Tuesday, January 11, 2005

Liquidation preferences / LendingTree

Here is my 2002 article on preferred stock - perhaps it still has some relevance, if not some vestige of humor:
(Update: Now with charts and pictures! :) )

VALUING STOCK OPTIONS AND PREFERRED STOCK: A CASE STUDY OF WALL STREET ANALYST REPORTS
Journal of Applied Corporate Finance - Vol. 15 - Number 2 - Fall 2002

In the past few years, a number of publicly traded companies have raised money from venture capital and private equity funds, often by issuing some form of preferred stock. Preferred stock allows the investor to earn an attractive return on an investment with lower risk (as compared to common stock, for example) and typically grants other kinds of investor rights frequently seen in traditional venture capital and private equity transactions, such as the right to veto mergers or the right to seats on the board. (1) But because these securities are not part of the “simple” (debt plus common equity) capital structure of a public company, SEC reporting rules are not designed to cover them adequately. As a result, they are sometimes overlooked by analysts who can fail to understand the implications of these securities for a company’s capital structure and how to account for them when valuing a company. To a lesser extent, but with far wider implications, the same issue affects the large number of public companies that have issued stock options either to employees or to customers, new investors, or other business partners—particularly when the issuing companies are incurring losses, because the accounting treatment then becomes even murkier.

In this paper, I analyze the capital structure of a public company that has both preferred stock and stock options outstanding, and examine investment banking equity analyst reports on the company following a major preferred stock financing. In the reports, analysts’ misunderstanding of capital structure resulted in a 36-74% miscalculation of the company’s market value.

The Case of LendingTree, Inc.

LendingTree, Inc. (Nasdaq: TREE) went public in February 2000, just a few months before the equity markets melted. Had the equity markets remained robust, LendingTree would almost certainly have done a secondary offering of common stock to raise the money it needed to execute its business plan. Instead, LendingTree’s stock price plummeted from $16.75 on its first day of trading to a low of $2 at the end of the year, before rebounding somewhat in 2001 as investors seemed to regain confidence that this “dot-com” has a sustainable business plan. To survive the market downturn, LendingTree went back to its principal venture capital investor, Capital Z, for financing. In September 2000, Capital Z invested $10 million in the company in return for preferred stock convertible into 1.25 million shares of common stock at a price of $7.98 per share. In March 2001, Capital Z led another $13.4 million preferred stock financing. As part of that deal, Capital Z also exchanged its September 2000 preferred shares for $10 million of the new preferred stock—resulting in a total of $23.4 million of preferred stock convertible into 6.69 million shares of common stock at $3.50 per share. (Additionally, the CEO of LendingTree received a $700,000 loan to purchase another $700,000 of preferred stock convertible into 200,000 common shares—giving him, in effect, an option on the preferred stock, and increasing the total amount of preferred stock issued to $24.1 million.) The preferred stock pays an 8% dividend, and has a one times (1×) “simple” liquidation preference.

The liquidation preference of preferred stock makes it in effect like super-subordinated debt—the preference amount is subordinated to all debt as well as to the creditors of the company, but senior to all equity in case of a liquidation of the company. The debt-like feature of the preferred stock is emphasized by an optional redemption schedule similar to bond offerings, starting at 118% of par in March 2004 and ending with a mandatory redemption in March 2006
for 105% of par. Prior to such redemption dates, preferred stock holders have the right to treat any merger or acquisition of the company as a “liquidation,” permitting them to receive the liquidation preference amount (in this case, one times the value of their investment) before common stock holders receive any consideration. Alternatively, preferred holders can convert to equity at will. So the preferred stock is at least as valuable as the number of common shares into which it would convert.

In this case, the preferred stock is convertible into 6.9 million shares of common stock (including the CEO’s shares). Together with the 18.7 million shares of common stock actually outstanding as of the 2001 financing, the total shares outstanding for LendingTree should have been at least 25.6 million on a fully diluted basis. However, most analysts reported shares outstanding of around 19 million in their equity reports, even after the filing of the company’s 2000 10-K which contained all the relevant deal documentation on the preferred financings. During the LendingTree earnings conference call on March 7, 2001, where 2000 earnings and the recent
financings were discussed by the company’s management, the first question asked by the audience—in this case, an analyst from Merrill Lynch—related to how one should calculate the total shares outstanding. In his answer, LendingTree’s CFO, Keith Hall, stated both numbers—the 18.7 million common shares outstanding as well as the 25.6 million shares on a fully diluted basis, but then he went on to say that according to GAAP, “18 million [should be] used for calculating a loss on net income. When we get to profitability we will be using all shares for calculating earnings per share.”

Therein lies the problem: the relevant SEC rule, FASB 128, which deals with reported earnings per share (EPS), does not permit a fully diluted number of shares to be used to calculate EPS when a company is showing a loss. The logic here is that the more shares outstanding, the lesser is the magnitude of the loss per share on a fully diluted basis. So in trying to be more conservative, FASB 128 requires the smaller number, or the total shares actually outstanding, to be used for the “fully diluted” EPS. (In other words, the “basic” and “fully diluted” number of shares is the same.) However, as is common practice, analysts often use the fully diluted number of shares outstanding not only to calculate EPS but also to calculate the value of a company, either by multiplying the number of shares outstanding by the per-share market price to get to the market’s valuation of the company, or by dividing their estimate of the equity value of the company (derived by using discounted cashflow analysis, for example) by the number of shares outstanding to get to a target share price. For calculating market values, it would not only be “conservative” but correct to use the larger number (that is, the fully diluted number of shares outstanding), rather than the smaller one. An analyst who used the smaller number of shares to calculate per-share market value would be overestimating the value of the shares, making the company’s stock look more attractive than it really is and tending to generate more “buy” recommendations.

This same problem arises with any options and warrants outstanding. Although options are largely disclosed in various public documents, they are in no way included in the fully diluted number of shares outstanding for a company showing losses. In the case of LendingTree, public documents from mid- 2001 showed that the company had 6.1 million options outstanding, which made the total potential shares outstanding 31.7 million—almost 70% more than the 18.7 million share number used by analysts. But is 31.7 million the right number to use in calculating the company’s market value?

Interestingly, venture capitalists often use the total potential shares outstanding (31.7 million in this case) to calculate the per-share value of a company. But for a fair valuation, the number of options would have to be discounted in some way to reflect the degree to which the options are in or out of the money. For companies incurring losses, FASB 128 does not allow any options to be counted in the fully diluted number of shares, again because doing so minimizes the loss per share. For companies showing profits, FASB suggests the treasury method based on the average share price of the stock over the past quarter to calculate the dilutive effect of the options. The treasury method effectively calculates a common stock equivalent number of shares based strictly on the intrinsic value—the difference between the strike price and the stock price—of all options that are in the money, without taking account of the value of their “optionality.”(2) Again, this approach may be appropriate for SEC financial statements, which like all financial statements are historical in nature, and show a snapshot of a company at a point in time. But the value of a company is forward looking in nature, and so investors and analysts need to calculate the full value of the options. The analysts’ job should be to find a method that uses available information to analyze and quantify such a value, rather than ignoring the problem. Note that this is a separate issue from the recent controversy regarding the methods used by companies to price options for the purpose of expensing them on their income statements. The latter problem concerns standardization and consistency of reporting across all companies, while the analyst’s challenge is to interpret those same standards and adjust them to paint a more accurate picture of an individual company.

In the case of LendingTree, I used the standard Black-Scholes option-pricing formula to calculate the value of the options at a given share price. But whatever method is used to value the options, that value is then divided by the current price of the stock to obtain a common stock equivalent (CSE) number of shares outstanding at the company. This CSE methodology, similar to the treasury method, provides the number of common shares outstanding that reflects the value of the options and, therefore, the dilution that would result from the potential realizable
value of those options.

Table 1 presents LendingTree’s fully diluted shares outstanding using the FASB’s treasury method as well as an estimation using Black-Scholes. In the case of the treasury method, the dilution is almost insignificant because the average stock price for the first quarter of 2001 was below the strike price of most of the options. But using Black-Scholes ascribes value to those 6.1 million options; they would be equivalent to 2.3 million shares of common stock. Added to the 6.9 million shares into which the preferred stock is convertible, this makes the total common stock equivalent number of shares 27.9 million, 49% more than the 18.7 million number of common shares actually outstanding.


An alternative method for arriving at a forward looking CSE value of the options is to use an adjusted treasury method, with either the current stock price or a target price. (This more conservative method would avoid any controversy arising from the use of a formula like Black-Scholes.) Clearly, if an analyst uses fundamental analysis to conclude that LendingTree’s stock is worth $10 per share, then those options that are “in the money” at that price would have some intrinsic value, and would be dilutive even if one didn’t assign any value to them beyond their intrinsic value (that is, any optionality value).

A look at equity research reports from five investment banks following the disclosure of the March 2001 preferred stock financing shows that none of them incorporated any dilution resulting from the preferred stock in their calculation of the firm’s market price or enterprise value. In effect, the analyst reports seem to suggest that LendingTree raised $23 million of convertible securities without causing any dilution to the common shareholders or incurring any debt on the balance sheet. Table 2 shows that as a result of this error (compared to the CSE methodology using Black-Scholes values for options as outlined above), the analysts miscalculated the market capitalization and enterprise value of LendingTree by 36-74%. Even on the basis of the adjusted treasury method and the analysts’ target prices, the margins of error are a very significant 41-73%.


In these comparisons, I have not included any value for the additional features of the preferred stock, such as the liquidation preference or dividends. (3) Dividends are significant if one is valuing a company based on an exit date a number of years in the future, because of any accrued dividends until that date. The liquidation preference can be viewed as a put on the common stock and could be included in the common stock equivalent capitalization table. Of course, the inclusion of these features would have shown the analysts’ miscalculation to be even more severe. (4)

As it is, if the options and warrants outstanding are not taken into account, the valuation error relating to the preferred stock alone is 28-60%. Or, as seen in Table 3, even if the analysts had correctly taken the preferred stock into account and used a 25.6 million share count, the error associated with the stock option calculation alone would have amounted to 7-13% of market capitalization using the adjusted treasury method (with the target price), and 6-12% using the CSE Methodology (with Black-Scholes).


Many analysts use the market capitalization of a company in relation to a financial measure like earnings or EBITDA (earnings before interest, taxes, depreciation, and amortization) to derive an earnings “multiple.” This multiple is then compared to multiples of other similar companies, on which basis a price target can be set for the stock. To the extent that analysts miscalculate the market capitalization of a company, their basis of comparison and thus their price targets are also mistaken. As seen in Table 4, for the three investment banks that discuss at least one of the measures they use to judge the comparative value of the firm, the correct share count would result in a 37-46% difference in the target price. Perhaps coincidentally, LendingTree’s multiples without these errors are much closer to those of the target or comparable companies mentioned by the analysts. So even though the analysts do not understand the capital structure of these companies, the “market” itself seems to.


Interestingly, many of the analysts did mention the dilution resulting from the preferred financing in their research reports relating to the financing, but for some mysterious reason the dilution effect failed to make its way from the text of these reports to the numerical analysis. The only exceptions (to some extent) were Putnam Lovell and Friedman Billings. Putnam Lovell included a pro forma calculation of shares outstanding to project its earnings per share estimates. In the reports from March and April 2001, the analyst gave an EPS estimate using the total shares outstanding of 20 million, but also a pro forma EPS which included full dilution of 25.5 million and another pro forma figure which included an additional 2 million shares “due to warrants and dividends on the convertible preferred stock and debt issuance.” However, as seen above, this same report failed to include the more accurate pro forma fully diluted shares outstanding figure in its calculation of the firm’s value. Furthermore, these pro forma share calculations disappeared from the Putnam Lovell reports by the end of 2001.

In the case of Friedman Billings, the analyst seemed to discover the importance of the preferred shares as time went by. In the reports from early to mid-2001, no reference was made to the dilution from the preferred stock. In October 2001, the Friedman Billings report stated that “we have not included approximately 7M in preferred stock in our share count for 2002 as GAAP rules would not permit it. However, as time goes on these shares will ultimately need to be incorporated… the preferred shares will start to count by 2003.” But by March 2002, the analyst had understood that it made little sense not to include the additional shares in the share count before 2003—they are part of the capital structure whether LendingTree is profitable or not. And so Friedman Billings came to the correct conclusion that “to address TREE’s valuation, we believe it is appropriate to include the 7.0M shares in the market capitalization of the company” (March 14, 2002).

In May and June 2002, two other Wall Street firms, Ladenburg Thalman and Salomon Smith Barney, initiated coverage of the stock and accounted for the preferred stock in their share count calculations. Perhaps due in part to the probing by these two banks, and in light of the company’s projected return to profitability in the third quarter, LendingTree’s reports for the second quarter of 2002 made the fully diluted shares outstanding in the company far more apparent than did previous reports. In the company’s earnings conference call, the management noted an increase in the shares outstanding for the company: The actual number of shares outstanding increased due to a new common stock share offering in the second quarter of 2002, as well as the exercise of stock options and the conversion of some of the preferred stock.
“You’ll note that our basic and diluted share count figures have increased for the third and fourth quarters. This reflects a number of stock option exercises and a few conversions of our convertible preferred stock into common shares [in the] 2nd quarter. The adjustments in the number of basic shares outstanding has caused the EBITDA per basic share for the third quarter to round down to 12 cents from the previous guidance of 13 cents. Note however that the absolute number for the EBITDA earnings of $2.6 million for the third quarter remains unchanged from our previous guidance.” In that same conference call, the chief financial officer also explained the projected 2002 EPS numbers as follows: “Let me just go into a little footnote there. The fully diluted share count next year is estimated [to be] an average of 34.8 million shares which includes the preferred shares outstanding. As we’ve discussed on previous conference calls, GAAP accounting requires the use of the highest share count when the company becomes profitable.” As per the company’s earnings release, that number includes an increased number of common shares at 22.1 million,(5) 6.6 million shares from the potential conversion of the preferred stock, and an increase in options outstanding resulting in 3.9 million additional common equivalent shares using the treasury method. But again, these additional shares are part of LendingTree’s capital structure and should be used to calculate market capitalization whether the firm is profitable or not.

Following management’s “footnote” that the share count had increased 79% from the first quarter’s “weighted average common share” count of 19 million to a projected 35 million in 2003, the investment banks dutifully followed LendingTree’s guidance and adjusted their reports to reflect this update. In its July 25th 2002 report, Merrill Lynch stated: “One key difference between our [original] forecast and the company’s was a much higher level of average shares in the company model than we were using ($0.03 per share impact).” Luckily, LendingTree’s better-thanexpected performance helped disguise the analysts’ miscalculation of share count. UBS Warburg’s July 25th report states: “As the company achieves profitability, our per share figures will reflect fully diluted shares as opposed to basic shares. In TREE’s case, this calculation was complicated by a series of convertible preferred shares TREE issued in March 2001… In short, these shares will be included in our 2003 fully diluted share count. As a result, our EBITDA per share estimate falls from $0.80 to $0.50. The impact of the greater marketing expense is $0.07; therefore, if there were no change in share count, our EBITDA/share estimate would have fallen to $0.73.” In other words, this represents a 32% decline in 2003 EPS due to the preferred shares that had been issued in 2001.

Conclusions

The significant share count errors encountered in the equity research reports for LendingTree have some important implications. First, the magnitude of the problem—up to 74% in measuring LendingTree’s enterprise value—suggests that analysts should pay greater attention to the implications of financial instruments like preferred stock and stock options. As shown in this paper, accounting rules for the number of shares outstanding in a company may be appropriate for historic EPS reporting purposes, but cannot be used blindly in valuing a company with a complex capital structure. The problem is especially severe for companies that show negative earnings, because in these cases the accounting rules exclude convertible securities in calculating EPS.

With regard to stock options, the magnitude of the error was no more than 13% in the case of LendingTree, but the problem may be more relevant in other companies. Given the number of publicly traded companies with negative earnings and significant option pools, as well as a cursory glance at various analyst reports on these companies, one can safely deduce that this problem may have been quite widespread in recent years and could certainly be a contributing factor to the overvaluation of venturefunded, option-laden technology stocks with negative earnings during the recent stock market bubble.

It is interesting that most of the recent controversy regarding the accounting for stock options has centered on the fact that the granted options are not expensed in a company’s income statement. One of the arguments often put forth by opponents of expensing stock options is that it would constitute double counting if options were expensed in the income statement in addition to being reflected in the fully diluted number of shares outstanding. But given the results outlined in this paper, advocates of expensing stock options could argue that for companies with negative earnings, stock options issued as compensation show up neither in the income statement nor as a dilution to the number of shares outstanding—at least as far as the analysts are concerned.

Finally, in a larger context, this paper may also shed some light on the recent controversy surrounding investment banks and their analysts’ stock recommendations. The errors highlighted here seem to suggest that naïveté or a lack of knowledge (rather than willful deception) can affect analysts’ perception of stock valuations. LendingTree’s analysts certainly show great vertical expertise in the industries they cover and great prowess in understanding the complicated business models and operating metrics used by a company like LendingTree — their reports are packed with sophisticated measures of the progress of the company’s business. Yet the same analysts can fail to accurately count the number of shares outstanding in the company. Friedman Billings was the only firm covering the stock at the time of the financing to use a correct share count without company prompting (albeit long after the financing was announced), which suggests that analysts may not be spending enough time addressing the capital structures of the companies they cover. The fact that most of the analysts only adjusted their share counts more than a year after the financing, after management led the way, suggests that the analysts were following the company’s earnings guidance numbers too closely, without applying much independent judgment or analysis. But if analysts ignore a critical, yet simple, measure like the number of shares outstanding, then it should come as no surprise that the investment community might fail to understand the implications of Enron’s special purpose funding vehicles, for example.

Investment banking analysts as well as auditors and the companies themselves must provide clearer disclosure on the balance sheet or income effects of complex financial instruments — including the common stock equivalent of any convertible instruments or the debt equivalent liability inherent in any offbalance sheet financing—even if some judgment is required in making the relevant calculations. The “common stock equivalent” share calculations in this paper could help to simplify the complexities of capital structures and defeat their potentially obfuscating nature.

_____________________________________
Footnotes:
(1) All such rights would typically appear as additional terms or provisions relating to the preferred stock of the company, in the documentation of the transaction.
(2) Methods like the Black-Scholes formula are used to assign additional value due to the ‘optionality’ feature. The treasury method effectively assumes an immediate exercise of the options and calculates the number of shares that could be purchased from the proceeds of the option exercise.
3. In another paper (“Preferred Stock in Venture Capital—A Valuation Methodology,” unpublished) I describe a simple methodology to interpret various features of more complex financial instruments that are quite typical in private equity and venture capital transactions. In brief, many complex structural features can be translated to debt, common stock, or options on common stock, and so one can value a company by reducing the capital structure to debt and common stock equivalents.
4. The amount of dilution due to the value of the put would be approximately 3.4 million shares, or 12% at a stock price of $4, but only 0.4 million or 1% at a stock price of $10.
5. The actual number of shares outstanding increased due to a new common stock share offering in the second quarter of 2002, as well as the exercise of stock options and the conversion of some of the preferred stock.


Labels: