The value relevance of financial statements within and across private and public equity markets

John R. M. Hand Tel: (919) 962-3173 Kenan-Flagler Business School Fax: (919) 962-4727 UNC Chapel Hill hand@unc.edu Chapel Hill, NC 27599-3490

Abstract This study finds that the value relevance of financial statements follows a ‘sawtooth’ pattern as firms mature within and across private and public equity markets. Specifically, I show that the strength of the associations between the equity values and financial statements of U.S. biotech companies rises as firms mature toward an IPO, falls at the IPO, and increases again after the IPO. I argue that this pattern is due to underlying sawtooth dynamics in firms’ investment opportunity sets and in the financial sophistication of the marginal investor in firms’ stock. I also find that the financial statements of U.S. biotech companies are value relevant in surprisingly similar ways in private as compared to public markets. In each market, equity values are positively related to R&D expense, cash, and noncash assets; are unrelated to revenues and SG&A expense; and are negatively related to long-term debt. And in each market, financial statements explain over one-third of the cross-sectional variance in equity values. Key words: Biotechnology; firm maturity; investment opportunity set; investor

sophistication; private equity; public equity; value relevance. JEL classifications: G12, M21, M41.

First draft: October 14, 2002 This draft: June 21, 2003

The comments of R.T. Ball, M. Barth, R. Beatty, B. Beaver, R. Bushman, A. Davila, G. Foster, W. Landsman, J. Raedy, D. Shackelford, K.R. Subramanyam, B. Trueman, X.-J. Zhang and seminar participants at Berkeley, Stanford, UNC and USC are appreciated. Thanks also to M. Mumma and G. Kong of InterSouth Partners and F. Worthy of A. M. Pappas & Associates for their insights into venture capital valuation, and Mark Edwards of Recombinant Capital for valuation data. The Edward O’Herron, Jr. Fund for Distinguished UNC Faculty, the UNC Center for Technology and Entrepreneurial Venturing, and the UNC Center for Technology & Advanced Commerce provided generous financial support for this research.

2

1. Introduction and summary

Much research has shown that financial statements are value relevant in public equity

markets—that is, they are associated with the equity values of, and returns to, publicly traded

firms (Lev and Ohlson, 1982; Bernard, 1989; Kothari, 2001). In this study, I break new ground

by examining the value relevance of financial statements within private equity markets, and

across the transition from private to public equity markets. Both settings are previously

unexplored but are economically important in the life of a firm.

Although the value relevance of financial statements within and across equity markets is

likely to depend on many factors (such as regulation, liquidity, price-setting mechanisms, and

degree of information asymmetry between managers and investors), I focus on the roles played

by firms’ future investment opportunities, and by the degree of sophistication of the marginal

investor in their stock, since both factors have been found to significantly influence publicly

traded firms’ financial decisions (Smith and Watts, 1992; Gaver and Gaver, 1993) and stock

returns (Bernard and Thomas, 1989, 1990; Hand, 1990).

The central hypothesis that I develop and test is that the value relevance of financial

statements will be a nonlinear ‘sawtooth’ function of firm maturity when measured within and

across private and public equity markets for firms that file to go public. Specifically, I predict

that value relevance will rise as a firm matures toward an IPO, fall at the IPO, and then increase

again after the IPO. My hypothesis is built on three propositions.

First, I propose that VA/(VA + VFIO), the fraction of a firm’s equity value that comes from

its net assets in place VA relative to its future investment opportunities VFIO will be a sawtooth

function of firm maturity when measured within and across equity markets for firms that file to

go public. The sawtooth is created by the interaction of an increase in VA/(VA + VFIO) as firms

mature within a given market, and a decrease in VA/(VA + VFIO) at the time of an IPO, when

firms transition across markets. The unconditional tendency for VA/(VA + VFIO) to increase

within a given market occurs because when firms are founded VA/(VA + VFIO) ≅ 0 since their

value consists almost entirely of ideas for profitable future investments, not actual assets in

place. As firms mature, ideas for profitable future investments are converted into actual assets in

place, leading VA/(VA + VFIO) → 1, unless the firm creates new and increasingly large future

investment opportunities which in expectation is economically implausible. VA/(VA + VFIO)

decreases at the time of an IPO for two reasons. One, private equity markets can be viewed as

3

tournaments that determine which young firms have the most valuable future investment

opportunities. Firms that win the tournament and file to go public will be those whose values of

VA/(VA + VFIO) are the lowest. Two, firms that go public benefit from having their stock traded

in a liquid market, thereby removing a private equity illiquidity discount of between 15% and

40%. This will lower VA/(VA + VFIO) because illiquidity discounts will be more severe for future

investment opportunities than for assets in place.

Second, I propose that the probability that the marginal investor in the firm’s stock will

be financially sophisticated, Psoph, will also be a sawtooth function of firm maturity when

measured within and across equity markets. The price-setting investor in a private equity market

will always be highly sophisticated because the limited partners in private equity funds have

strong incentives and deep resources with which to hire the most experienced and financially

sophisticated general partners, such as venture capitalists (Gompers and Lerner, 2000).

However, when a firm undertakes an IPO, it almost always sells some of its stock to

unsophisticated retail investors. On average this lowers Psoph (or equivalently, lowers the

expected level of sophistication of the marginal investor). After the IPO, Psoph improves as the

firm grows and attracts analyst coverage and institutional shareholders. Moreover, I argue that

any tournament- or liquidity-based decline at the IPO in VA/(VA + VFIO) will be reinforced by

mispricing pressure from unsophisticated investors who overestimate the value of the firm by

overestimating the intangible VFIO proportionally more than the tangible VA (Miller, 1977;

Daniel, Hirshleifer and Subrahmanyam, 1998).

Third, I suggest that the value relevance of firms’ financial statements will be positively

related to VA/(VA + VFIO), and positively related to Psoph. To illustrate this, let value relevance be

defined as the goodness of fit in a regression of equity value on financial statement data. When

the marginal investor is financially sophisticated, equity value will be the rational and efficiently

discounted sum of future cash flows attributable to assets in place and future cash flows

attributable to future growth opportunities/investments in positive NPV projects (Myers, 1977).

However, because U.S. GAAP only allows financial statements to record assets in place, not

future investment opportunities, value relevance will be increasing in VA/(VA + VFIO), assuming

that assets in place are non-negatively correlated with future investment opportunities. To the

extent that the marginal investor in the firm’s stock is less than fully financially sophisticated, the

value relevance of financial statements will be reduced because the lack of financial

4

sophistication creates equity values that are noisier than they otherwise would be. This noise

lowers the value relevance of financial statements because it increases the residual variance of

the regression, thereby diminishing the goodness of fit (value relevance) of the regression.

Combining the three propositions above yields the central hypothesis of the paper—that

the value relevance of financial statements will be a nonlinear sawtooth function of firm maturity

when measured within and across private and public equity markets. Value relevance is

predicted to rise as a firm matures toward an IPO, fall at the IPO, and then increase again after

the IPO.

I test the sawtooth hypothesis using longitudinal data from 186 U.S. biotechnology

companies over the period 1992–2003. Biotech firms are chosen because they have large

expected future investment/growth opportunities relative to assets in place; share a similar

production function in the R&D-intensive search for new drugs; are typically funded in well-

defined stages by organized private equity; and tend to go public quite rapidly. Biotech company

pre-money private equity valuations were obtained from Recombinant Capital and those of firms

that filed to go public were retained. Pre-money valuations are equity values before accounting

for the capital put in by private equity investors in the current funding round.

The resulting set of IPO-conditional private equity valuations were then matched with up

to five years of annual pre-IPO financial statements obtained from the IPO filing documents,

yielding 458 pairs of private equity valuations plus the preceding fiscal year’s financial

statements. The pre-IPO valuation points range from Series A funding rounds to Series F

funding rounds and beyond. A similar matching procedure was applied to the IPO filing date,

where the firm’s equity value was defined using the offer price reported in the IPO filing

documents (i.e., not including the funds targeted to be raised through the IPO). This yielded 96

pairs of IPO filing equity valuations plus the preceding fiscal year’s financial statements.

Finally, three post-IPO public equity market valuations were used, namely three months after

each of the first three fiscal years after the IPO. Post-IPO valuations were matched with the

preceding fiscal year’s financial statements filed with the SEC to yield 441 pairs of post-IPO

public equity valuations plus the preceding fiscal year’s financial statements.

Pooled and financing round by financing round cross-sectional regressions of equity

values on key components of firms’ balance sheets and income statements were then conducted.

In those regressions, I find evidence of the predicted sawtooth pattern in the value relevance of

5

financial statements within and across markets. Specifically, I find that the regression adjusted

R2 that is uniquely attributable to financial statements increases as firms mature toward their

IPO, declines at the IPO, and increases again after the IPO. I define firm maturity not by the age

of the firm per se, but the firm’s position in the normal private equity financing sequence (Series

A, Series B, etc.) because valuation data is most precisely available at the date of a particular

financing round, rather than at any given age of the firm in years since founding. I demonstrate

that the sawtooth pattern in adjusted R2 is not a spurious effect of scale or other factors that can

impede comparisons of adjusted R2 across samples.

Consistent with the proposition that the sawtooth pattern in value relevance stems from

underlying sawtooth patterns in the maturing of firms’ future investment opportunities and the

probability that the marginal investor in firms’ stock is sophisticated, I show that the ratio of total

assets to the market value of the firm, a commonly used proxy for the importance of the

investment opportunity set (Smith and Watts, 1992; Gaver and Gaver, 1993), increases as firms

mature in the private equity market, falls at the IPO filing, and increases after the IPO. A

commonly used proxy for the probability that the marginal investor in a firm’s stock is

sophisticated, the percentage of shares held by institutional shareholders (Hand, 1990; Bartov,

Radhakrishnan and Krinsky, 2000; Balsom, Bartov and Marquart, 2002), increases post-IPO.

In addition to these results, I also find that despite intermarket differences in regulation,

liquidity, price-setting mechanisms, and information asymmetry, financial statements are as

value relevant in private equity markets as they are in public equity markets. Qualitatively, I

observe that in both types of markets biotech firms’ equity values are reliably positively related

to R&D expense, cash, and noncash assets; are unrelated to revenues and SG&A expenses; and

are negatively related to long-term debt. Quantitatively, financial statements explain over one-

third of the cross-sectional variance in biotech equity values in both private and public markets.

In conclusion, this paper contributes to the value relevance literature in several ways. It

is the first work to study the associations between financial statement data and the equity values

of privately funded companies. In doing so, the paper demonstrates that at least for biotech

companies, financial statements are as qualitatively and quantitatively value relevant in private

equity markets as they are in public equity markets. Most importantly, the paper develops, tests

and finds evidence consistent with the hypothesis that interactions between the maturing of

firms’ investment opportunity sets and the dynamics of the financial sophistication of the

6

marginal investor create a nonlinear ‘sawtooth’ pattern in the value relevance of financial

statements when measured within and across markets for firms that file to go public. For biotech

firms, the value relevance of their financial statements rises as they mature toward an IPO, falls

at the IPO, and increases again after the IPO.

The remainder of the paper proceeds as follows. Section 2 describes the market for

private equity and venture capital. Section 3 develops in more detail the hypothesis on which the

paper is centered. Section 4 outlines the novel data set created for and used in the study and

reports descriptive statistics, while section 5 motivates the regression models and experimental

designs employed. Section 6 reports the regression results and the results of several other

supporting empirical analyses. Section 7 concludes.

2. Private equity markets 2.1 Institutional background

Private equity markets differ from public equity markets in many ways (Gompers and

Lerner, 2000). For example, public equity markets are highly regulated by the S.E.C. and stock

exchanges, while private equity markets are not. Private equity firm valuations are set through

face-to-face negotiations between management and a small number of wealthy, professional and

risk-tolerant investors. Valuations in a public equity market are set anonymously without direct

contact with management through the interactions of large numbers of investors, many of whom

are risk-averse and do not have significant wealth or professional investing experience. Public

equity markets are highly liquid, while private equity markets are illiquid. Private equity

investors can extract management’s private information because of their frequent interactions

with management and holding of board seats, but public equity investors must rely almost

exclusively on public data and face significant information asymmetries.

The private equity market is made up of four submarkets: organized, angel, informal, and

Rule 144A (Fenn, Liang and Prowse, 1995). This paper focuses on organized private equity,

defined as unregistered investments in the equity of private companies by professionally

managed entities, made either directly by professional investors such as pension funds and

endowments, or indirectly by these investors through intermediaries, particularly venture capital

7

partnerships.1 Funds invested in organized U.S. private equity have grown substantially over the

past two decades, rising from $5 billion in 1980 to almost $300 billion in early 2001 (Lerner,

2001). Private equity funds invest in a wide variety of vehicles and objectives, including young

companies, leveraged buyouts, consolidation, mezzanine financing, and distressed debt. Of

private equity funds, venture capital partnerships are the most common.

2.2 Venture capital

Venture capital is independently managed, dedicated capital invested in young, often

startup or early-stage technology businesses that are highly risky but also have very strong future

growth and profit potential (Gompers and Lerner, 2000). Venture capital funds are usually

structured as partnerships of venture capitalists that raise money in staged amounts from wealthy

private investors, companies and institutions. A fund usually has a ten-year life and invests in a

portfolio of private companies, often restricted to one or two sectors such as biotechnology or

software. Although venture capitalists put up only about 1% of the limited partnership’s capital,

they manage the fund through their role as general partners. In exchange for finding, screening,

and deciding upon the companies to invest in, venture capitalists are paid an annual management

fee that is usually between 1.5% and 3% of the fund’s committed capital or net asset value, and

they receive about 20% of the profits made by the fund’s investments.

The typical investment made by a venture capitalist is in illiquid preferred stock that is

only convertible into liquid common stock or cash at one of two major exit points: either an IPO

or the sale of the company to another entity. This usually occurs within a targeted window of a

certain number of years. Although venture capitalists often provide a firm a measure of long-

term financing by investing in several financing rounds, they also provide business expertise and

connections. The venture capitalist usually serves on the firm’s board of directors; provides the

entrepreneur with financial sophistication, operating services, and a network of business

contacts; helps recruit key personnel; and imparts financial and strategic discipline.

Research into venture capital has blossomed over the past decade as researchers have

1 Angel private equity consists of investments made by wealthy individuals, typically arranged by matchmakers such as lawyers and accountants. Informal private equity is similar to angel capital except that firms sell unregistered equity securities to both institutional investors and wealthy individuals across a larger number of such investors. Rule 144A private equity is underwritten private equity offerings under the SEC’s Rule 144A, which establishes the rules and conditions under which private securities are permitted to be traded among certain classes of institutional investors (Fenn, Liang and Prowse, 1995).

8

exploited the fact that venture capitalists face many of the same problems as public investors, but

to a more extreme degree (Lerner, 2000, 2001). An excellent summary is provided by Gompers

and Lerner (2000), spanning topics such as the compensation of venture capitalists (Gompers and

Lerner, 1999); the optimal investment, monitoring and staging of venture capital (Gompers,

1995); the decision to go public (Lerner, 1994a); and the long-run performance of venture-

backed IPOs (Brav and Gompers, 1997). Kaplan and Stromberg (2002a, b) empirically relate the

characteristics of venture capital contracts to theories of financial contracting.

3. The ‘sawtooth’ hypothesis 3.1 Dynamics of the investment opportunity set The value of a firm’s equity, V, can be broken into two parts: the present value of net

assets in place, VA, and the present value of future profitable investment/growth opportunities,

VFIO (Miller and Modigliani, 1961). As noted by Myers (1977), growth opportunities are the real

options that a firm has or may create to make future investments that earn a rate of return in

excess of its opportunity cost of capital. Such growth opportunities are frequently referred to as

the firm’s investment opportunity set and have been shown to be an important determinant of

many corporate decisions (Adam and Goyal, 2003). For example, the relative mix of assets in

place versus future investment opportunities affects a firm’s accounting policies (Skinner, 1993),

capital structure (Myers, 1977; Smith and Watts, 1992), compensation contracts (Smith and

Watts, 1992; Gaver and Gaver, 1993), and dividend policy (Smith and Watts, 1992).

I propose that for firms that file to go public, the fraction of their total equity value that

comes from assets in place relative to future investment opportunities will be a sawtooth function

of firm maturity when measured within and across equity markets. In figure 1, the sawtooth

shape of the thin solid line VA/(VA + VFIO) is created by the interaction of an unconditional

increase in VA/(VA + VFIO) as a firm matures within a given market and a conditional decrease in

VA/(VA + VFIO) when the firm transitions from a private equity market to a public equity market

through an IPO. The unconditional increase in VA/(VA + VFIO) within a given market occurs

because when firms are founded, VA/(VA + VFIO) ≈ 0. At founding, the value of a firm consists

almost entirely of ideas the founders have for profitable future investments, not recognized on-

balance-sheet assets in place (Myers, 1977, p.151). However, as firms mature, the VFIO that is

present at founding is converted into VA through the process of raising capital and investing that

9

capital into a variety of current, fixed, and intangible assets and liabilities (e.g., cash, PP&E,

R&D, and debt). The result of this idea-to-cash conversion process is that as a firm matures,

VA/(VA + VFIO) → 1. Although technically it could be the case that a maturing firm could

maintain VA/(VA + VFIO) ≅ 0, it could only reasonably do so by continually creating new and

increasingly large future investment opportunities, something that in expectation is economically

implausible in the long-term.2

I propose that the conditional decrease in VA/(VA + VFIO) at the IPO occurs for one or

both of two reasons: removal of the private equity illiquidity discount and/or the ‘tournament’

nature of private equity markets. Firms that go public benefit from having their stock traded in a

liquid market, thereby removing a private equity illiquidity discount of between 15% and 40%

(Houlihan Valuation Advisors/VentureOne, 1998). This will lower VA/(VA + VFIO) because the

illiquidity discount will likely be more severe for the firm’s future investment opportunities than

for its assets in place, since there are more liquid markets for the latter than the former.

Alternatively, a private equity market can be viewed as a tournament to determine which young

firms have the most promising investment opportunities. The prize for “winning” is that public

equity markets provide the winners with the large amounts of capital needed to convert their

intangible investment opportunities into tangible cash flows (and perhaps also an increase in firm

value through a lower cost of capital). The cost, however, is dilution through having to share

their future cash flows with public investors. Firms that file to go public are those that

underwriters, in their capacity as expert screeners and across-market agents for public investors,

determine to have the most valuable investment opportunities. As a result, unless there is no

uncertainty as to which privately funded firms have the best investment opportunities (and

therefore which firms will win the private equity market tournament), firms that file to go public

will have values of VA/(VA + VFIO) that are lower than both the average of all firms that could 2 This view of the dynamics of assets in place and future investment opportunities corresponds to the conventional four-phase model of the firm’s life cycle: start-up, growth, maturity, and decline (Smith, Mitchell and Summer, 1985; Black 1998). In the start-up phase there are few if any assets in place and the largest fraction of a firm’s value stems from its ideas, intellectual property and growth opportunities. In the growth phase, financing has been obtained, investments have been made and operating activities are underway. Although the fraction of firm value attributable to assets in place is higher than in the start-up phase, growth opportunities remain large and near-term financing requirements remain substantial. In the maturity phase of the life cycle, the firm’s growth has slowed substantially, its assets in place dominate its valuation, and its cash from operations is sufficiently strong that it does not need external financing. Finally, in the stagnant or decline phase, the firm is either in a no-real-growth steady state where its earlier abnormal profits have all been competed away, or the firm is destroying shareholder value in its investments and operations, making it a candidate for restructuring, a takeover, or liquidation (Smith, Mitchell and Summer, 1985).

10

have been taken public, and their own historical VA/(VA + VFIO).

3.2 Probability that the marginal investor in a firm’s stock is financially sophisticated Traditionally, research in finance has assumed that rational and sophisticated investors set

security prices. Over the past 15 years, however, the field of behavioral finance has emerged to

challenge this traditional assumption. In broad terms, behavioral finance argues that some

financial phenomena can be better understood using models in which some investors are less

than fully rational (Barberis and Thaler, 2002; Daniel, Hirshleifer and Teoh, 2002).

Accounting research, which has both borrowed from and contributed to behavioral

finance, has characterized imperfect rationality as a lack of sophistication in understanding

financial statements. For example, Bernard and Thomas (1990) suggest that investors do not

fully understand the time-series properties of quarterly EPS. Hand (1990) finds evidence that the

probability that the marginal investor reacts to stale income statement information depends on

how financially sophisticated he or she is. Bartov, Radhakrishnan and Krinsky (2000) suggest

that the trading activity of unsophisticated investors underlies the predictability of stock returns

after earnings announcements. Balsom, Bartov and Marquart (2002) report that unsophisticated

investors recognize earnings management by firms more slowly than do sophisticated investors.

In this study, I hypothesize that Psoph, the probability that the marginal investor will be

financially sophisticated will be a key determinant of the value relevance of historical financial

statements for two reasons. First, arriving at an unbiased and efficient expectation of the value

of future cash flows from assets in place and investment opportunities requires a sophisticated

understanding of financial accounting. I propose that the price-setting investor in private equity

markets will always be highly sophisticated because the limited partners in private equity funds

have strong incentives and deep resources with which to hire the most experienced and

financially sophisticated general partners, such as venture capitalists (see fig. 1). However, when

a firm undertakes an IPO, it will almost always sell some of its stock to unsophisticated

individual investors. On average, this lowers Psoph (or equivalently, lowers the expected level of

sophistication of the marginal investor). After the IPO, Psoph will improve as the firm grows and

attracts analyst coverage and long-term institutional shareholders.

The second role played by Psoph in determining the value relevance of financial statements

is that of reinforcing the tournament effect described in section 3.1. I suggest that any

11

tournament- or liquidity-based decline at the IPO in VA/(VA + VFIO) will be reinforced by

mispricing pressure from unsophisticated investors who overestimate the value of the firm by

overestimating the intangible VFIO proportionally more than the tangible VA. Prior research

lends support to this proposition. For example, Miller (1977) shows that investors who demand

shares offered in an IPO are likely to be those that are most optimistic about the firm’s prospects.

Purnanandam and Swaminathan (2003) find that over the period 1980–1997, the median IPO is

overvalued at the offer price by between 14% and 50% and that this overvaluation arises

because, in a manner consistent with the investor overconfidence theory of Daniel, Hirshleifer

and Subrahmanyam (1998), financially unsophisticated investors place too much weight on

forecasts of future revenue and earnings growth and pay too little attention to current

profitability.3 Yetman (2003) finds that periods of high investor sentiment are associated with

less rational and less efficient use of accounting information at IPOs. Shiller (1990) reports that

only about one-quarter of respondents in a large survey undertook any fundamental analysis

when evaluating whether or not to invest in firms going public.

3.3 Value relevance of historical financial statements within and across equity markets

Financial statements are value relevant if they are reliably associated with equity prices

or returns. I arrive at the hypothesis that the value relevance of financial statements will rise as

firms matures toward an IPO, fall at the IPO, and then increase again after the IPO by proposing

that value relevance is positively related to VA/(VA + VFIO) and positively related to Psoph. These

propositions convert the sawtooth relations between the maturing of the firm’s future investment

opportunities (section 3.1) and the dynamics of probability that the marginal investor will be

sophisticated (section 3.2) into a sawtooth pattern between value relevance and firm maturity.

To illustrate, let value relevance be defined as the adjusted R2 in a regression of equity

values on financial statement data. If the marginal investor is financially sophisticated, then

equity value, the dependent variable in the regression, will be the rational and efficiently

discounted sum of future cash flows attributable to assets in place and future cash flows

attributable to future investments in positive NPV projects (Myers, 1977). U.S. GAAP only

allows financial statements to record assets in place, not future investment opportunities.

3 The tendency for overvaluation at the IPO to be corrected in the long run leads to the long-run underperformance of IPOs documented by Ritter (1991) and Loughran and Ritter (1995).

12

Therefore, if assets in place are (as seems reasonable to suppose) non-negatively correlated with

future investment opportunities, then value relevance will be increasing in the size of a firm’s

assets in place relative to its future investment opportunities. In addition, to the extent that the

marginal investor is not financially sophisticated, equity values will be noisier, the regression

adjusted R2 worse, and the value relevance of financial statements less.

4. Data In this section I describe the construction and characteristics of a novel database that

combines private equity valuations and relevant financial statement data. Because the database

consists entirely of biotech firms, I begin with a brief explanation of biotechnology and

economic research into the business of biotechnology.

4.1 Biotechnology

Biotechnology is the application of technology to the life sciences, wherein living cells or

their processes are used to solve problems and to perform specific industrial or manufacturing

processes. Biotech applications include the production of drugs, synthetic hormones and bulk

foodstuffs, the bioconversion of organic waste, and the use of genetically altered bacteria.

Biotech firms are highly dependent on the intellectual property (ideas, discoveries, patents)

generated through their large R&D expenditures, and as such are among the most intangible-

intensive of businesses. The value chain of the typical biotech firm stretches some 10–15 years

from founding through patenting to successful FDA approval and product sales (fig. 2). Biotech

is therefore a very risky but potentially very lucrative equity investment.

I choose the biotechnology sector to test the sawtooth hypothesis described in section 3

because biotech firms have large investment opportunity sets, share a similar production function

in the R&D-intensive search for new drugs, and are usually funded in well-defined stages by

organized private equity. The typical young biotech firm is in an intense R&D race against

competitors to discover and patent a new drug. As such, it has large capital needs over a long

period of time. In the early stages of its life, capital needs are met by private equity, usually in

the form of venture capital and strategic investments from big pharmaceutical companies.

However, capital needs quickly become so large that they can only be satisfied through an IPO

or a merger with a big pharmaceutical company. Successful biotech firms therefore tend to go

13

public very rapidly, and it is not uncommon for a biotech firm’s S-1 filing with the S.E.C. to

contain financial statement data that span its entire life. As the biotech sector has been in

existence for some 30 years, this has led to a steady stream of young, start-up biotech firms as

well as more mature biotech firms that are going, and have gone, public.

Economic research into biotechnology has spanned three areas: intellectual human capital

(Zucker and Darby, 1996, 1998; Zucker, Darby and Brewer, 1998), strategic alliances (Robinson

and Stuart, 2000), and valuation (Stuart, Hoang and Hybels, 1999; Nicholson, Danzon and

McCullough, 2002; Darby, Liu and Zucker, 1999). In accounting, Joos (2002) finds that the

level and rate of growth in R&D expense, R&D success, and competitive structure all help

explain cross-sectional variation in market-to-book ratios for pharmaceutical drug manufacturers

that operated in the pharmaceutical preparation industry over the period 1975–1998. Ely, Simko

and Thomas (2003) conclude that the average stage at which a firm’s portfolio of drugs resides

significantly conditions the value relevance of the firm’s R&D expenditures. Guo, Lev and Zhou

(2003) find that biotech firms’ disclosures affect their bid-ask spreads and stock return volatility.

Finally, Hand (2003a) finds that balance sheet, income statement and statement of cash flows

data explain cross-sectional variance in biotech firms’ post-IPO equity market values, and that

the mapping of biotech firms’ R&D expenditures into equity market value is a function of the

location of R&D in the biotech value chain of discovery, development and commercialization, as

well as the growth rate in R&D spending.

4.2 Sample selection, equity valuations and financial statement data The private equity market data used in this study are integrated from two sources. The

starting point is a set of pre-IPO pre-money valuations purchased from Recombinant Capital

(www.Recap.com). Recap has collected what it indicates is a full set of round-by-round

financings for over 600 biotech companies that begins in the early 1980s. Pre-money are used

rather than post-money valuations because pre-money valuations are independent of the amount

invested in the firm during the current financing round (Lerner, 1994b).

For those firms in Recap’s database that filed to go public, I obtained financial statement

data from S-1 and 424B documents available online at www.sec.gov. In doing so, I exploit the

fact that when a firm files to go public, it has to provide five years’ worth of audited (albeit

coarse) historical financial statement data. Then, on a firm-by firm basis, each year’s financial

14

statement data was matched with the first pre-money valuation following the fiscal year-end, as

long as the valuation date was less than a year beyond the fiscal year-end (see fig. 3).

Equity valuations and financial statement data at and after the IPO filing were obtained

from sources including www.siliconinvestor.com and www.sec.gov. In addition to the IPO

filing, I used three post-IPO valuation dates, namely three months after the first, second and third

fiscal year-ends following the IPO offering date (if an IPO occurred, since not all biotech firms

in the sample that filed to go public successfully accomplished an offering). These post-IPO

equity valuations, the latest of which is on 3/31/03, were matched with firms’ financial

statements from the preceding fiscal year, the latest of which is 12/31/02. I required that a firm’s

cash balance, SG&A expense, and R&D expense be positive for the financial statement data to

be usable.4 The results of the sample selection process were 458 valid firm-year pre-IPO

observations, 96 at-IPO observations, and 441 post-IPO observations covering 186 biotech firms

over the period 1992–2003.

The strengths of this database are that it is large; not unduly clustered in time; it is rich in

financial statement data; and it contains a wide variety of longitudinal financing points (Series A

through Series F and beyond, successful and unsuccessful IPO filings, and post-IPO equity

valuations). The weaknesses of the database are twofold. First, all of the IPOs filed are filed

between 9/29/95 and 12/14/01, which is only a six-year period. This period could be lengthened,

but at the significant cost of purchasing hard copy S-1 or 424B documents from commercial data

sources. Second, the database does not contain private equity valuations or financial statement

data for firms that did not file to go public because they went bankrupt, merged, or chose to

remain private. While not filing to go public is not synonymous with failure, it is likely more

associated with failure than is going public. Thus, if financial statement data are more value

relevant for companies that succeed than those that fail, the fact that the private equity database

is restricted to those firms that filed to go public may result in an upward bias in the measured

pre-IPO value relevance of accounting information. Moreover, if the impact of selection bias is

larger the further away is the valuation date from the IPO filing selection date, then the bias will

be larger the earlier is the valuation date in the firm’s life. These sample selection issues may

therefore work to simultaneously increase the intercept and decrease the slope of the pre-IPO

4 A biotech firm that has no cash, or no SG&A, or no spending on R&D is highly unusual. Approximately 2% of observations had were deleted as a result of these restrictions.

15

portion of VA/(VA + VFIO), the fraction of equity value that comes from net assets in place

relative to future investment opportunities (fig 4.). If so, then the empirical tests will be biased in

favor of finding that accounting is value relevant in private equity markets, particularly in a

firm’s earliest financing rounds [because the average pre-IPO-measured value relevance of

VA/(VA + VFIO) will be higher], but biased against finding that value relevance increases as a

function of firm maturity within the private equity market (because the slope is lower).

4.3 Descriptive statistics Table 1 lists and defines the variables used in the study. There is a significant amount of

valuation information contained in, and/or computable from, Recap’s private equity market

valuation database (panel A). The public equity market data is standard (panel B). There is a

large amount of pre-IPO accounting information reported in the S-1 and 424B documents that

firms file with the SEC when they register to go public (panels C and D).

Table 2 provides descriptive statistics on the 458 private equity financings done by the

sample of 186 U.S. biotechnology firms. The financings are spread out over the ten-year period

1992–2001 (panel A). Most biotech firms file to go public quite rapidly, primarily between four

and six years after they were founded (panel B). The IPO filing date for the median firm occurs

during preclinical testing/Phase I trials (fig. 2), although a minority of firms takes much longer to

go public. The majority of sample companies are in one of two subsectors of biotechnology—

pharmaceutical preparations (SIC 2834) or commercial, physical and biological research (SIC

8731). Despite this concentration, sample firms comprise a total of 22 different 4-digit SIC

codes. Of firms, 46% had their headquarters at the time of the IPO filing in California and 14%

were located in Massachusetts. Such headquarter clustering arises because California and

Massachusetts contain many top private and public universities with star scientists on their

faculties (Zucker, Darby and Brewer, 1998).

Table 3 reports percentiles for the private equity financing observations.5 Almost every

variable is highly right-skewed. For example, the 5th, 50th and 95th percentiles of pre-money

valuations are $3.9 million, $31 million, and $163 million, respectively (panel A). The 5th, 50th

and 95th percentiles of end-of-fiscal-year cash are $0.1 million, $2.7 million, and $20 million,

respectively (panel B), and the 5th, 50th and 95th percentiles of annual R&D spending are $0.4

5 Equivalent data at and after the IPO filing are available from the author on request.

16

million, $3.1 million, and $12 million, respectively (panel C).6

Panels B and C of table 3 highlight the effects that the biased accounting for intangibles

have on firms’ aggregate financial statement data. Of firms, a mere 3% have positive core

income (defined as revenues less cost of sales, SG&A and R&D expenses), and 22% of firms

have negative total shareholder equity. This is due to the fact that biotech firms’ production

functions are very R&D-intensive and that U.S. GAAP requires that R&D expenditures (and

other intangibles such as advertising and branding) must be expensed, even though a large part of

those expenses are in reality assets.

The median values of key variables measured within and across private and public equity

markets are reported in table 4. With the exception of financings coded as OTHER, table 4 is

organized from left to right in the normal sequence followed by a firm funded by private equity,

beginning with the Series A round and culminating in the IPO filing. OTHER financings are

financings that were not explicitly identified by Recap as being “Series A,” “Series B,” etc.

Such valuations occur anywhere in the firm’s pre-IPO life, whereas Series C financings always

follow Series B financings, etc.7 Although all firms follow the Series A, B, C, etc. sequence,

some go public as early as after their Series A round, while others go public after their Series G

or H round. Some firms file to go public but end up withdrawing their registration.

As would be expected, the time between a firm’s founding and when it undertakes a

particular financing round (AGE) increases as the funding Series increases.8 The median gap

between the financing date on which the pre-money valuation is established and the end of the

previous fiscal year from which the financial statement data is taken is less than five months, and

is smallest for Series A and largest for the IPO. This helps mitigate the concerns that financial

statement data is stale by the time the firm’s current round of financing is undertaken, and is

staler the earlier the round. It is also the case that pre-money valuations, the capital raised, the

firm’s cash balance, and its revenues and R&D expenditures all increase from Series A to IPO,

though at what appear to be decreasing rates.

6 Untabulated analysis reveals that this skewness is also present at each financing round, as well as in the pooled data reported in table 3. 7 OTHER valuation points include investments of common equity, private placements, and debt-related financing such as bridge and convertible notes. 8 One exception is that the median AGE of Series B financings is slightly less than that of Series A financings. Closer examination of this anomaly indicated that this was not due to data errors, but arose by chance because the procedure by which valuations and financial statement data are combined does not guarantee that AGE will be increasing in the Series level. It is also the case that sometimes a firm’s first financing is labeled as Series B.

17

Table 4 also reveals that round-to-round returns earned by equity holders decline from

67% at the Series B round to 28% at Series E and 3% at Series ≥ F. The declining trend is

broken by the very large 86% median return that is in theory earnable by equity holders at the

IPO filing. The 86% figure is consistent with the sawtooth proposition advanced in section 3.1

that predicts that firms that file to go public will experience a decrease in VA/(VA + VFIO).

Although an increase in equity value does not necessitate a decrease in VA/(VA + VFIO), because

VA could increase proportionately more than VFIO, it seems more likely that the lion’s share of

any increase in the equity values of intangible-intensive young biotech firms in the round prior to

their filing for an IPO will come from an increase in the value of their investment opportunities

rather than their actual assets in place.

5. Regression methods This section defines the value relevance of financial statements, explains why

components of firms’ financial statements rather than book equity and net income are used to

determine value relevance, and explains why a log-linear specification is employed as the main

regression model.

5.1 Value relevance

Financial statements are said to be value relevant if they are reliably associated with

equity prices or returns. Following many other studies, I define the degree of the value relevance

of financial statements using the adjusted R2 statistic—specifically, the adjusted R2 that can be

uniquely attributed to financial statements in a regression of equity values on financial statement

data and control variables.9 However, mindful of the inferential dangers of comparing adjusted

R2s across samples (Goldberger, 1971; Brown, Lo and Lys, 1999; Gu, 2002), I conduct auxiliary

analyses to test whether such dangers are significant in my setting. I conclude that they are not.

5.2 Financial statement data used as explanatory variables The regressions that I estimate model firms’ equity values as a function of the major

components of their book equity and net income rather than their book equity and net income per

9 Studies by Harris, Lang and Moller (1994), Collins, Maydew and Weiss (1997), Francis and Schipper (1999), Lev and Zarowin (1999), and Ball, Kothari and Robin (2000) all use adjusted R2 to measure value relevance.

18

se. Zhang (2000) demonstrates analytically that accounting conservatism combined with rapid

growth in intangible assets can dramatically distort the associations between aggregate financial

statement data and equity value. For example, if the firm’s investments in intangible assets are

sufficiently intense, reported losses can be negatively associated with equity values. Hand

(2003b) confirms the predictions of Zhang’s model for intangible intensive Internet companies.

Hand suggests that one solution to the problem of distorted relations between equity values and

aggregate financial statement data for intangible intensive firms is to replace book equity and net

income with their key components—individual or major categories of assets, liabilities,

revenues, and expenses. This substitution prevents the associations between intangible

assets/expenses and equity value from contaminating the associations between tangible

assets/expenses and equity value. In essence, using individual financial statement data items

rather than aggregated book equity and net income removes the restriction that the accounting for

all assets, liabilities, revenues, and expenses be equally biased or unbiased. For example, U.S.

GAAP requires that R&D be expensed even though economically it is an asset (Lev and

Sougiannis, 1996). As a result, if R&D expense is forced into the computation of net income,

the coefficient on net income will be a blend of a positive marginal association between R&D

and equity value arising from R&D being an asset, and a negative marginal association between

cost of sales or other true expenses and equity value. I therefore use the components approach

because young biotech firms make highly intensive and rapidly growing investments in R&D

and therefore have significantly biased financial statements.

The balance sheet data used are three primary components of total shareholder equity—

cash, noncash assets, and long-term debt. The definitions of assets and liabilities and the

relations observed between assets, liabilities, and firms’ equity values in public markets lead to

the expectation that the coefficients on cash and noncash assets in both the private and the public

equity markets will be positive and that the coefficient on long-term debt will be negative. The

income statement data used are the main components of net income for biotech firms—revenues,

cost of sales, SG&A costs, and R&D expense. I expect to observe a positive coefficient on

revenues to the extent that revenues are not entirely transitory, and a positive coefficient on R&D

because the bulk of the benefits from R&D emerge in future periods. I expect to observe a

negative coefficient on cost of sales, recognizing that because cost of sales is only recorded for

product sales and young biotech firms recognize revenues from several sources beyond product

19

sales (collaborations, contracts, grants, licenses, and research), few firms will report positive cost

of sales. I make no sign prediction on SG&A because SG&A is a mixture of period expenses

that would be expected to be negatively related to equity value such as the rent on the firm’s

facilities, and costs that provide future benefits such as salaries for senior management and key

scientific personnel that would be expected to be positively related to equity value.

In addition to financial statement items, most regressions include indicator variables

covering the major financing rounds that are identified in the Recap database (Series A through

Series ≥F, and OTHER), and the years 1992–2001. The financing round indicators help mitigate

the potential selection bias arising from the fact that only firms that file for an IPO are included

in the sample (see section 4.3) because selection bias is likely to be more severe across rounds

rather than within rounds. Financing round indicators also help control for discount rates that

decline as firms mature (table 4). Year indicators control for time-dependent economy-wide

interest rates and other macroeconomic factors.

5.3 Regression models

The economics and accounting of biotech companies are likely to differ substantially

from the assumptions made in equity valuation models such as Ohlson (1995). For example, the

Ohlson (1995) model assumes that accounting is unbiased. This is certainly not the case for

biotech firms (see table 3), and Zhang’s model of equity valuation when accounting is

conservative predicts that there will be nonlinearities in the relations between equity values and

financial statement data. In addition, although the Ohlson model does not per se rule out positive

future NPV opportunities (Ohlson, 2000), neither does it accommodate firms’ investment

opportunity sets in any clear-cut manner. As firms’ investment opportunity sets are by definition

real options (Myers, 1977), any relations between current financial statement data and firms’

investment opportunity sets will likely be nonlinear.

Because of the high likelihood of nonlinearities between equity values and financial

statements, my main empirical tests employ a log-linear regression specification. I subsequently

show that inferences reached using a log-linear model are robust to alternative regression

specifications, including rank regressions. A log-linear model between equity value and

financial statement data implies that a firm’s pre-log equity value is a Cobb-Douglas type

production function of its pre-log financial statement data. Although log-linear models have

20

been employed extensively in economics, particularly for valuing R&D (Hall, 1993, 2000; Hall,

Jaffe and Trajtenberg, 2001), they are rare in accounting and finance.10 In addition to

accommodating nonlinearities, log-linear models are econometrically robust because the log

transforming of the dependent and independent variables substantially dampens the influence of

anomalous observations or outliers, and typically yields a greater degree of homoscedasticity in

regression residuals.11 These are significant concerns for biotech firms because biotech firms’

equity market values, net income, R&D, revenues, etc. are highly skewed (table 3).

The regression models include financial statement data from only the most recent fiscal

year immediately prior to the valuation date. Including earlier financial statement data and/or

instruments for expected future net cash flows, particularly as the firm matures and more years of

historical financial statement data exists, could expand such a simplistic view. However, I limit

myself to financial statement data for the year immediately preceding the valuation date for

reasons of parsimony and data availability: past financial statement data are highly correlated

with current financial statement data, and direct measures of expected future net cash flows such

as analyst forecasts are unavailable for pre-IPO firms.12

Each variable Z in the log-linear regressions is transformed using:

LZ = loge[Z + 1] if Z ≥ 0, where Z is expressed in $000s. (1)

Equation (1) is information-preserving in the sense of being monotone and one-to-one. The

addition of $1,000 to Z ensures that LZ is zero when Z is zero. Negative values of Z do not arise

10 Kaplan and Ruback (1995) and Berger, Ofek and Swary (1996) are two infrequent instances of the use of log-linear models in valuation contexts in finance. Nicholson, Danzon and McCullough (2002) explore the impact of deals on biotech venture capital and IPO valuations using a log-linear model, as do Stuart, Hoang and Hybels (1999) in their examination of the impact of interorganizational networks on growth and valuation. In the accounting arena, Ye and Finn (2000) motivate a log-linear model of firms’ equity market values, book equity and net income by demonstrating that if the log of one plus the return on equity follows an AR(1) process, and net dividends are zero, then equity market value emerges as a multiplicative function of book equity and net income. Beatty, Riffe and Thompson (2001) derive a log-linear valuation model under the assumptions that stock valuation is first degree homogenous in underlying valuation drivers, that accounting constructs measure such valuation drivers with multiplicative measurement error that is conditionally lognormal, and that the unconditional distribution of stock values is either diffuse or lognormal. 11 To finesse the reasonable concern that a minority of the data drives the magnitude and/or significance of parameter estimates, most researchers who apply OLS regression to nonlogged data first identify and then winsorize or delete outliers. This potentially ad hoc process is all but unnecessary within a log-linear model because the log transformation dramatically dampens the values of previously extreme observations. 12 Untabulated tests indicate that lagged financial statement data are rarely incrementally associated with current period equity values.

21

in this study. Unreported statistics indicate that the log transformation dramatically reduces the

right-skewness of the raw data shown in table 3, and achieves greater homoscedasticity in the

regression residuals relative to most models that are not log-linear (e.g., ranks of non-log

transformed data also achieve more homoscedasticity in regression residuals).

6. Empirical tests

This section reports the results of a series of empirical tests. I begin by estimating

regressions in which all available observations are pooled within a given market. This provides a

high-level assessment of the value relevance of biotech firms’ financial statements in private and

public equity markets. I then directly test the sawtooth hypothesis by estimating value relevance

regressions financing round by financing round as firms mature within and across markets, being

careful to address the inferential dangers inherent in comparing adjusted R2s across samples.

The section concludes with tests of the propositions that the underlying drivers of the sawtooth

hypothesis—the fraction of equity attributable to future investment opportunities, and the

probability that the marginal investor in the firm’s stock is financially sophisticated—are

themselves sawtooth functions of firm maturity.

6.1 Univariate correlations

As background, table 5 reports the Pearson correlations between firms’ log-transformed

pre-IPO pre-money valuations LPREMV and financial statement data (panel A, n = 458), and

firms’ log-transformed post-IPO equity market values LMVE and financial statement data (panel

B, n = 441). Table 5 includes all available observations pooled across private equity financing

rounds and public equity end-of-fiscal-year dates, except for IPO filing observations (n = 96).

The correlations indicate that both private and public equity values are reliably highly positively

correlated with firms’ cash assets, noncash assets, long-term debt, revenues, SG&A expenses,

and R&D costs. There are also high correlations among several accounting items.

6.2 Regressions estimated using observations that are pooled within a given market

Table 6 reports the results of estimating value relevance regressions where all available

observations are pooled within a given market. In each regression, an intercept, year indicators,

and series indicators are included but for the sake of compactness are not reported. Valuations at

22

the IPO filing are not included in any of the regressions in order to unequivocally separate

private equity valuations from public equity market valuations.

Three models are estimated for each type of market. Models 1 and 4 are the primary

regressions, in which the dependent variables are LPREMV and LMVE, respectively, and the

financial statement independent variables are log transformed per equation (1) of section 5.3.

Models 2–3 and 5–6 report the results of specification checks where unscaled and nonparametric

regressions are estimated on the same data as models 1 and 4. In models 2 and 5 the dependent

and independent variables are as reported (i.e., are not log transformed), and in models 3 and 6

the dependent and independent variables are the ranks of the data used in models 2 and 5.

From models 1 and 4, it can be seen that firms’ equity values are related to financial

statements in very similar ways in private and public markets. First, untabulated tests of the

equality of the estimated coefficients (elasticities) on financial statement variables in these two

models indicate that of the 12 estimated coefficients on financial statement data, only the

coefficient on cash differs across markets, being reliably higher in public equity markets (p-value

< 0.01). Second, conforming to economic expectations, firms’ equity values are reliably

positively related to their most recent fiscal year-end cash balances, noncash assets and R&D

spending in both models 1 and 4. The strong significance of the coefficients on R&D and cash is

consistent with the prominence that biotech venture capitalists accord to R&D and cash in both

the survivability and success of biotech firms. The estimated coefficients on cash, noncash

assets and R&D are also very similar in size across markets.

Third, equity values are not significantly related to revenues, most likely because of the

somewhat transitory nature of many biotech revenues. Few biotech companies generate

substantial amounts of recurring product revenues until their drugs have passed through the FDA

hurdles, which can be as much as 15 years into the life of a biotech firm (see fig. 1). Instead,

young biotech firms generate revenues from fixed-period collaborations, contracts, grants,

licenses, and research. One result of the paucity of product sales is that only a minority of firms

has a cost of sales, which perhaps accounts for the lack of significance on the coefficient on cost

of sales.

Fourth, SG&A expense is not reliably related to firms’ equity values. I suggest that this

is likely because SG&A contains a mixing of period expenses that would be expected to be

negatively related to equity value, such as the rent on the firm’s facilities, and costs that provide

23

future benefits, such as salaries for senior management and key scientific personnel, that would

be expected to be positively related to equity value.

Finally, conforming to economic expectations, firms’ equity values are reliably

negatively related to long-term debt. The reliably negative coefficient on long-term debt

dampens the potential criticism that the positive coefficients observed on cash, noncash assets,

revenues and R&D are illusory or overstated because such variables merely capture scale effects

(Christie, 1987; Barth and Kallapur, 1996; Lo and Lys, 2000). Were that criticism valid, the

positive univariate correlations between equity values and long-term debt (table 5, panels A and

B) would be likely to lead to positive coefficients on long-term debt, not the reliably negative

coefficients observed in models 1 and 4.

The inferences obtained from estimating log-linear regressions (models 1 and 4) are

generally robust to estimating unscaled regressions (models 2 and 5) and rank regressions

(models 3 and 6). For example, of the 24 regression coefficients in models 2–3 plus models 5–6,

only those on long-term debt in model 2 and SG&A in model 5 are different from those obtained

in the log-linear regressions. In model 2, the estimated coefficient on long-term debt is not

reliably negative (while it is in models 1 and 3–6), and in model 5 the estimated coefficient on

SG&A is reliably negative (while it is insignificantly different from zero in models 1–4 and 6).

Examination of the regression diagnostics for models 2 and 5 indicates that these differences

arise from the undue influence of a few outliers. Untabulated supplementary regressions also

show that the results reported in table 6 are robust to redefining the dependent variable as equity

value less the amount of cash on hand, or as equity value less the book value of equity, and

restricting private equity observations to those for which the financing round was led by venture

capital investors.

6.3 Regressions estimated on a financing-round-by-financing-round basis

The central hypothesis of this paper is that sawtooth dynamics in the maturing of firms’

investment opportunity sets and in the financial sophistication of the marginal investor in firms’

stock combine to create a sawtooth pattern in the value relevance of financial statement data

within and across private and public equity markets (see fig. 1). Specifically, the value relevance

of financial statements is predicted to rise as a firm matures toward an IPO, fall at the IPO, and

then increase again after the IPO.

24

Table 7 reports the results of maturity-based regressions that directly test this hypothesis.

Firm maturity is measured in private equity markets by financing round, and in public equity

markets in yearly intervals beginning three months after the end of the fiscal year following the

IPO.13 I refer to these maturity-based regressions as round-by-round regressions. Compared to

pooled regressions, I expect round-by-round regressions to mitigate the impacts of selection bias

(if any), because round-by-round regressions control for a key factor associated with the

likelihood of filing for an IPO, namely the firm’s age (Cochrane, 2001). I also expect round-by-

round regressions to mitigate the time-varying effects of not controlling for risk.14

The regressions estimated in table 7 do not contain all the financial statement variables

shown in table 5’s correlation matrices. Mindful of the small sample sizes that can arise when

pooled data is divided into subsamples (particularly at the youngest stages of a firm’s life) and

the collinear relations among some financial statement variables, I include only those financial

statement variables whose coefficient estimates in the pooled regressions reported in table 6 are

significantly nonzero—namely cash, noncash assets, long-term debt, and R&D expense.15

Revenues, cost of sales, and SG&A were therefore not included in the round-by-round analysis.

Results of untabulated regressions that include these variables indicate that they are only

infrequently significantly different from zero, in contrast to the variables included in table 7.

Several of table 7’s regression results deserve highlighting. First, consistent with the

sawtooth hypothesis, the value relevance of financial statements, as measured by the adjusted R2

that is uniquely attributable to financial statements, increases as firms mature toward an IPO,

falls at the IPO, and then increases again after the IPO. Figure 4 visually depicts this result.

Second, the sawtooth in value relevance does not appear to be due to scale or sampling variation.

If the sawtooth in value relevance were due to scale, then value relevance should be positively

correlated with the coefficient of variation in the scale factor, such as LPREM, LMVE, or

LCASH (Brown, Lo and Lys, 1999, 2002). However, the Spearman correlations between the

adjusted R2 that is uniquely attributable to financial statements and {LPREM, LMVE} and

LCASH are 0.04 (one-tailed p-value > 0.90) and –0.42 (one-tailed p-value = 0.11), respectively.

Alternatively, if the pattern in value relevance were due to the sample variances of the financial 13 As a result, table 7 only includes private equity observations that were explicitly identified in the Recap database as being “Series A,” “Series B,” etc. 14 Evidence that risk varies systematically with firm maturity is found in table 4, where the round-by-round returns earned by equity holders declines monotonically from 67% at the Series A round to 3% at the Series ≥F round. 15 In addition, parsimony led me to not include LLTD or year indicators in the Series A round regression.

25

statement variables, then value relevance should be positively correlated with the standard

deviation of the regression residuals (Gu, 2002). However, the Spearman correlation between

the adjusted R2 that is uniquely attributable to financial statements and the standard deviations of

round-by-round regression residuals is –0.14 (one-tailed p-value > 0.50).

Third, as measured by the p-values on the F-statistics that test the hypothesis that the

coefficients on all four financial statement variables are zero, financial statements are value

relevant at every stage of the firm’s life after founding, even at the earliest, or Series A, financing

round, where the p-value of the F-statistic is 0.03. However, as indicated in section 4.2, there

may be a selection bias at work, because the dataset does not contain private equity valuations or

financial statement data for firms that did not file to go public.

The same caveat applies to the fourth result of note in table 7, namely the finding that

financial statements explain over one-third of the cross-sectional variance in equity values in

each market. The average adjusted R2 that is uniquely attributable to financial statements is 40%

in private equity markets versus 35% in public equity markets. However, if there is a selection-

based upward bias in value relevance at work, then the average value relevance of financial

statements in private equity markets as measured by my dataset will be upward biased.

Fifth, there is evidence that the sawtooth pattern in value relevance stems from sawtooth

dynamics in both the maturing of firms’ future investment opportunities, and in the probability

that the marginal investor in firms’ stock is sophisticated. To assess the maturing of firms’

investment opportunities, I employ a commonly used proxy for the relative importance of the

investment opportunity set, namely TA/(EV + LTD), the ratio of total assets to the market value

of the firm (Smith and Watts, 1992; Gaver and Gaver, 1993).16 The second to last row of table 7

shows that the median value of TA/(EV + LTD) increases while the firm is in the private equity

market, declines at the IPO filing, and increases to a higher level after the IPO. Moreover, the

Spearman correlation between the adjusted R2 that is uniquely attributable to financial statements

and the median value of TA/(EV + LTD) is 0.63 (one-tailed p-value = 0.03).

Testing for the presence of a sawtooth in the probability that the marginal investor in

firms’ stock is sophisticated is harder because of a lack of data, particularly prior to the IPO.

16 Adam and Goyal (2003) evaluate four proxies for a firm’s investment opportunity set: the market-to-book assets ratio, the market-to-book equity ratio, the earnings-price ratio, and the ratio of capital expenditures to the net book value of PP&E. They conclude that the market-to-book assets ratio is the best variable to proxy for investment opportunities. Based on this, I use (the inverse of) the market-to-book assets ratio as my proxy.

26

One piece of information that becomes available once the firm begins to trade publicly is the

percentage of shares held by institutional shareholders. This variable that has been used in

several studies as a proxy for the probability that the marginal investor in a firm’s stock is

sophisticated (Hand, 1990; Bartov, Radhakrishnan and Krinsky, 2000; Balsom, Bartov and

Marquart, 2002). I report the median percentage of shares held by institutions in the last row of

table 7. Consistent with the prediction that the probability of a sophisticated marginal investor

rises as the firm matures within the public equity markets, the median percentage of shares held

by institutions increases from 0.22 to 0.28 to 0.31 over the three years following the IPO.17

Finally, I note that the value relevance of financial statements at the IPO filing, as

measured by the adjusted R2 that is uniquely attributable to financial statements, is a mere 12%,

lower even than the 19% observed at the Series A financing round. This is consistent with the

view that unsophisticated investors lower the value relevance of financial statements at the IPO

point, reinforcing the simultaneous decline in value relevance arising from the increase in the

importance of firms’ future investment opportunities (see section 3.3). The low value relevance

of biotech firms’ financial statements at their IPO is consistent with the results of Kim and Ritter

(1999), who find that valuing IPOs in general on the basis of the price-to-historical-earnings,

price-to-historical-sales and market-to-book ratios of comparable firms is of very limited use.18

6.4 Caveats and limitations of the paper

Beyond the standard concern of how well the results found here for biotech firms

generalize to the population of private firms, this paper has three main limitations—differences

between private and public equity markets that have not been controlled for, the uncertain

impacts of selection bias, and the exclusion of nonfinancial variables.

Private equity markets differ from public equity markets along many dimensions. The

focus of this paper—the first research in the area—has been on the investment opportunity set

and the sophistication of the marginal investor. However, other factors such as regulation,

liquidity, price-setting mechanisms, and degree of information asymmetry between managers 17 The median of 0.28 two years after the IPO is reliably larger than the median of 0.22 one year after the IPO (Wilcoxon p-value < 0.01). The median of 0.31 three years after the IPO is not reliably larger than the median of 0.28 one year after the IPO (Wilcoxon p-value = 0.47). 18 In univariate regressions of IPO firms’ price-to-historical-earnings, price-to-historical-sales and market-to-book ratios on the median price-to-historical-earnings, price-to-historical-sales and market-to-book ratios of comparable firms, Kim and Ritter (1999, table 4) report adjusted R2 of between 5% and 8.5% and mean absolute valuation errors of between 31% and 52%.

27

and investors differ across private and public equity markets and are therefore likely to affect the

value relevance of financial statements. Selection bias arises because private equity valuations

are only included in the study if the firm is successful enough to file for an IPO. I attempted to

control for this by estimating round-by-round regressions, since the likelihood of a firm filing for

an IPO increases with its age (Cochrane, 2001). In terms of nonfinancial information, I chose

not to collect and include data such as the number of patents held by the firm, the stages of the

firm’s patents through the FDA approval process, or the quality of the firm’s key scientists,

simply because of the very high costs involved. As such, I cannot rule out the possibility that the

inferences I have drawn regarding the value relevance of financial statement data are overstated

because the financial statement data are acting as proxies for the more value relevant

nonfinancial information. Addressing all three limitations just discussed remains a promising

avenue for future research.

7. Conclusions

This study has contributed to the value relevance literature by examining the value

relevance of financial statements within private equity markets, and across the transition from

private to public equity markets. Both settings previously unexamined but are economically

important stages in the life of a firm. My results demonstrate that, at least for biotech companies,

financial statements are as qualitatively and quantitatively value relevant in private equity

markets as they are in public equity markets. In each market, equity values are positively related

to R&D expense, cash, and noncash assets; unrelated to revenues and SG&A expense; and

negatively related to long-term debt. And in each market, financial statements explain over one-

third of the cross-sectional variance in equity values in each market.

Most importantly, the paper developed, tested and found evidence supportive of the

hypothesis that interactions between the maturing of firms’ investment opportunity sets and the

dynamics of the probability that the marginal investor is sophisticated create a nonlinear

‘sawtooth’ pattern in the value relevance of financial statements measured within and across

markets. For firms that file to go public, the value relevance of their financial statements as

predicted rises as firms mature toward an IPO, falls at the IPO, and increases again after the IPO.

28

References Adam, T., and V.K. Goyal, 2003. The investment opportunity set and its proxy variables:

Theory and evidence. Working paper, Hong Kong University of Science & Technology.

Ball, R., Kothari, S.P., and A. Robin, 2000. The effect of international institutional factors on properties of accounting earnings. Journal of Accounting & Economics 29: 1-51.

Balsam, S., Bartov, E., and C. Marquardt, 2002. Accruals management, investor sophistication, and equity valuation: Evidence from Form 10-Q filings. Journal of Accounting Research 40: 987-1012.

Barberis, N., and R.H. Thaler, 2002. A survey of behavioral finance. Working paper, University of Chicago.

Barth, M.E., and S. Kallapur, 1996. The effects of cross-sectional scale differences on regression results in empirical accounting research. Contemporary Accounting Research 13: 527-567.

Bartov, E., Radhakrishnan, S., and I. Krinsky, 2000. Investor sophistication and patterns in stock returns after earnings announcements. The Accounting Review 75: 43-64.

Beatty, R., Riffe, S., and R. Thompson, 2001. The log-linear form of stock value models based on accounting information. Working paper, Southern Methodist University.

Berger, P.G., Ofek, E., and I. Swary, 1996, Investor valuation of the abandonment option. Journal of Financial Economics 42, 257-287.

Bernard, V.L., 1989. Capital markets research in accounting during the 1980s: A critical review. In: Frecka, T.J. (Ed.), The State of Accounting Research As We Enter The 1990s. Urbana, IL: University of Illinois at Urbana-Champaign.

Bernard, V.L., and J.K. Thomas, 1989. Post-earnings-announcement drift: Delayed price response or risk premium? Journal of Accounting Research 17 (Supplement): 1-48.

Bernard, V.L., and J.K. Thomas, 1990. Evidence that stock prices do not fully reflect the implications of current earnings for future earnings. Journal of Accounting & Economics 13: 305-340.

Black, E.L., 1998. Life cycle impacts on the incremental value relevance of earnings and cash flow measures. Journal of Financial Statement Analysis, Fall: 40-56.

Brav, A., and P.A. Gompers, 1997. Myth or reality? The long-run underperformance of initial public offerings: Evidence from venture capital and nonventure capital-backed companies. Journal of Finance 52: 1791-1822.

Brown, S., Lo, K., and T. Lys, 1999. Use of R-squared in accounting research: Measuring changes in value relevance over the last four decades. Journal of Accounting & Economics 28: 83-115.

Brown, S., Lo, K., and T. Lys, 2002. Erratum to “Use of R-squared in accounting research: measuring changes in value relevance over the last four decades.” Journal of Accounting & Economics 33: 141.

29

Christie, A., 1987. On cross-sectional analysis in accounting research. Journal of Accounting & Economics 9: 231-258.

Cochrane, J.H., 2001. The risk and return of venture capital. Working paper, University of Chicago.

Collins, D.W., Maydew, E.L., and I.S. Weiss, 1997. Changes in the value relevance of earnings and book values over the past forty years. Journal of Accounting & Economics 24: 39-67.

Cramer, J. S., 1987. Mean and variance of R2 in small and moderate samples. Journal of Econometrics 35: 253-266.

Daniel, K., Hirshleifer, D., and A. Subrahmanyam, 1998. Investor psychology and security market under- and over-reactions. Journal of Finance 53: 1839-1886.

Daniel, K., Hirshleifer, D., and S.H. Teoh, 2002. Investor psychology in capital markets: evidence and policy implications. Journal of Monetary Economics 49: 139-209.

Darby, M.R., Liu, Q., and L.G. Zucker, 1999. Stars and stakes: The effect of intellectual human capital on the level and variability of high-tech firms' market values. Working paper, UCLA.

Ely, K., Simko, P., and L.G. Thomas, 2003. The usefulness of biotechnology firms’ drug development status in the evaluation of research and development costs. Journal of Accounting, Auditing and Finance, Winter (forthcoming).

Fenn, G.W., and N. Liang, 1995. The Economics of the Private Equity Market. Washington D.C.: Board of Governors of the Federal Reserve System.

Francis, J., and K. Schipper, 1999. Have financial statements lost their relevance? Journal of Accounting Research 37: 319-352.

Gaver, J.J., and K.M. Gaver, 1993. Additional evidence on the association between the investment opportunity set and corporate financing, dividend and compensation policies. Journal of Accounting & Economics 16, 125-160.

Goldberger, A., 1991. A Course in Econometrics. Boston: Harvard University Press.

Gompers, P.A., 1995. Optimal investment, monitoring, and the staging of venture capital. Journal of Finance 50: 1461-1490.

Gompers, P.A., and J. Lerner, 1999. An analysis of compensation in the U.S. venture capital partnership. Journal of Financial Economics 51: 3-44.

Gompers, P.A., and J. Lerner, 2000. The venture capital cycle. Cambridge, MA: MIT Press.

Gu, Z., 2002. Cross-sample incomparability of R2s and additional evidence on value relevance changes over time. Working paper, Carnegie Mellon University.

Guo, R., Lev, B., and N. Zhou, 2003. Competitive costs of disclosure by biotech IPOs. Working paper, University of Illinois.

Hall, B.H., 1993. The stock market's valuation of research and development investment during the 1980s. American Economic Review 83: 259-264.

30

Hall, B.H., 2000. Innovation and market value. In Barrell, R., Mason, G. and M. O'Mahoney (Eds.), Productivity, Innovation and Economic Performance, Cambridge: Cambridge University Press.

Hall, B.H., Jaffe, A., and M. Trajtenberg, 2001. Market value and patent citations: A first look. NBER working paper No. 7741.

Hand, J.R.M., 1990. A test of the extended functional fixation hypothesis. The Accounting Review 65: 739-763.

Hand, J.R.M., 2003a. The market valuation of biotechnology firms and biotechnology R&D. In McCahery, J. and L. Renneboog (Eds.), Venture Capital Contracting and the Valuation of High-Technology Firms, Oxford: Oxford University Press.

Hand, J.R.M., 2003b. Profits, losses and the nonlinear pricing of Internet stocks. In Hand, J.R.M. and B. Lev (Eds.), Intangible Assets: Values, Measures, and Risks. Oxford: Oxford University Press.

Harris, T., Lang, M., and H.P. Möller, 1994. The value relevance of German accounting measures: An empirical analysis. Journal of Accounting Research 32: 187-209.

Houlihan Valuation Advisors/VentureOne, 1998. The pricing of successful venture-backed high-tech and life-science companies. Journal of Business Venturing 13: 331-351.

Joos, P., 2002. Explaining cross-sectional differences in market-to-book ratios in the pharmaceutical industry. Working paper, University of Rochester.

Kaplan, S.N., and R.S. Ruback, 1995. The valuation of cash flow forecasts: An empirical analysis. Journal of Finance 50, 1059-1093.

Kaplan, S.N., and P. Stromberg, 2002a. Characteristics, contracts, and actions: Evidence from venture capitalist analyses. Working paper, University of Chicago.

Kaplan, S.N., and P. Stromberg, 2002b. Financial contracting meets the real world: An empirical analysis of venture capital contracts. Review of Economics Studies (forthcoming).

Kim, M., and J. Ritter, 1999. Valuing IPOs. Journal of Financial Economics 53: 409-437.

Kothari, S.P., 2001. Capital markets research in accounting. Journal of Accounting & Economics 31, 105-231.

Lerner, J., 1994a. Venture capitalists and the decision to go public. Journal of Financial Economics 35: 293-316.

Lerner, J., 1994b. The importance of patent scope: An empirical analysis. Rand Journal of Economics 25: 319-333.

Lerner, J., 2000. Venture Capital and Private Equity: A Casebook. New York: John Wiley & Sons.

Lerner, J., 2001. Venture capital and private equity: A course overview (2001 update). Working paper, Harvard Business School.

Lev, B., and J. Ohlson, 1982. Market based empirical research in accounting: A review, interpretations, and extensions. Journal of Accounting Research 20 (Suppl.): 249-322.

31

Lev, B., and T. Sougiannis, 1996. The capitalization, amortization, and value-relevance of R&D. Journal of Accounting & Economics 21: 107-138.

Lev, B., and P. Zarowin, 1999. The boundaries of financial reporting and how to extend them. Journal of Accounting Research 37: 353-385.

Lo, K., and T. Lys, 2000. The Ohlson model: Contribution to valuation theory, limitations, and empirical applications. Journal of Accounting, Auditing and Finance 15, 337-367.

Miller, E., 1977. Risk, uncertainty, and divergence of opinion. Journal of Finance 32: 1151-1168.

Miller, M.H., and F. Modigliani, 1961. Dividend policy, growth and the valuation of shares. Journal of Business 34: 411-433.

Myers, S., 1977. Determinants of corporate borrowing. Journal of Financial Economics 5: 147-175.

Nicholson, S., Danzon, P.M., and J. McCullough, 2002. Biotech-pharmaceutical alliances as a signal of asset and firm quality. NBER working paper 9007.

Ohlson, J.A., 1995. Earnings, equity book values, and dividends in equity valuation, Contemporary Accounting Research, 661-687.

Ohlson, J.A., 2000. Positive (zero) NPV projects and the behavior of residual earnings. Working paper, NYU.

Purnanandam, A.K., and B. Swaminathan, 2003. Are IPOs really underpriced? Working paper, Cornell University.

Rink, D.R., and J.E. Swan, 1979. Product life cycle research: A literature review. Journal of Business Research, 219-247.

Ritter, J.R., 1991. The long run performance of initial public offerings. Journal of Finance 46: 3-28

Ritter, J.R., and T. Loughran, 1995. The new issues puzzle. Journal of Finance 50: 23-51.

Robinson, D.T., and T. Stuart, 2000, Just how incomplete are incomplete contracts? Evidence from biotech strategic alliances. Working paper, University of Chicago.

Shiller, R., 1990. Speculative prices and popular models. Journal of Economic Perspectives 4: 55-65.

Skinner, D., 1993. The investment opportunity set and accounting procedure choice: Preliminary evidence. Journal of Accounting & Economics 16: 407-445.

Smith, C.W., and R.L. Watts, 1992. The investment opportunity set and corporate financing, dividends, and compensation policies. Journal of Financial Economics 32: 262-292.

Smith, K.G., Mitchell, T.R., and C.E. Summer, 1985. Top level management priorities in different stages of the organizational life cycle. Academy of Management Journal 28: 729-820.

Stuart, T.E., Hoang, H., and R.C. Hybels, 1999. Interorganizational endorsements and the performance of entrepreneurial ventures. Administrative Science Quarterly 44: 315-349.

32

Ye, J., and Finn, M., 2000. Nonlinear accounting-based equity valuation models. Working paper, Baruch College.

Yetman, M.H., 2003. Accounting-based value metrics and the informational efficiency of IPO early market prices. Working paper, University of Iowa.

Zhang, X-J., 2000. Conservatism, growth, and the analysis of line items in earnings forecasting and equity valuation. Working paper, University of California, Berkeley.

Zucker, L.G. and M.R. Darby, 1996. Star scientists and institutional transformation: Patterns of invention and innovation in the formation of the biotechnology industry. Proceedings of the National Academy of Sciences, 93(23): 12709-12716.

Zucker, L.G. and M.R. Darby, 1998. Capturing technological opportunity via Japan’s star scientists: Evidence from Japanese firms’ biotech patents and products. Working paper, UCLA.

Zucker, L.G., Darby, M.R., and M.B. Brewer, 1998. Intellectual human capital and the birth of U.S. biotechnology enterprises. American Economic Review 88(1): 290-306.

33

Figure 1

Schematic representation of the hypothesized ‘sawtooth’ pattern in the value relevance of financial statements within and across private and public equity markets

100% Psoph = probability that marginal investor is financially sophisticated Value relevance of financial statements VA/(VA + VFIO) = fraction of equity value attributable to assets in place relative future investment opportunities Founding Series Series Series Series IPO Firm maturity A B C D Note: Fig. 1 is illustrative; it is not drawn to scale. Some firms may go public after a Series B round, others after a Series G round etc.

34

Discovery and Preclinical Testing: The drug development process usually begins with the scientific

discovery of a gene or other biological pathway involved in a disease. Discovery can take 2–10 years. From discovery, a target for therapeutic intervention is established. Preclinical tests are conducted in the lab using individual cells or sometimes animals to evaluate the safety and potential for effectiveness in humans. If the target is determined to be legitimate, the company files an Investigative New Drug (IND) application with the FDA for clearance to begin testing on humans. Even after these first few years of research and testing, however, most new drug candidates will never make it to the market.

Phase I Trials: Human testing begins. The purpose of a Phase I trial is to use a small number of patients to establish basic safety and maximum dosage parameters.

Phase II Trials: This stage of clinical study is much more involved, requiring many months to plan, set up and recruit trial participants. Phase II is conducted on a larger group of patients with the targeted disease to study the efficacy of the drug at various doses and confirm its safety. They typically use blinding and placebo controls to achieve scientifically sound results. Phase II often lasts two years, and sometimes a drug will undergo multiple Phase II trials for different indications (for example, to treat different types of cancer). This may be the most critical phase in terms of sorting winners from losers. As a rule of thumb, drugs that complete Phase II and move on to Phase III have about a 50% success rate of reaching the market, though some studies suggest the rate is higher.

Phase III Trials: These tests are designed with a specific endpoint—a measurable result that clearly demonstrates success in combating the targeted disease. The endpoint must be agreed upon by the FDA as an outcome that will lead to marketing approval. The trial involves a large group from the targeted patient population and uses controls such as double-blinding (neither patient nor doctor knows who is getting a placebo). Multi-center trials are common to show that results are reproducible when administered in different clinical settings. This pivotal phase often lasts two to three years from initial design to study completion, and here again it is common for drugs to undergo more than one Phase III trial for different indications or to support different therapy combinations.

FDA Approval Process: If a drug successfully completes Phase III, the company gathers all of its clinical data and files an application for marketing approval with the FDA. It often takes three to six months just to prepare the application. Another six to twelve months can pass before an FDA advisory panel reviews the application and makes a recommendation. This advisory panel has expertise in the drug's specific area of therapeutic or disease characteristics, and its recommendation for denial or approval is normally followed by the FDA (though another six to twelve months can pass before that happens).

Modified from an article by James Hale (http://www.theonlineinvestor.com/industries.phtml?content=is_bio2)

D is c o v e ry (2 -1 0 ye a rs )

P re c lin ic a l T e s t in g (L a b a n d a n im a l te s tin g )

P h a s e I (2 0 -3 0 h e a lth y v o lu n te e rs u s e to c h e c k fo r s a fe ty & d o s a g e )

P h a s e II (1 0 0 -3 0 0 p a tie n t v o lu n te e rs u s e d to c h e c k fo r e f f ic ie n c y & s id e e f fe c ts )

P h a s e I II (1 ,0 0 0 -5 ,0 0 0 p a tie n t v o lu n te e rs u s e d to m o n ito r re a c t io n s to lo n g - te rm d ru g u s e )

F D A R e v ie w & A p p ro v a l

P o s tm a rk e t in g T e s t in g

0 2 4 6 8 1 0 1 2 1 4 1 6 Y e a rs

S o u rc e : E rn s t & Y o u n g L L P , B io te c h n o lo g y In d u s try R e p o r t C o n v e rg e n c e , 2 0 0 0 .

B io te c h v a lu e c h a inF ig u r e 2

35

Figure 3

Stylized timeline of private equity financing rounds, public equity market valuations, and annual historical financial statements

Codes: = Usable private equity financing round and pre-money valuation = Usable public equity valuation 12/96 12/97 12/98 12/99 12/00 3/01 12/01 3/02 Time Founding Angel [==FY 97 financials==] A Private placement [== FY 98 financials==] B C Option exercise [== FY 99 financials==] IPO filing [== FY 00 financials==] IPO+1 [== FY 01 financials==] IPO+2 Note: Fig. 3 is illustrative. Rounds that are not in a circle or box are not usable in the study.

36

Figure 4

Value relevance of financial statements before, at and after an IPO, measured by the regression adj. R2 uniquely explained by financial statements (solid lines are trendlines). Sample is 186 U.S.

biotech firms, 1992-2003.

0%

25%

50%

75%

A B C D E >=F IPO filing IPO+1 IPO+2 IPO+3

37

Table 1

Valuation and financial statement variables computed for U.S. biotechnology firms, 1992–2003 (n = 186)

Panel A: Valuation variables for private equity market settinga Variable Label Description

Pre-money valuation PREMV Valuation of total equity (common + preferred) before the infusion of private capital in the current financing round. Preferred shares are converted into common stock at the conversion ratios specified in the financing agreements.

Capital investment RAISED Private capital invested in current financing round. Post-money valuation POSTMV PREMV + RAISED. Dilution DIL RAISED ÷ POSTMV. Equity price per share PRICE POSTMV ÷ SHSOUT, where SHSOUT is the

number of common + preferred-as-if-converted-into-common shares outstanding immediately after the capital investment made in the current round. SHSOUT is adjusted for stock splits.

Round-to-round step-up STEPUP PREMV ÷ (POSTMV lagged one valuation round). in valuation Round-to-round return RET (PRICE – [PRICE lagged one valuation round]) ÷ to equity holders [PRICE lagged one valuation round]. Pre-money valuation date VALDT mm/yy date of beginning of the current round. Age of the firm AGE Number of years from founding to VALDT. Panel B: Valuation variables for public equity market settingb Variable Label Description

Equity valuation MVE Value of common equity three months after the fiscal year-end.

Equity price per share PRICE MVE ÷ SHS, where SHS is the number of common shares outstanding three months after the fiscal year-end, adjusted for stock splits after the IPO.

Annual equity return RET (PRICE – [PRICE lagged one year]) ÷ to equity holders [PRICE lagged one year]. Valuation date VALDT mm/yy date of three months after the fiscal year-end. Age of the firm AGE Number of years from founding to VALDT.

38

Table 1 (continued) Panel C: End-of-fiscal year balance sheet variables (whether in private or public market)c Variable Label Description

Total equity TEQ Total shareholder equity + book value of preferred stock not in equity (e.g., redeemable conv. preferred).

Preferred stock PFD Carrying value of all preferred stock, whether classified in shareholder equity or not.

Retained earnings RE Retained earnings (usually “accumulated deficit”). Cash assets CASH Cash + short-term financial investments. Total assets TA Total assets. Noncash assets NCASS TA – CASH. Long-term debt LTD Long-term debt + capitalized long-term leases. Panel D: Annual income statement variables (whether in private or public market)c Variable Label Description

Revenue REV Product + collaborative + contract + grant + license + research revenues.

Cost of sales COS Cost of product sales. SG&A expense SGA G&A + selling & marketing expense components are

provided; else simply the line item Selling, General and Administrative expense. SGA excludes RD.

R&D expense RD Research & development expense. Core (pre-tax) income CI REV – COS – SGA – RD. Net income to common NIC Net income or loss to common shareholders. Revenue growth rate REVGRW REV ÷ [REV lagged one valuation round]. R&D growth rate RDGRW RD ÷ [RD lagged one valuation round]. Financial statement gap FRGAP Years between VALDT and FYE prior to VALDT. Panel E: Statement of cash flows variables (whether in private or public market)c Variable Label Description

Cash from operations CFOPS Net cash flow from or used by operating activities. Cash from investing CFINV Net cash flow from or used by investing activities. Cash from financing CFFIN Net cash flow from or used by financing activities. a Private equity valuation data are from Recombinant Capital’s biotechnology valuation database. b Public equity valuation data are from CRSP. c Pre-IPO filing financial statement data are from those firms in Recombinant Capital’s biotechnology valuation

database whose IPO filings were available online at www.sec.gov (earliest IPO filing = 5/96, most recent = 2/02). Post-IPO financial statement data are from 10-K filings available online at www.sec.gov.

39

Table 2

Descriptive statistics for 458 pre-IPO private equity market financings undertaken by 186 U.S. biotechnology firms, 1992–2001

Panel A: Number of financings, by year Year # financings Year # financings 1992 15 1997 68 1993 27 1998 64 1994 44 1999 54 1995 58 2000 51 1996 72 2001 4 Panel B: Number of years from founding to IPO filing Years # obs. Years # obs. 0 to 1 year 0 7 to 8 years 15 1 to 2 years 4 8 to 9 years 13 2 to 3 years 10 9 to 10 years 5 3 to 4 years 20 10 to 11 years 4 4 to 5 years 42 11 to 12 years 7 5 to 6 years 40 12 to 13 years 2 6 to 7 years 22 13 to 15 years 2 Panel C: 4-digit SICs represented 4-digit SIC SIC industry description # firms % firms 2834 Pharmaceutical preparations 60 32% 8731 Commercial, physical and biological research 46 25% 2836 Biological products (except diagnostic substances) 20 11% 3845 Electromedical and electrotherapeutic apparatus 12 6% 2835 In vitro and in vivo diagnostic substances 11 6% All others Various (representing 17 four-digit SIC) 37 20% 186 100% Panel D: State in which firm’s headquarters were located at time of the IPO filing State # obs. % obs. California 86 46% Massachusetts 26 14% Pennsylvania 11 6% All others (20 states) 63 34% Note: Sample comprises biotechnology firms with pre-IPO pre-money valuations in Recombinant Capital’s biotech valuation database and valid pre-financing financial statement data in SEC IPO filing documents (see fig. 3).

40

Table 3

Statistics on major valuation and financial statement variables for 458 pre-IPO private equity market financings made by 186 U.S. biotechnology firms, 1992–2001

Panel A: Valuation data

Percentile PREMV RAISED POSTMV DIL PRICE STEPUP RET VALDT AGE 100% $ 572 $ 159 $ 597 87% $ 100 132 1,900% 7/01 13.4 95% 163 31 183 61 13 6.7 400 3/00 8.5 75% 64 15 75 35 7.3 2.3 112 12/98 5.2 50% 31 7.4 42 20 5.0 1.5 39 1/97 3.7 25% 13 3.5 20 9.2 3.2 1.1 1.0 6/95 2.5 5% 3.9 0.9 7.5 2.2 1.4 0.8 –24 3/93 1.3 0% 0.6 0.03 1.7 <0.2 0.6 0.3 –80 1/92 0.3 % > 0 100% 100% 100% 100% 100% 100% 76% 100% 100% # obs. 458 458 458 458 458 356 356 458 458 Panel B: Balance sheet and statement of cash flows data

Percentile TEQ PFD RE TA CASH NCASS LTD CFOPS CASH÷TA 100% $ 132 $ 185 $ 17 $ 147 $ 114 $ 39 $ 23 $ 7.3 100% 95% 20 34 –0.4 28 20 10 5.2 –0.3 96 75% 7.0 12 –3.0 10 6.9 2.9 1.2 –1.9 84 50% 2.5 2.5 –7.6 4.9 2.7 1.4 0.3 –3.5 71 25% 0.2 0 –16 2.0 0.8 0.5 0 –6.1 40 5% –3.6 0 –35 0.3 0.1 0.1 0 –13 7.2 0% –22 0 –197 0.03 <0.01 0 0 –37 0.3 % > 0 78% 82% 2% 100% 100% 99% 71% 2% 100% # obs. 458 422 452 458 458 458 458 334 458 Panel C: Income statement dataa

Percentile NIC CI REV COS SGA RD IREV REVGRW RDGRW 100% $ 12 $ 8.2 $ 26 $ 22 $ 26 $ 33 $ 6.6 1,000% 1,000% 95% –0.2 –0.3 10 2.4 5.9 12 0.8 1,000% 400% 75% –1.7 –1.8 1.5 0 2.5 5.9 0.3 225% 138% 50% –3.9 –3.9 0.2 0 1.5 3.1 0.1 60% 62% 25% –6.7 –6.7 0 0 0.8 1.4 0.04 –17% 24% 5% –17 –14 0 0 0.2 0.4 <0.01 –100% –20% 0% –58 –44 0 0 <0.01 0.04 0 –100% –68% % > 0 3% 2% 61% 19% 100% 100% 95% 69% 90% # obs. 458 458 458 458 458 458 350 186 342 Note: Sample comprises biotechnology firms with pre-IPO pre-money valuations in Recombinant Capital’s biotech valuation database and valid pre-financing financial statement data in SEC IPO filing documents (see fig. 3). Dollar amounts are in millions. For variable definitions, see table 1. a REVGRW and RDGRW are winsorized at 1,000%.

41

Table 4

Medians of key variables at valuation points prior to, at, and after the IPO filing for 186 U.S. biotech firms, 1992–2003 Series Series Series Series Series Series IPO Percentile A B C D E ≥F OTHERa filingb IPO+1c IPO+2 IPO+3

AGE (yrs) 2.6 2.5 3.2 4.0 4.8 6.8 3.9 5.7 5.8 6.9 7.8 FRGAP (yrs) 0.18 0.34 0.30 0.34 0.26 0.34 0.18 0.42 0.25 0.25 0.25

PREMV; MVE $ 5.9 15 26 41 51 56 49 110 140 143 114 RAISED $ 5.6 6.0 9.2 8.6 9.9 11 4.8 37 n.app. n.app. n.app. STEPUP n.app. 1.8 1.9 1.5 1.3 1.1 1.5 2.2 n.app. n.app. n.app. RET n.app. 67% 60% 45% 28% 3% 29% 86% n.app. –14% –32% DIL 50% 27% 28% 19% 18% 16% 8% 25% n.app. n.app. n.app. CASH $ 0.2 1.5 2.6 3.6 5.5 9.4 2.6 11.8 40 39 34 NCASS $ 0.4 0.7 1.3 1.9 2.5 2.4 1.9 3.6 8.3 14 17 LTD $ 0 0.2 0.2 0.5 0.4 0.8 0.1 0.5 0.6 0.6 1.1 REV $ 0.03 0 0.04 0.2 1.2 2.1 0.4 3.4 5.9 7.7 9.6 RD $ 0.9 1.5 2.9 4.5 5.3 6.0 2.6 6.8 11 15 17 CI $ –1.2 –2.1 –3.5 –5.4 –6.7 –6.9 –3.0 –5.9 –11 –17 –20

CFOPS $ –1.6 –2.0 –3.6 –4.6 –5.7 –5.3 –2.5 –4.0 –7.9 –13 –14 RDGRWe n.app. 74% 107% 74% 46% 36% 35% 47% n.app. 36% 15% REVGRWe n.app. 44% 62% 67% 105% 38% –3% 93% n.app. 37% 28%

# obs. 33 86 101 79 54 33 73 96 157 151 133 Note: Sample comprises biotechnology firms with pre-IPO pre-money valuations in Recombinant Capital’s biotech valuation database and valid pre-financing financial statement data in SEC IPO filing documents (see fig. 3). Dollar amounts are in millions. For variable definitions, see table 1. a OTHER financing rounds include common equity, private placements, and debt related financings (such as bridge notes and convertible notes). b Includes all IPO filings, whether ultimately successful or withdrawn. c “IPO+n” denotes three months after the nth fiscal year-end following the IPO offering date, if the IPO offering was successfully undertaken. d REVGRW and RDGRW are winsorized at 1,000%.

42

Table 5

Correlations among log-transformed financial statement data and the equity valuations in pre-IPO private equity markets and post-IPO public equity markets for 186 U.S.

biotechnology firms, 1992–2003

Panel A: Pearson correlations for pre-IPO private equity financings (n = 458) LCASH LNCASS LLTD LREV LCOS LSGA LRD

LPREMV 0.63* 0.60* 0.30* 0.32* 0.09* 0.59* 0.67* LCASH 0.44* 0.22* 0.16* –0.03 0.47* 0.59* LNCASS 0.61* 0.48* 0.29* 0.74* 0.69* LLTD 0.34* 0.11* 0.45* 0.45* LREV 0.42* 0.46* 0.32* LCOS 0.36* –0.05 LSGA 0.69*

Panel B: Pearson correlations for post-IPO public equity valuations (n = 441) LCASH LNCASS LLTD LREV LCOS LSGA LRD

LPREMV 0.62* 0.40* 0.13* 0.20* –0.02 0.28* 0.44* LCASH 0.40* 0.19* 0.13* –0.13* 0.31* 0.59* LNCASS 0.42* 0.51* 0.36* 0.69* 0.37* LLTD 0.28* 0.03 0.20* 0.31* LREV 0.43* 0.39* 0.07 LCOS 0.49* –0.33* LSGA 0.35* Note: Sample comprises biotechnology firms with pre-IPO pre-money valuations in Recombinant Capital’s biotech valuation database and valid pre-financing financial statement data in SEC IPO filing documents (see fig. 3). Neither panel contains observations at the IPO filing point. A log transformation is applied to all non-indicator variables Z according to LZ = loge[Z +1] where Z ≥ 0 is in $000s. Table 1 provides definitions of data items prior to their being log transformed. An asterisk denotes that the correlation is reliably nonzero at the 5% level under a two-tailed test.

43

Table 6

Pooled regressions of firm equity value on financial statement data and indicator controls for 186 U.S. biotechnology firms, 1992–2003

Independent Predicted sign Pre-IPO private equity marketa Post-IPO public equity market b variables on coefficient Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

CASH (+) 0.17 3.58 0.35 0.55 1.70 0.54 (7.2)c (16.6) (9.4) (11.4) (6.9) (11.5) NCASS (+) 0.16 2.53 0.17 0.20 0.90 0.21 (4.5) (4.7) (3.2) (4.5) (6.0) (3.8) LTD (–) –0.04 –0.75 –0.14 –0.06 –1.72 –0.20 (–3.3) (–0.9) (–3.7) (–4.5) (–2.0) (–5.3) REV (+) 0.01 0.89 0.05 0.01 0.01 0.05 (1.2) (1.6) (1.4) (0.8) (0.0) (1.2) COS (–) –0.01 –1.54 –0.02 0.01 2.42 0.07 (–0.8) (–1.1) (–0.4) (0.7) (1.7) (1.2) SGA (?) 0.05 0.99 0.09 0.05 –4.67 –0.06 (1.0) (0.9) (1.7) (0.7) (–3.1) (–1.1) RD (+) 0.23 1.24 0.21 0.27 4.51 0.27 (4.2) (2.2) (4.0) (4.3) (5.2) (5.2)

Year indicators included?d Yes Yes Yes Yes Yes Yes Series indicators included?e Yes Yes Yes Yes Yes Yes # observations 458 458 458 441 441 441 Adj. R2 65% 71% 67% 58% 58% 58% Median absolute valuation errorf 37% 39% n.app. 45% 46% n.app. Note: Sample comprises biotechnology firms with pre-IPO pre-money valuations in Recombinant Capital’s biotech valuation database and valid pre-financing financial statement data in SEC IPO filing documents (see fig. 3). Dollar amounts are in millions. For variable definitions, see table 1. Regression intercept and indicator coefficients are estimated but not reported. Equity valuations at the IPO filing are not included in the regressions. a Model 1: Dependent variable is LPREMV, the log-transformed pre-IPO pre-money equity value of the firm per

equation (1) of section 5.3 . Financial statement variables are also all log-transformed. Model 2: Dependent variable is PREMV, the raw pre-IPO pre-money equity value of the firm. Financial statement variables are also raw (i.e., not log-transformed). Model 3: Dependent variable is RPREMV, the rank of PREMV. Financial statement variables are also ranks.

b Model 4: Dependent variable is LMVE, the log-transformed post-IPO equity value of the firm per equation (1) of section 5.3 . Financial statement variables are also all log-transformed. Model 5: Dependent variable is MVE, the post-IPO equity value of the firm. Financial statement variables are also raw (i.e., not log-transformed). Model 6: Dependent variable is RMVE, the rank of MVE. Financial statement variables are also ranks.

c t-statistic relative to a null coefficient value of zero are in parentheses. d Year indicators are for 1993–2003 (e.g., 1993 indicator = 1 if valuation date is in calendar 1993, zero otherwise). e Pre-IPO series indicators are for Series A, B, C, D, E and ≥F (e.g., Series A = 1 if round is Series A, zero otherwise).

Post-IPO series indicators are for IPO+2 and IPO+3. f Valuation error = (Equity value less fitted equity value) divided by fitted equity value.

44

Table 7

Log-linear regressions of equity values on financial statement variables and indicator controls for 186 U.S. biotechnology firms, 1992–2003

<============ Pre-IPO market ============> At <== Post-IPO market ==> Independent Predicted Series Series Series Series Series Series IPO Variablea coefficient A B C D E ≥F filingb IPO+1c IPO+2 IPO+3

LCASH (+) 0.01 0.21 0.25 0.15 0.25 0.73 0.13 0.39 0.64 0.52 (0.1)d (3.5) (4.4) (3.1) (3.0) (6.0) (2.4) (4.9) (7.6) (5.6)

LNCASS (+) 0.18 0.13 0.18 0.09 0.21 0.28 0.10 0.21 0.25 0.25 (1.3) (1.3) (2.9) (1.3) (1.8) (2.4) (2.1) (4.2) (4.6) (3.3)

LLTD (–) n.app. –0.04 –0.04 –0.04 –0.10 –0.11 –0.03 –0.04 –0.07 –0.05 (–1.0) (–1.6) (–1.5) (–2.5) (–2.7) (–1.6) (–2.4) (–3.4) (–2.0)

LRD (+) 0.18 0.25 0.22 0.43 0.43 –0.33 0.17 0.18 0.22 0.31 (1.0) (1.9) (2.2) (4.3) (2.3) (–1.6) (1.9) (2.4) (2.3) (2.7)

# obs. 32 86 101 79 54 33 96 157 151 133

Pr > F(all slopes = 0)e 0.03 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

Adj. R2 [total]f, of which: 19% 42% 65% 65% 61% 79% 67% 57% 63% 54% Adj. R2 [common] 1% 19% 17% 15% 1% 34% 20% 6% –4% Adj. R2 [unique to year indicators] 7% 11% 5% 8% 5% 21% 16% 11% 20% Adj. R2 [unique to financial stmts.] 19% 34% 35% 43% 38% 73% 12% 21% 45% 39% Adj. R2 [only finl. stmts. in regn.]g 19% 35% 54% 60% 53% 74% 46% 40% 52% 34% CV of LPREM, LMVEh 0.13 0.11 0.09 0.08 0.08 0.09 0.08 0.08 0.10 0.12 CV of LCASH 0.42 0.24 0.17 0.18 0.14 0.11 0.17 0.09 0.10 0.13 Residual standard deviation 1.02 0.84 0.62 0.53 0.59 0.48 0.51 0.72 0.81 1.15 Median abs. valuation errori 70% 55% 36% 29% 38% 32% 41% 37% 52% 54%

Median TA/(EV + LTD)j 0.09 0.17 0.16 0.17 0.19 0.20 0.14 0.39 0.43 0.46 % shares held by 13-F institutionsk 0.22 0.28 0.31

45

Table 7 (continued)

Log-linear regressions of equity values on financial statement variables and indicator controls for 186 U.S. biotechnology firms, 1992–2003

Note: Sample comprises biotechnology firms with pre-IPO pre-money valuations in Recombinant Capital’s biotech valuation database and valid pre-financing financial statement data in SEC IPO filing documents (see fig. 3). Dollar amounts are in millions. For variable definitions, see table 1. Regression intercept and indicator coefficients are estimated but not reported. Equity valuations at the IPO filing are not included in the regressions. A regression intercept is always estimated but not reported. Financing rounds that are not labeled as being from a particular Series (e.g., Series A, Series B, etc.) are excluded. Such OTHER financing rounds include the sale of straight common equity, private placements, and debt related financings (such as bridge notes and convertible notes). Year indicator variables are included in all regressions except for Series A financings. Year indicators are for 1993–2003 (e.g., 1993 dummy = 1 if valuation date is in 1993, zero otherwise). Year indicators are included if a separate regression in which the only explanatory variables are the year dummies yields a positive adjusted R2 statistic. This is the case for all rounds except Series A. a A log transformation is applied to all nonindicator variables Z according to LZ = loge[Z +1] where Z ≥ 0 is in $000s. The regressions do not include the

variables LREV, LCOS and LSGA for parsimony, given that the results in table 6 indicate that at the pooled level they are not reliably different from zero in either private or public equity markets. If those variables are included, their coefficients are significantly different from zero only infrequently.

b Includes all IPO filings, whether ultimately successful or withdrawn. c “IPO+n” denotes three months after the nth fiscal year-end following the IPO offering, if successfully undertaken. d t-statistic relative to a null value on the coefficient of zero. e Pr > F(all slopes = 0) is the p-value testing the restriction that the coefficients on the set of financial statement variables LCASH, LNCASS, LLTD and LRD

are all zero. f Adj. R2 [total] = Adj. R2 [unique to financial statements] + Adj. R2 [unique to year indicators] + Adj. R2 [common to financial statements and year indicators]. g Adj. R2 [only financial statements in regression] is a regression where only financial statements are included (year indicators are not included). h CV denotes the sample coefficient of variation (= sample standard deviation divided by sample mean). i Valuation error = equity value less fitted equity value divided by fitted equity value. j Proxy for the relative importance to equity value of assets in place versus future investment opportunities. TA is total assets, EV is equity value (PREMV in

private equity markets; MVE in public equity markets), and LTD is long-term debt. See table 1 for variable definitions. k Proxy for the probability that the marginal investor is financially sophisticated, defined as the number of shares held by 13-F filing institutions at the end of the

fiscal year on or immediately preceding the valuation date divided by number of common shares outstanding. Source is the Disclosure SEC database.