieving revenue targets like clockwork looks awfully strange. Here’s how they did it.
Autonomy’s UK-based senior managers directed a program swelling revenues by almost $200 million. Autonomy sold its software through “value-added” resellers, legitimate businesses providing additional services and support to product end users while also selling Autonomy’s software. Just five resellers, in 30 transactions, provided services to Autonomy that couldn’t be called legitimate.
When Autonomy was negotiating a sale to an end user, but couldn’t close the sale by quarter’s end, Egan would approach the resellers on or near the last day of the quarter, saying the deal was nearly done. Egan coaxed the resellers to buy Autonomy software by paying them hefty commissions. The resellers could then sell the software to a specified end user – but Autonomy maintained control of the deals and handled negotiations with the end user without the resellers’ aid. There’s no way these transactions could be revenue.
Autonomy retained risks, rewards and ownership of the goods – not the resellers, and not the end users. Autonomy was still exercising continuing involvement to an abnormal degree for a real transfer of ownership to occur. And the benefits of the deals didn’t accrue to Autonomy until they were sold to an end user. These transactions “grew” Autonomy’s revenues by as much as 15% in some periods. They were critical: they enabled the firm to report financial results within the boundaries of analyst expectations.
Between 2009 and 2011, after a quarter’s close, Autonomy’s most senior finance executive directed Egan to procure backdated purchase orders from resellers; once, the executive obtained the dirty documentation personally. This resulted in revenue “pulled” from a later quarter into the one just closed – sometimes just enough revenue to let Autonomy hit its revenue target. In anyone’s book, backdating purchase orders is a falsification of facts. Viewed through an IFRS accounting lens, these transactions couldn’t be called revenues because Autonomy didn’t transfer the risk and rewards of software ownership to the buyer, at quarter end. That happened in a later quarter.
Autonomy needed to get cash to the resellers so they could pay Autonomy for the sham sales, creating a paper trail for the auditors demonstrating payments on the sales – a necessary optic, to avoid arousing their suspicions. The illusion was created with round-trip reseller transactions. Autonomy purchased various surplus products from the resellers, with nearly simultaneous payment by the reseller back to Autonomy on debt owed to them. The round-trip transactions were improvised by Autonomy’s senior-most finance executive, and they amounted to at least $45 million of phony paybacks to Autonomy.
On the orders of the most senior financial executive, Egan was involved in some round-trip transactions: he served as the conduit between Autonomy and the resellers, and knew, or should have known, that they didn’t represent genuine purchases by Autonomy. He knew Autonomy did not price the cost of such purchases with other vendors of the same products; he didn’t negotiate agreement terms.
What to make of these facts?
There’s more than one way to skin a cat
Often, the most senior financial executive of Autonomy developed most of the illicit transactions; and being out the country, was not directly within the SEC’s reach. Egan was more of a pawn than a mastermind, and carried out his boss’s plans – inside the United States, where at least one gigantic investor relied on the doctored financial statements. Egan was reachable by the SEC, and he provided them with assistance in ferreting out the falsifications.
Would the sham transactions have been uncovered without HP buying Autonomy – and without Egan’s assistance? It probably wouldn’t have been uncovered. The bogus transactions were designed to look real and throw suspicious parties off the trail. Backdated documents don’t look different from ones that aren’t backdated; the cash paid from the round-trip transactions was arranged to make dummy receivables look real. That would likely satisfy auditors, if their suspicions were unaroused otherwise.
Should the Autonomy auditors have caught the fake transactions? Possibly. Had they investigated the reseller transactions to see if the initial sales to resellers resulted in sales to end users – not all of them did, remember – they might have probed further. There were only 30 such transactions over a 10-quarter period: the auditors would have been awfully lucky to find transactions that piqued their curiosity out of such a small population and over such a long time frame. If odd transactions always showed in the last few days leading up to the quarter end, however, they should have noticed – and dug further.
This occurred under IFRS rules. Could it have happened under GAAP? No doubt. The problem here isn’t with the standards: the problem is the intent of the players involved.
Accounting standards can’t prevent the creation of false documents or backdated purchase orders. No set of rules, not even the new revenue recognition standard, can prevent mischief by managers expected to play by the rules.
Jack T. Ciesielski is president of R.G. Associates, Inc., an asset management and research firm in Baltimore that publishes The Analyst’s Accounting Observer, a research service for institutional investors. Neither Hewlett Packard nor Autonomy are clients of R.G. Associates Inc.