Monday, November 26, 2012

HP's Deal from Hell: The mark-it-up and write-it-down two-step

I don't think that there can be any disagreement that Hewlett Packard (HP) had a terrible day on November 20. In a surprise announcement, the company announced that it was taking a write off of $8.8 billion of the $11.1 billion that it paid to acquire Autonomy, a UK based technology company, in October 2011, and that a large portion of this write off ($ 5 billion) could be attributed to accounting improprieties at Autonomy. Even by the standards of acquisition mistakes, which tend to be costly to acquiring company stockholders, this one stood out on three dimensions:
  1. It was disproportionately large: While there have been larger write offs of acquisition mistakes , this one stands out because it amounts to approximately 80% of the original price paid. 
  2. The preponderance of the write off was attributed to accounting manipulation: Most acquisition write offs are attributed either to over optimistic forecasts at the time (the investment banker made us do it..) of the merger or changes in operations/markets after the acquisition (it was not our fault). HP's claim is that the bulk of the write off ($5 billion of the 8.8 billion) was due to accounting improprieties (a polite word for fraud) at Autonomy.
  3. The market was surprised: Most acquisition write offs, which take the form of impairments of goodwill, are non-news because they lag the market and have no cash flow effects. In other words, by the time accountants get around to admitting a mistake from an acquisition, markets have already admitted the mistake and moved on. In HP's case, the market was surprised and HP's stock price dropped about $ 3 billion (12%) on the announcement. Put differently, the market had priced in an acquisition mistake of $5.8 billion into the value already and was surprised by the difference.
The blame game
I am sure that this case will be examined and reexamined over time in books like this one, but at this moment, every one involved in the merger is blaming someone else for the fiasco. So, here is a roundup of the suspects:
  • Meg Whitman, the current CEO of HP, blamed the prior top management at the company, and said that "(t)he two people that should have been held responsible are gone ".
  • Leo Apotheker, the prior CEO who orchestrated the acquisition, claimed to be shocked at the "accounting improprieties" at Autonomy. 
  • Michael Lynch, the founder of Autonomy, said that two major auditors had performed "due diligence" on the financial statements and had found no improprieties at the company. 
  • Deloitte LLP, the auditor for Autonomy, denied all knowledge of accounting misrepresentations and claimed to be cooperating with authorities. 
  • The advisers on the deal (Perella Weinberg & Barclay's Capital for HP, Quatalyst, UBS, Goldman Sachs, Chase & BofA for Autonomy) have all been mysteriously silent, though none have offered a refund of their advisory fees. 
So, who is telling the truth here and who is to blame? Perhaps, the only way to answer this question is to go back to the original deal, which occurred just over a year ago.

Building up to an acquisition price: The original deal
Before we look at the numbers, it is worth reviewing the history of the two companies involved. Autonomy was a company founded at the start of the technology boom in 1996, which soared and crashed with that boom and then reinvented itself as a business/enterprise technology company that grew through acquisitions between 2001 and 2010. Hewlett Packard, with a long and glorious history as a pioneer in computers/technology, had fallen on lean times as it's PC business became less competitive/profitable and due to top management missteps.

On August 18, 2011, HP's then CEO, Leo Apotheker (who had worked at SAP) announced his intent to get out of the PC business and expand the enterprise technology business by buying Autonomy. While the deal making began on his watch, the actual deal was officially completed on October 3, 2011, with Meg Whitman as CEO. If she was a reluctant participant in the deal, it was not obvious in the statement she released at the time where she said that "(t)he exploding growth of unstructured and structured data and unlocking its value is the single largest opportunity for consumers, businesses and governments. Autonomy significantly increases our capabilities to manage and extract meaning from that data to drive insight, foresight and better decision making."

One of the perils of assessing "big" merger deals is that the fog of deal making, composed of hyperbole, buzzwords and general uncertainty, obscures the facts. So, let me stick with the  facts that were available at the time the deal was done (a time period that stretched from August 18, 2011, to October 3, 2011):
  1. Acquisition Price: While there have been varying numbers reported about what HP paid for Autonomy, partly reflecting when the story was written (between August & November) and partly because of exchange rate movements (HP paid £25.50/share), the actual cost of the deal was $11.1 billion. 
  2. Market Price prior: Autonomy's market cap a few days prior to the deal being announced was approximately $5.9 billion.
  3. Pre-deal accounting book value: The book value of Autonomy's equity, prior to the deal, was estimated to be $2.1 billion. (Source: Autonomy's balance sheet from its annual report for 2010)
  4. Post-deal accounting book value: After acquisitions, accountants are given a limited mission of reappraising the value of existing assets and this appraisal led to an adjusted book value of $ 4.6 billion for Autonomy. (Source: HP's 2011 annual report, page 99)
The advantage of working with these numbers is that differences between them are revealing. In the figure below, I attempted to deconstruct  the $11.1 billion paid by HP into its constituent parts:

You can see the build up to the price paid by HP as a series of premiums:
  1. The accounting "write up" premium for book value: One of the residual effects of the changes that have been made to acquisition accounting is that accountants are allowed to reassess the value of a target company's existing assets to reflect their "fair" value. For technology companies such as Autonomy, this becomes an exercise in putting values to technology patents and other intangible assets and that exercise added $2,533 million to the original book value of equity.
  2. The pre-deal "market" premium over book value ($1.3 billion over post-deal book value): Even if accountants write up the value of assets in place to fair value, markets may still attach a premium for growth potential and future investments. As with any market number, this number can be wrong, too high for some companies and too low for others. Prior to the HP deal, the market was attaching a value of $6.2 billion to Autonomy, $3,833 million higher than the original book value of equity and $1.3 billion more than the post-deal accounting book value of equity. 
  3. The acquisition premium ($5.2 billion): To justify this premium, HP would have to had to believe that one or more of the following held:  (i) the market was undervaluing Autonomy, i.e., that the true value of Autonomy was much higher than the $ 5.9 billion, (ii) there are synergies between HP and Autonomy that have value, i.e., that there are value-enhancing actions that the combined firm (HP+Autonomy) can take that could not have been taken by the firms independently and/or (iii) that Autonomy was badly run and that changing the way it was run could make it more valuable, i.e., there is a control premium. Even without the benefit of hindsight, neither undervaluation nor the control premium seemed to fit as motives in this acquisition. First, Autonomy was being priced by the market richly in August 2011; the market cap of $ 5.9 billion was roughly 6 times revenues and 15 times earnings and neither number looked like a bargain. Second, for a company that had been as badly run as HP to be talking about inefficiencies at other companies (and control premiums) strikes me as absurd.
The reaction to the deal was negative, a the time that it was done. The analysts and experts were generally down on the deal, but more importantly, the markets voted against the deal by pushing down HP's stock price. Between August 18, 2011 (the date the deal was announced) and October 3, 2011 (when the deal was consummated), HP's market cap plummeted by $15 billion from $58.5 to $43.5 billion. It would be unfair to attribute this meltdown to the Autonomy deal alone, since HP was announcing spectacular failures on so many different fronts, but it would be fair to say that markets did not share HP's hopeful assessments of synergy in this deal.

The cost of accounting impropriety & breaking down blame
Let's fast forward to today. In the conference call on November 18, the CFO of HP attributed the write off of $8.8 billion to two primary sources: $5 billion to accounting improprieties at Autonomy and $3.8 billion to a drop in HP's stock price. The latter rationale does not really hold up since it mistakes cause and effect; the stock price went down because of HP's misstep (though it has made so many that I am not sure which one) and is not the cause of the write off. Thus, it makes sense to attribute the entire write off to the deal. Applying HP's write off of $8.8 billion to the acquisition price of $11.1 billion, brings Autonomy's estimated value (according to HP) back down to $2.3 billion (almost equal to the pre-deal book value of $2.1 billion). In effect, HP is arguing that almost all of the premiums in the original deal (the accounting write up, the pre-deal market premium, the acquisition premium) were not justified.

So, what form did these accounting improprieties take? Based on news reports, HP's contention is that Autonomy was "overstating" and "mis-categorizing" revenues (they were allegedly booking low margin hardware sales as higher margin service/software sales). Assume for the moment that HP is right, and that Autonomy's revenues in 2010 were overstated by 15% and that its true operating margin was the industry average of 31% (and not the 36% that Autonomy was reporting in 2010). Since the accounting misstatements predated the HP acquisition, the pre-deal market value of $5.9 billion should already have been inflated because of the misstatements. Using the pre-deal market value of $5.9 billion as a base, I extracted an expected revenue growth rate of 14.25%. I then substituted in the lower revenues (15% drop) and lower margin (31%) into the valuation and estimated a value for the equity of $4.2 billion. Put differently,  if you buy into HP's story fully, the value effect of the accounting misstatement was $1.7 billion (the difference between the pre-deal market value and the adjusted value)  on the pre-deal value.

HP's argument would be that the synergy premium of $5.2 billion that they paid was also overstated because of the accounting improprieties. Since we do not have access to the detailed synergy estimates (assuming that they were made), we assumed that the accounting overstatement would comprise the same percent of the synergy value it was of the pre-deal market value ($1.7 billion is 28.81% of $5.9 billion) and that half of this value is Autonomy's share (since synergy accrues to the combined firm):
Estimated accounting impropriety portion of synergy value =0.5( 28.81% of $5,200) = $749 million
I know that this may not be fair and that I am going with incomplete information, but here is my breakdown of blame for the $ 8.8 billion write down:

As I see it, everyone involved in this process owns part of this disaster. Leo Apotheker, the CEO who pushed this deal through, gets the lion's share with $4,451 million, though the investment bankers who advised him were his enablers in the process. To the extent that HP is right in its contention that Autonomy cooked the books (inflating revenues and margins), Autonomy's founder/ managers and its auditor (Deloitte) are responsible for $2,449 million in value destruction. That leaves us still with an additional $1,900 million in write offs, which I can attribute to either a deterioration of Autonomy's business in the eleven months since HP took it over (a form of reverse synergy) or game playing on the part of HP, where taking bigger losses now will allow them to claim improvements and look better in the future. In either case, I would hold HP's current management responsible for that portion ($1,900 million) damage. As HP stockholders, though, don't expect any of these parties to offer to cover their fair share.

This deal offers important lessons about headstrong CEOs, the ineffectual accounting for acquisitions and the flaws in the acquisition process that allow bad deals like this one to get through, but this post has become too long to expand on these issues. So, more in my next few posts....

The Acquisition Series
HP's deal from hell: The mark-it-up and write-it-down two step
Acquisition Archives: Winners and Losers
Acquisition Hubris: Over confident CEOs and Compliant Boards
Acquisition Advice: Big deal or good deal?
Acquisition Accounting I: Accretive (Dilutive) Deals can be bad (good) deals
Acquisition Accounting II: Goodwill, more plug than asset

Monday, November 19, 2012

Much ado about liquidity? Lockup expirations and stock prices

Last week shaped up as a big one for Facebook. On Wednesday (November 14), the company faced the steepest of its lockup expiration cliffs so far, with 777 million shares released for sale by insiders. Its two earlier lockup expirations, of 271 million shares on August 16 and 234 million shares on October 29, did cause stock price pullbacks of about 5% and 3% respectively. Consequently, there was concern that Facebook’s stock price would take a beating on November 14, but the stock price climbed 12.6% on that day.

There are a host on intriguing questions that derive from lockups, their expiration and the market reaction to them, and I think it is worth taking a look at them.

The mechanics of and rationale for a 'lockup"
            Facebook is not unique. Most initial public offerings (IPOs) have lockup agreements that prevent insiders (which include owners/founders and VC investors) from selling their shares for a period after the offering. These lockups are not mandated by regulatory authorities but are contractual agreements between issuers and underwriters, with the terms disclosed in the IPO prospectus. While the most common lockup period for IPOs in the US is 180 days, there are quite a few firms that stagger their lockup dates (as Facebook did).  So, why do we see lockup periods in initial public offering? There are at least three reasons for the practice.
  1. Skin in the game:  There are many risks that investors face when investing in newly public companies, but one is that the investors and founders of the company will cash out and leave behind a vacuum. Lockup periods ensure that the venture capitalists and owners of a business have skin in the game at least for a limited period after a stock goes public.
  2. Signal of company quality: As a related point, having a lockup period makes the offering price more credible for outside investors, since insiders have to wait to sell their shares (rather than dump them at the offering price). A study of British IPOs found that longer lockup periods are associated with better performing IPOs (both in terms of profitability and stock price performance). In fact, it is worth remembering that Facebook, in its pre-offering hubris cut the lockup period to three months for some holders just before the offering date.
  3. Stage management of offering: Having a lockup period for insiders also ensures that only a small fraction of the outstanding shares  hit the market on the offering date. This, in turn, enables investment bankers to "discount the price" at the offering and allows them to use the initial offering more as a marketing event leading up to the main event (which is the sale of shares when the lockup period expires). As this study notes, there is evidence that investment bankers and insiders both have an interest in underpricing the offering shares to create price momentum, which can then be ridden (hopefully) to the end of the lockup period. 
Lockup expirations: The insiders' choice to sell (or not)
When a lockup expires, insiders get the right to sell their shares, though they can choose not to sell. While there may be other motives at play, there are three reasons why insiders may sell at the end of the lockup period:
  1. Need for cash: Some insiders, while wealthy in terms of the market value of their holdings, a can still be cash poor if the bulk of their wealth is tied up in the shares of the company. If they need the cash (to either fund conspicuous consumption or to pay taxes), they may have to liquidate at least some of their holdings.
  2. Diversification : In a related point, the founders and even some of the venture capitalists in a company that has just gone public may find that they have too much of their wealth tied up on that company. Having weighed the desire for control of having a concentrated position against the peace of mind that comes from diversification, some of them may choose to cash out on at least a portion of their holdings and invest that cash elsewhere.
  3. Information: Perhaps the trickiest part of the equation is that insiders do have access to information that the rest of the market does not about how a company's operations are performing. If they feel that the market price is too high (relative to their judgment of value), they will be inclined to sell their holdings.
The rest of us get to observe the actions (whether insiders sell at the end of the lockup period and how much they sell) but not the motives. Not surprisingly, we still try to find signals in the actions and react to how much insider selling there is, relative to our expectations.

Lockup expirations: The evidence and analysis
Lockup periods are therefore par for the course in initial public offerings and insiders have multiple motives for selling when lockup periods end. So, what typically happens when lockup periods end? To understand the market reaction when lockup periods expire, let's look at the effects, both positive and negative, of these events:
  1. Liquidity effect: In the short term, the end of the lockup period releases shares for sale into the market, creating a demand/supply story that goes as follows: the end of the lockup releases news shares into supply, and holding the demand for these shares constant, this should reduce price. In the longer term, the release of the “locked up” shares to the market increases the shares that are available for trading (the float) and may should improve liquidity. 
  2. Information effect: When insiders exercise their right to sell their shares, they are also conveying their views on what they think about the market price. As we noted in the last section, you are more likely to see heavy insider selling, if insiders view the stock to be over valued and less if it is under valued. In particular, markets form expectations about how much insider selling you should observe and if there is less (more) insider selling than anticipated, it is viewed as good (bad) news. 
  3. Backstop effect: While investment banks may be under no legal obligation to provide support services beyond the immediate IPO, there is evidence that issuing banks continue to provide at least partial support for an offering until the lockup date. That support can range from buying shares, if the stock price goes into free fall, to favorable recommendations from the issuing banks’ equity research analysts. The removal of the investment banking support system may have negative consequences for stocks. 
Looking at the trade off, the net effect should vary then across companies. You would expect the most negative price impact from lockups ending at small lightly-traded firms (where the near term trading imbalance can overwhelm the long term liquidity benefits), where there are few institutional investors or analysts tracking the firm (making insider trading that much more informative) and where issuing banks have been active in providing support (ensuring that the removal of the backstop will have more consequences.  The effect should be more muted with larger firms that are already actively traded by institutions and tracked by analysts. Since these firms are already heavily traded, the liquidity impact is likely to be smaller, the institutional and analyst following should reduce the information impact of insider trading and the size of these firms will make it impractical for investment bankers to provide more than surface level backstop support.

The studies that have looked at this phenomenon seem to reach consensus on two broad conclusions::
1. The price impact is negativeWhen lockups expire, stock prices drop. The drop is statistically significant (between 2 and 5%) and there is no rebound from this price drop.  In case you are tempted to try to take advantage of this price drop by selling short prior to lockup expirations, it is too small (relative to transactions costs) and too unpredictable (about a third of lockups end with increases in stock prices) to build a profitable investment strategy around.
2. The trading volume surges: Trading volume increases  on the expiration of lockup periods, with volume increasing substantially both at the time of the expiration and the periods after.  One study finds that the price impact on the lockup expiration is related to the change in liquidity in the post-lockup period, with more positive (negative) stock price reactions correlating with increases (decreases) in liquidity.

Looking at Facebook
Based on the last section, I would argue that analysts who have used the lockup expirations as a rationale for Facebook's stock price performance since its IPO  have been reaching for straws. Facebook is a large market cap firm (market cap > $50 billion), with substantial trading volume and a heavy analyst following. It is not the type of company where you would have expected to see dramatic up or down moves on lockup expirations.

That is not to say that lockup expirations are non-events, since even at Facebook, there have been sizable price reactions to the first three lockup expirations - negative (-5%) for the first one, slightly less negative on the second one (-3%) and positive (13%) for the third.  Rather than search for elaborate rationale for the different market reactions, I would point to price momentum around each of the expirations. The first lockup period expired in the immediate aftermath of a bad earnings report in August, with the stock price already sliding, and the market reaction was negative. The second lockup period expired after the stock had spiked on the third quarter earnings report but was retracing its steps in the days after. The lockup period that expired last week (on November 14) was after the second earnings report, which was viewed as good news, and when the stock had upward momentum. Looking at this small sample, it seems to me that at least in Facebook, insiders have behaved like other momentum investors, selling when everyone is selling and holding if the prevailing sentiment is positive.  If there is a broader lesson to be learned from these experiences, it is that we attribute too much wisdom and knowledge to insiders, at least at young growth companies.  Rather than being ahead of the market in their assessments of value, insiders at these companies are often just as uncertain as the rest of the market and just as likely to follow the crowd. As a stockholder in Facebook now (my limit order did come through), I am less inclined to pay attention to what Zuckerberg thinks or does about the company and more to the fundamentals that will drive its value over time.

Thursday, November 15, 2012

Storms and Stocks: Dealing with Disruptive Shocks

Sandy, the super-storm that terrorized New York and its environs is now history, but it left a trail of destruction. As communities around the city and New Jersey dealt with power failures, gas shortage and transportation chaos, I was thinking about the lessons that I learned from the experience and their application to financial markets, which have been buffeted with their own storms for the last five years.
  1. Once is an accident, twice is bad luck, but three times is a pattern: For much of the time that I have lived in the United States, power failures were not only unusual but when they did occur, lasted at most for a few hours. However, in 2011, we lost power for several days twice, once after Hurricane Irene and once after a freak ice storm, and this year, we lost power again for a week. While it is entirely possible to attribute these occurrences to chance, it is also possible that weather systems have changed and that the last two years may be more the rule than the exception going forward. The last five years in financial markets have been characterized by “unusual” macro events (banking crisis, Greece, Spain etc.) but they are unusual only because we are viewing them through the lens of recent history (the prior six decades in developed markets). As with the weather, it is possible (and I think it is likely) that these macro crises are not an aberration but are part and parcel of markets for the foreseeable future, and that investment strategies and risk management systems have to be adapted accordingly.  
  2. History provides little guidance: When there is a disruptive shock (and the storm definitely qualified), it is human nature to use past history to fill in the gaps, even if it does not quite fit. Thus, my neighbors argued that since train service was up and running a couple of days after the storm last year or the terrorist attacks in 9/11, it was likely to be back up after this one too. In financial markets, investors have used the crutch of historical data (equity risk premiums from the past, PE ratios over time) to evaluate when and where to invest these last five years. In both cases, extrapolating the past would have yielded poor predictions. 
  3. Misinformation fills the news vacuum: In the immediate aftermath of the storm, there was an information vacuum where the power and transportation companies had no useful guidance to customers and rumors filled in the gap. With each macro crisis over the last few years, we have seen the same phenomenon in markets, where rumors of deals made and unmade have moved markets substantially. 
  4. It is good to have back up systems: About 15 months ago, none of the houses in my neighborhood had back-up generators, as the cost of installing one seemed to be well in excess of any potential benefits. After this storm, I would not be surprised to hear generators starting up at a third of the houses the next time we lose power. The problem, though, is that these generators are themselves dependent upon fuel (natural gas or gasoline) to work and may end up being idle in their absence. Risk managers (at companies and financial service firms) have devised their own back up systems to protect themselves against the “last” crisis but these systems may themselves break down, in the face of the next crisis. 
  5. But it is better to design resilient systems: One reason that this portion of the East Coast was hit so hard by the storm was that it was never designed to withstand it. In particular, large power-dependent houses with finished basements, power stations that are close to the ocean or rivers and overhead power lines are all rich targets for storms like Sandy. If these storms are the new norm, we have to think about building houses that are livable without power (those older houses have lower ceilings, unfinished basements and fireplaces for a reason) and a more defensible power system. In investing we have to think about a similar redesign of how we invest, with dynamic asset allocation (reflecting the constant shifts in the macro environment) and a stock selection process that is less dependent upon rules of thumb (many of which were constructed for a past that no longer applies). 
More generally, in the face of the increasing frequency of disruptive shocks, I would pick:
(a)Simpler over more complex systems: Over the last few decades, lulled by the growth of technology and access to data, we have built more and more complex systems (in both day-to-day living and investing) that are dependent upon both. Since disruptive shocks cut off both technology and data, simpler systems will survive and bounce back faster in the face of these shocks. I am glad that my investing strategy is based on intrinsic valuation and that I can value a company with an annual report and a calculator (or even an abacus) and that I am not dependent on access to real time data or computerized trading for investments. I am even happier that I could go two weeks without tracking either the market or looking at the investments in my portfolio, without fear of a meltdown. 
(b) Decentralized over centralized systems: The “hub and spoke” system, where you centralize resources does have its advantages, primarily related to efficiency, at least in normal times. The problem with these systems is that failures at the system’s center can shut the entire system down, as passengers on United and Delta discovered, when their hubs (Newark for United and LaGuardia for Delta) shut down during the storm. Decentralizing these systems may create more coordination headaches during normal time periods but allow for faster recovery after disruption.

I am glad that the storm has passed, that I have power and that I am able to type this post on my train ride home from work. At the same time, I realize that as an investor, there are more storms coming, both from within (the fiscal cliff) and from outside (Asia’s slowdown and the EU’s future) and that I need to become more agile to weather these storms. Time to get to work….