HP Thinks That Autonomy Extracted A Bit Too Much Meaning From Its Enterprise Data

Author:
Updated:
Original:

I don't know much about this Autonomy thing - in brief, Hewlett-Packard acquired British software company Autonomy last year for $10.3 billion and today wrote that investment down by $8.8 billion, blaming $5 billion of that on "accounting improprieties, misrepresentations and disclosure failures" at Autonomy - but this sure sounds fake doesn't it?

Today, Autonomy is firmly established as the leading provider of Pan-Enterprise Search and Meaning Based Computing (MBC) solutions. Autonomy’s unique Intelligent Data Operating Layer (IDOL) platform enables organisations to harness the full richness of human information by extracting meaning from the mass of unstructured information they handle every day, which analysts estimate to constitute over 80% of all enterprise data.

That's from Autonomy's last annual report as an autonomous company1 before HP bought it. Retrospective red flag perhaps? I would be wary of companies whose business involves "extracting meaning." Meaning doesn't come from software.2

Now of course HP is going to sue everyone and demand fraud investigations on two continents. Many people look bad here - HP first of all, whether or not its claims are true, then (if they're true) Autonomy, Deloitte (Autonomy's auditors), E&Y (HP's auditors), HP's bevy of bankers and others involved in due diligence who seem to have been unduly undiligent, and to some extent Autonomy's bankers who marketed it to HP.3 I have plenty of sympathies with both sets of bankers, of course; their job is mainly to harness the full richness of human information by extracting meaning from the mass of stuff that companies make public, not to know whether that stuff is true.4 Bankers are an intelligent data operating layer; if you give them bad data then their operating layer is less intelligent. It's possible that some of those words mean things.

That excuse sounds sort of lame - lots of people thought Autonomy was doing shady stuff! - but here we are. The bankers aren't supposed to be able to spot well-concealed accounting problems; if they could, they'd be doing so. Like Jim Chanos noticed accounting imperfections at Autonomy but then he's paid to do so. More than the bankers are I mean.

Let's just assume, because it's more fun than the alternative, that Autonomy was in fact a massive accounting fraud.5 The way that normally happens is, you're an okay business, but you miss a few targets here and there and you don't want to lose investor confidence so you fudge the numbers just a bit, booking revenue too early and so forth, planning to pay yourself back later, but then of course when you miss the next target things are even worse because you owe yourself for the prior quarter, and things snowball, and one day you wake up and your reality is $5 billion behind your accounting and it seems unlikely you'll ever dig yourself out.

What do you do? Autonomy's answer seems to have been "shop the hell out of yourself" - to HP, certainly, and maybe also to Oracle6 - so as to be acquired. In some ways, this is obvious: getting bought out for cash gets you, y'know, cash, at a significant premium in Autonomy's case, which allows you to monetize your fraud and take off for non-extradition countries if that's your thing. (Note that it wasn't particularly at Autonomy; its CEO took his $800 million on the deal but stayed around to run the place until HP fired him.)

But it's not entirely obvious. Being a public company subjects you to a certain amount of scrutiny, from auditors and investors and whatnot, but Autonomy seems to have managed fooling its auditors just fine. And putting yourself on the block exposes you to due diligence: a prospective buyer doesn't need to be satisfied with public information, but can kick the tires more thoroughly, demanding nonpublic information and doing its own review of your accounts. There's a chance they'll discover any fraud. And if people are worrying about your aggressive accounting anyway, and a potential acquirer publicly walks away from a deal (as Oracle kind of did!), that's not going to help your quest to avoid detection.

But Autonomy did it anyway. Maybe because there's no fraud and HP is mistaken? Meh. Or maybe because it figured the consequences of being caught, as an HP subsidiary, would be less bad than the consequences of being caught as a public-company fraud? After all, no Autonomy shareholders were directly harmed by management's fraud: they all got bought out at a premium! (HP's shareholders, on the other hand, are losing out, but they have a bevy of other failures to shield them from the consequences of the Autonomy failure. Who knows what part of Autonomy's performance decline, post-acquisition, was due to HP's management and what part was due to the fact that pre-acquisition performance was fictional?) And when HP goes and sues everyone at Autonomy for fraud, Autonomy can - semi-reasonably! - say "well, HP had every opportunity to do due diligence, why were they relying on what we told them?" Public shareholders just seem more sympathetic, more able to prove reasonable reliance on Autonomy's accounts, more clearly damaged by that reliance, and more likely to interest the sorts of people who put executives in jail, than HP does. HP garners much less sympathy.

Also, of course, shopping yourself is only dangerous if due diligence is likely to actually discover fraud. That didn't happen and I guess there's no reason why it would - again, the M&A process is about extracting meaning from information, not checking its accuracy. Perhaps Autonomy decided that being a public company exposed them to motivated short sellers who would figure out their accounting problems much faster than HP and its bankers and auditors would.

HP says discovered accounting "failures" at Autonomy unit [Reuters]
HP Issues Statement Regarding Autonomy Impairment Charge [HP]

1.Ugh, I know, sorry.

2.That's why I didn't buy the Facebook IPO.

3.I guess I look bad too; I did my own forensic accounting last year to guess that Autonomy's version of its silly spat with Oracle sounded more plausible than Oracle's. I have updated my priors.

4.The internet today is full of people defending their professions. I think if you asked the average

  • person, they'd say "why didn't the bankers and lawyers doing due diligence catch this?";
  • banker or lawyer they'd say "we rely on audited financials - why didn't the auditors catch this?";
  • auditor they'd say "we just audit the work of internal accountants - why did they lie to us?"

This is a problem of the M&A professions. M&A bankers have skills, but they are not the sort of skills that can tell you whether you should buy a company or how much you should pay for it. They just look like they are.

Also! Remember this story, where the shareholders of Dragon, the speech-recognition software, are suing Goldman because Goldman let them sell themselves to a massive fraud in an all-stock deal without catching that problem in diligence, and it turned out that the Goldman deal team was led by some VP who was perpetually on vacation? I bet he feels better, anyway.

5.Let's also assume that it came from the top; HP's press release blames "some former members of Autonomy’s management team [who] used accounting improprieties, misrepresentations and disclosure failures to inflate the underlying financial metrics of the company" but it's not clear if that means the CEO or others who had a say in shopping the company. Maybe it was some sales manager whose fake reports were just reported up the chain in good faith.

6.See footnote 3, supra.

Related