Bay Bridge Bio
Are biopharma markets rational?
Biopharma stock returns have diverged from fundamentals over the last 15+ years. Investors reward pharma with rising valuation multiples despite deteriorating fundamentals, and biotech IPOs are priced consistently above intrinsic value (measured ex post).
It is tempting to frame this as a pharma problem: discovering good drugs is hard, clinical trials are expensive, Eroom's Law.
But in our opinion, this isn't a pharma problem. It is driven by the structure of financial markets and behavior of investors. Pharma companies are just playing by the rules set by markets.
There is a large body of literature explaining how biopharma stocks could be misvalued. This post will quickly highlight some of the key ideas from the literature.
This is not an exhaustive review. It is just a collection of papers that, in my opinion, explain biotech markets better than the Econ textbooks I studied back in undergrad.
This is not a pharma problem
The biopharma industry is not without its problems. Eroom's law, rising clinical trial costs, regulatory challenges.
But the specific problem of capital allocation -- which touches everything from R&D productivity, M&A and BD, and venture funding -- is not a pharma problem. It could just as well be an investor problem, or a problem with financial markets.
For example: we found that pharma stock returns from 2009-2025 were mostly driven by multiple expansion, and that in-house R&D contributed little to stock price growth, while ROIC steadily declined and R&D was well below the cost of capital.
That suggests investors were rewarding the sector with higher valuation multiples despite deteriorating fundamentals. Why is that? Is there something about pharma that has caused this? Are investors anticipating the industry's fundamentals will turn around?
It doesn't seem like it. Pharma multiple expansion over the 2009-2025 period is in-line with the S&P 500. Further, pharma returns over the period can largely be explained by systemic factors: investors did not treat pharma any better or worse than the rest of the market.
In other words, pharma stocks rose because the rest of the market did. The market didn't seem to care much about the underlying fundamentals.
We also found that biotech IPOs were overpriced relative to their intrinsic value, and that big pharma overpaying for M&A accounts for all of the excess return above the risk-free rate from investing in those IPOs.
That phenomenon cannot be explained by rising multiples or systemic factors, mostly because unprofitable, early-stage biotech companies don't really have valuation multiples, and their stock returns can't be explained by systemic factors.
But there are aspects of this that can be explained by the financial economics literature.
Why might biotech IPOs be overpriced?
Annual return from investing in 439 US biotech IPOs, 2009–21, held to exit
Pooled money-weighted return (XIRR) on the shares sold in each IPO, held to acquisition, failure, or 31 March 2026 (the analysis cutoff). "Fundamental return" values each acquired company based on the estimated discounted cash flows generated by its products, discounted at the biotech sector CAPM rate at the time. For acquired companies with no approved products or insufficient revenue data for approved products, the company is valued at its market value 42 trading days before the deal was announced. Source: Bay Bridge Bio retrospective DCF of 2009–21 biotech IPOs.
Our earlier analysis found that US biotech IPOs from 2009–21 returned -2.4% on a fundamental basis.1 The IPO pop brought the return to 2.0% (contributing +4.3% to returns), which is roughly equal to the risk-free rate over the period. Returns from big pharma overpaying in acquisitions of these companies contributed +5.5% to annual returns, bringing total return to 7.5% against an 8.4% cost of capital.
In textbook economics, informed investors would short an overpriced stock until the price came back down to intrinsic or fundamental value, i.e., the expected present value of future cash flows.
But that often does not happen in practice. The literature identifies several kinds of stocks that tend to be overpriced, and most of them describe a typical biotech IPO:
- Hard to short. Overpricing lasts when betting against a stock is costly or risky, and newly public biotech stocks are among the hardest to short.
- Lottery-like payoffs. Investors overpay for a small chance of a very large win.
- Distant, uncertain cash flows. When value depends on cash flows a decade away that no one can forecast well, little anchors the price.
- Disagreement about value. When investors disagree and pessimists cannot short, the most optimistic buyers set the price.
Further, IPOs and M&A are both transactions where prices can be biased upward. Companies choose to go public when valuations are high. In an acquisition, the winning bidder tends to be the one that most overestimates the target, and its managers can have their own reasons to overpay.
Importantly, *none of these are specific to biotech*. Most of the studies supporting these findings analyzed all sectors, not just biopharma. So we can expect these factors to cause overvaluation in other asset classes sharing these characteristics: venture capital, tech stocks, crypto, etc.
Select ideas from the literature
- Short-selling cannot always correct overvaluation. The phenomenon here can be described by Keynes' quote "The market can stay irrational longer than you can stay solvent". The idea that there are limits to the ability of short-selling to correct overvalued stocks has been established in the literature since for decades. Professional short sellers use other people's money, and expert short sellers (or other arbitrageurs) know more about their investments than do the clients who give them money. Clients withdraw funds after temporary losses, even if the short seller is right, which is when the mispricing is largest. Stocks with high company-specific risk are the costliest to hold short, so overpricing is corrected more slowly than underpricing. Biotech stocks have some of the highest company-specific risk of any asset, and newly public stocks are among the costliest to borrow. Stocks that are expensive to short have had high valuations and low returns afterward. In biotech a short can lose overnight if a pharma company buys the target at a premium. Professional money often joins a bubble instead of fighting it. In the dot-com bubble, hedge funds were heavily invested in technology stocks and cut back only shortly before the fall. A trader who cannot know when others will start selling does better riding the price up than betting against it early.Shleifer & Vishny 1997; Pontiff 2006; Stambaugh, Yu & Yuan 2015; Jones & Lamont 2002; Geczy, Musto & Reed 2002; De Long et al. 1990; Abreu & Brunnermeier 2003; Brunnermeier & Nagel 2004
- Investors pay extra for lottery tickets. People overweight small chances of a very large win, so sectors with a few huge winners and many losers trade above their average payoff. Most stocks lose to Treasury bills over their lifetime; a small number produce all the gains. Biotech IPOs have this shape, and IPOs with more lottery-like payoffs have larger first-day jumps. Stocks with high expected skewness, or a recent extreme daily gain, go on to earn low returns, and for small over-the-counter stocks the evidence points to buyers misjudging the odds. Issuers of lottery-like stocks are incentivized to keep supplying lottery tickets, even if expected returns are negative for buyers: markets serve what buyers will pay for, including their mistakes.Tversky & Kahneman 1992; Barberis & Huang 2008; Kumar 2009; Boyer, Mitton & Vorkink 2010; Bali, Cakici & Whitelaw 2011; Green & Hwang 2012; Eraker & Ready 2015; Bessembinder 2018; Akerlof & Shiller 2015
- The optimists set the price. When investors disagree and shorting is hard, pessimists simply stay out. The price reflects the most hopeful buyers, not the average view. If beliefs shift over time, the price can exceed what even the most optimistic investor would pay to hold the stock forever, because owners value the option to resell to whoever is most optimistic later. This is strongest when few shares trade, as in the months after an IPO. In the dot-com period, prices fell as lockups expired and more shares reached the market. In M&A, the price is set by an auction with no arbitrage, so the highest bidder sets the price, and the winning bidder tends to be the one that most overestimates the target. Overconfident CEOs make more deals and get a worse market reaction.Miller 1977; Harrison & Kreps 1978; Hong, Scheinkman & Xiong 2006; Ofek & Richardson 2003; Thaler 1988; Roll 1986; Malmendier & Tate 2008
- Distant, hard-to-forecast cash flows attract the most mispricing. Stock prices are more volatile than they should be based on the volatility of cash flows. When a valuation is based on cash flows a decade away that nobody can forecast well, there is little to anchor the price, and years pass before the market learns who was right. Investor sentiment moves the prices of young, small, unprofitable, volatile companies that pay no dividends the most, because they are both hard to value and hard to bet against. Those characteristics fit the typical biotech IPO. Such stocks earn low returns after periods of high sentiment. Stocks whose cash flows lie furthest in the future have earned lower returns than stocks paying out sooner; the gap is larger after high-sentiment periods, appears only among stocks that are hard to short, and coincides with analysts extrapolating past growth. More generally, stocks with more uncertain information show stronger overreaction, stocks on which analysts disagree most earn lower returns afterward, and very volatile stocks have earned unusually low returns, the opposite of what their risk would suggest.Shiller 1981; Baker & Wurgler 2006; Weber 2018; Dechow, Sloan & Soliman 2004; Zhang 2006; Diether, Malloy & Scherbina 2002; Ang et al. 2006
- IPOs are sold when sentiment is high. Companies choose when to go public. IPOs have been priced above comparable listed companies, and the most overpriced ones had the biggest first-day gains and the worst returns afterward. IPOs as a group have lagged similar stocks for three to five years, and venture-backed biotech companies have gone public when biotech valuations were high. Private markets behave the same way: when money pours into venture funds, startup valuations rise, with no sign that the companies funded go on to do better.Ritter 1991; Loughran & Ritter 1995; Lerner 1994; Purnanandam & Swaminathan 2004; Gompers & Lerner 2000
- IPO buyers expect to sell to pharma. If acquirers reliably pay more than the products return ($0.62 of product value per $1 paid in our data), then a price above cash flows is correct for an investor who expects to be bought out.Our own deal analysis, 83 acquisitions of biotech IPOs from 2009-2021
- Acquirers' managers are paid to do deals. Bidder returns are lower when a deal serves the managers' goals, such as diversifying or buying growth. CEO pay and wealth tend to rise after an acquisition even when the acquirer's shareholders lose, and boards pay deal bonuses that track the effort of closing a deal, not how it turns out. Younger CEOs, with more years of the higher pay ahead of them, make more acquisitions. The cost of overpaying often shows up only years later, as goodwill write-downs.Morck, Shleifer & Vishny 1990; Harford & Li 2007; Grinstein & Hribar 2004; Yim 2013; Gu & Lev 2011
Opposing evidence
- Uncertainty raises fair value. When future growth is very uncertain, the upside is unlimited and the downside is capped at zero, so a young company with no earnings can be rationally worth more than its expected path suggests (Pástor & Veronesi 2003, 2006).
- Hindsight is one draw from the universe of potential outcomes. When looking forward, there is a distribution of potential outcomes. When looking backwards, we view only one draw from that distribution. Assets that look overpriced could have been priced correctly ex ante, and simply followed a less-than-favorable draw from the distribution of potential outcomes. In a highly skewed distribution, a single missed blockbuster would change the average. If dividends are smoothed or trend over time, prices can rationally move more than realized-dividend values (Kleidon 1986; Marsh & Merton 1986). Some of the IPO underperformance literature has been attributed to measurement, or to small growth stocks in general (Brav & Gompers 1997; Schultz 2003).
- IPOs may not be mispriced Our finding that 2009-2021 IPOs were valued above what the companies went on to deliver does not prove mispricing. Proving an asset is "mispriced" requires proving the actual expected return differs from its risk-adjusted benchmark, i.e., that it generates excess returns. Stock returns are noisy, so to get a reliable estimate of expected return, you need centuries of data (Merton 1980, Black 1986).
- Markets may underprice R&D, not overprice it. Firms that spend heavily on R&D relative to their market value, and firms with a record of turning R&D into sales, have earned higher returns afterward, which suggests investors undervalue long-dated research rather than overvalue it (Chan, Lakonishok & Sougiannis 2001; Cohen, Diether & Malloy 2013).
- Overpaying may still fund useful science. Hot markets fund riskier, more novel companies. More of them fail, but the ones that succeed are more innovative. Investors can lose money while society gains from the experiments they paid for (Nanda & Rhodes-Kropf 2013; Perez 2002; Janeway 2012).
Why did pharma's multiple rise while its fundamentals slipped?
Pharma's multiple rose while it's ROIC declined and R&D productivity was below the cost of capital. The rise in multiple seems to be due to general market factors, not anything pharma-specific.
Pharma's multiple rose about in line with the S&P 500's over the 2009-2025 period. A factor analysis shows that a basket of big pharma stocks returned no excess return above what standard risk factors predict. This supports an explanation that pharma's rising multiple was driven by market-wide factors rather than anything specific to pharma. Essentially, investors value big pharma like any large, defensive stock.2
Specifically, big pharma's multiple expansion concentrated in 2010-2015 (the post-GFC period through the 2015 biotech bubble). Over that period, revenue grew 1.10x and the multiple grew 1.78x. This looks to have been driven by a broad rally in low-beta, defensive stocks broadly, not anything specific to pharma.3
But why did pharma grow with the market, despite deteriorating fundamentals? Several ideas from the literature offer explanations. In general, most large investors are long-only or long-biased. Further, passive index investing does not discriminate based on characteristics of individual companies. Additionally, there are limits to how much arbitrageurs can correct overpricing. So when more money enters the market, a lot of it goes to passive funds or long-only funds that bid up asset prices, and short-sellers are not able to offset this increase in price. Lastly, the "Fed put" since the mid-1990s has tended to support markets after declines, without an offsetting tightening when markets rise. If investors anticipate the Fed put continuing in the future, this pushes up asset prices independent of cash flows.
Of course, the multiple expansion could also be driven by investors anticipating future growth that is higher than that of the recent past. The multiple expansion could also be a function of a decreasing discount rate: the discount rate we used did not decline over the period, but the actual discount rate could have declined more than common methods for calculating the discount rate indicate.
Select ideas from the literature
- Money flowing in moves prices a lot. The textbook assumption is that if a large buyer pushes the price up above fair value, others sell and the price barely moves. But most large investors hold fixed stock allocations and do not sell. In other words, the market is inelastic: demand for stocks is insensitive to price changes. One estimate is that $1 moved into stocks raises total market value by about $5. A later estimate puts it near $2. Index funds buy at any price, though the price jump when a stock joins the S&P 500 has shrunk as index funds have grown.Gabaix & Koijen 2021; Hartzmark & Solomon 2025; Koijen & Yogo 2019; Haddad, Huebner & Loualiche 2025; Greenwood & Sammon 2025
- Falling interest rates lift distant cash flows most. Stocks pay out over decades, so they behave like very long bonds. One study finds that a bond portfolio matched to the timing of the stock market's payouts did as well as stocks over the past half century. Later work gives rates a smaller role: about a third of the fall in rates passed into valuations as pure discounting, and 40% of the 1989–2017 rise in US equity value came from profits shifting away from labor and toward shareholders. Low rates also push investors, especially those who are leverage-constrained, toward riskier assets to reach a target return. Low-volatility stocks carry bond-like exposure to rates, which fits the defensive rally above.van Binsbergen 2020; Gormsen & Lazarus 2026; Greenwald, Lettau & Ludvigson 2025; Lian, Ma & Wang 2019; Driessen et al. 2019
- Investors expect the central bank to cushion losses. Since the mid-1990s the Federal Reserve has tended to ease after stock market declines. If investors count on that, stocks are less risky to hold and deserve a higher price, regardless of fundamentals. A backstop that limits losses also removes part of the discipline that falling prices would otherwise supply. Easier policy also raises investors' appetite for risk, beyond the effect of the rate itself. One channel is redistribution: a surprise rate cut shifts wealth toward the investors who hold the most risky assets, and they bid prices up further. Valuations have moved in long regimes that line up with the Fed's policy rate.Miller, Weller & Zhang 2002; Cieslak & Vissing-Jorgensen 2021; Kekre & Lenel 2022; Bianchi, Lettau & Ludvigson 2022; Bauer, Bernanke & Milstein 2023; Knox & Vissing-Jorgensen 2026
- A high price forces managers to buy growth. Once a stock is priced for growth the business cannot produce, managers make acquisitions to meet the expectation, including ones that lose money. This potentially describes pharma's dealmaking. Across all US acquirers, shareholders lost about $240 billion on deals announced in 1998–2001, mostly in a few large deals by highly valued buyers. In pharma, large mergers have followed patent expiries and pipeline gaps without fixing them, and companies that lose a product raise R&D spending mainly by buying other companies' pipeline drugs.Jensen 2005; Moeller, Schlingemann & Stulz 2005; Danzon, Epstein & Nicholson 2007; Krieger, Li & Thakor 2022
Opposing evidence
- The required return may simply be low. Discount rates were relatively unchanged over the measurement period, so we attribute multiple expansion to an increase in growth expectations beyond what pharma has delivered. However, our discount rate measurement may be incorrect. We use Damodaran's estimate for the equity risk premium. It may be the case that the actual equity risk premium has decreased over the period, which would explain the rise in valuation multiple without requiring an expectation of high future growth.
- The flow estimates are disputed. The studies showing that capital flows increase price in an inelastic manner are disputed. Estimates from changes in fund holdings find a smaller price effect per dollar that partly reverses over time (van der Beck 2022). Two methods papers argue that the usual comparisons across stocks cannot pin down the market-wide number (Haddad et al. 2026; Fuchs, Fukuda & Neuhann 2025). Flows also explain why prices move, but less clearly why they would stay too high.
- The Fed backstop may have faded. In option prices the backstop is large before 2008 and absent afterward, which is the period our data cover (Dahiya et al. 2026).
- Low rates are not required. The 2000 biotech and internet bubble happened with short-term rates above 6%. Stocks of companies rich in intangible assets have been found to react less to rate changes, not more (Döttling & Ratnovski 2023).
- Deals may be worth more than their products. Buying a competitor's drug can protect the acquirer's existing sales, a benefit that never shows up as revenue from the acquired product (Cunningham, Ederer & Ma 2021). Earlier work found acquirers' shares rose when they bought to fill pipeline gaps (Higgins & Rodriguez 2006).
Is this the best we can do?
Biopharma markets exhibit what looks like irrational behavior: overpaying for M&A, big pharma getting higher valuations despite deteriorating fundamentals, and investors consistently valuing biotech IPOs above intrinsic value.
Some of it is only irrational if we assume markets price assets based on intrinsic value.
But why do we expect markets to be rational in the first place?
Textbook economics describes people as homo economicus, making decisions based on rational self-interest. But decades of literature, theoretical and empirical, including an entire field of finance that has created multiple Nobel Prize winners (behavioral finance), call into question this view of humanity.
Markets may sort out some individual irrational behavior, the way that large samples smooth out noise. But a large body of literature calls into question whether markets price assets based on their intrinsic cash flows, and whether markets are rational all of the time.
If society's medicines are all developed through an industry that doesn't allocate capital efficiently, what should we do about that?
On the one hand, if that irrationality leads more money to be invested into R&D, that could be good for society, even if it is not invested efficiently. On the other hand, if investors don't need biopharma to be good at R&D in order to make money, there is little incentive to improve R&D productivity.
In any event, I think it is worthwhile to ask: can we do better?
Notes
Fundamental return. The return from the closing price on day 1 after the IPO to the date of acquisition or the terminal date, valuing acquired companies based on the discounted cash flows generated by their products. ↩
Big pharma’s returns look like any large, defensive stock. Fourteen large pharma companies, weighted by market value, monthly returns 2006–2025 (240 months). Eight factors together explain 43% of the variation. Total return 10.3% a year against 11.1% for the US market. From 2012, adding biotech factors shows a link to the biotech index beyond the market (loading 0.34, t-stat 6.5). The basket is today's largest companies, so it leaves out companies that were acquired along the way or shrank.
↩Exposure Loading t-stat Reading Market 0.69 11.7 Moves about two-thirds as much as the market Size (small minus big) −0.32 −3.4 Behaves like large companies Investment (conservative minus aggressive) 0.35 2.0 Behaves like slow-growing companies Value, profitability, momentum −0.15 to 0.00 under 1.3 No reliable exposure Low beta (AQR betting against beta) 0.13 1.2 Weak Quality (AQR quality minus junk) 0.20 1.5 Weak Unexplained return +1.1% a year 0.5 Indistinguishable from zero 2010–15: the re-rating rode a rally in defensive stocks. Most of pharma's multiple expansion happened in 2010–15. Over those six years revenue grew 1.10× and the multiple 1.78×. In that window pharma loads heavily on the low-beta and defensive factors. Same fourteen-company basket, weighted by market value, 72 monthly returns. The defensive factor is AQR's US defensive stocks portfolio; the global model is AQR's market, size, value, momentum, betting-against-beta and quality-minus-junk factors.
↩2010–15 model Loading (t-stat) Unexplained return, % a year (t-stat) Market only — +6.2 (1.7) Market + AQR US defensive 0.47 (4.1) −0.5 (−0.1) AQR six factors, global Low beta 0.75 (4.6) −3.5 (−1.2)
References
- Barberis & Huang (2008), American Economic Review. Bessembinder (2018), Journal of Financial Economics. Green & Hwang (2012), Management Science. Kumar (2009), Journal of Finance.
- Miller (1977), Hong, Scheinkman & Xiong (2006), Ofek & Richardson (2003), Shleifer & Vishny (1997), Stambaugh, Yu & Yuan (2015), all Journal of Finance. Pontiff (2006), Journal of Accounting and Economics.
- Ritter (1991), Loughran & Ritter (1995), Brav & Gompers (1997), Schultz (2003), Pástor & Veronesi (2003), Journal of Finance. Purnanandam & Swaminathan (2004), Review of Financial Studies.
- Gabaix & Koijen (2021), NBER working paper 28967. Koijen & Yogo (2019), Journal of Political Economy. Haddad, Huebner & Loualiche (2025), American Economic Review.
- van Binsbergen (2020), NBER working paper 27367. Lian, Ma & Wang (2019), Cieslak & Vissing-Jorgensen (2021), Greenwood & Shleifer (2014), Review of Financial Studies.
- De la O & Myers (2021), Journal of Finance. Bordalo, Gennaioli, La Porta & Shleifer (2024), Journal of Political Economy. Jensen (2005), Financial Management.
- Koijen, Philipson & Uhlig (2016), Econometrica. Döttling & Ratnovski (2023), Journal of Monetary Economics. Cunningham, Ederer & Ma (2021), Journal of Political Economy. Higgins & Rodriguez (2006), Journal of Financial Economics.
- Akerlof & Shiller (2015), Phishing for Phools, Princeton University Press. De Long, Shleifer, Summers & Waldmann (1990), Journal of Political Economy. Abreu & Brunnermeier (2003), Econometrica. Brunnermeier & Nagel (2004), Moeller, Schlingemann & Stulz (2005), Journal of Finance. Gompers & Lerner (2000), Nanda & Rhodes-Kropf (2013), Journal of Financial Economics.
- Perez (2002), Technological Revolutions and Financial Capital, Edward Elgar. Janeway (2012), Doing Capitalism in the Innovation Economy, Cambridge University Press. Miller, Weller & Zhang (2002), Economic Journal. Sharpe (1991), Financial Analysts Journal.
- Bali, Cakici & Whitelaw (2011), Eraker & Ready (2015), Nagel & Xu (2023), Journal of Financial Economics. Hartzmark & Solomon (2025), American Economic Review. Greenwood & Sammon (2025), Bianchi, Lettau & Ludvigson (2022), Journal of Finance. Greenwald, Lettau & Ludvigson (2025), Journal of Political Economy. Kekre & Lenel (2022), Econometrica. Bauer, Bernanke & Milstein (2023), Journal of Economic Perspectives. Driessen, Kuiper, Nazliben & Beilo (2019), Dahiya, Kamrad, Potì & Siddique (2026), Journal of Banking & Finance. Danzon, Epstein & Nicholson (2007), Managerial and Decision Economics. Krieger, Li & Thakor (2022), Management Science.
- Baker & Wurgler (2006), Zhang (2006), Diether, Malloy & Scherbina (2002), Ang, Hodrick, Xing & Zhang (2006), Chan, Lakonishok & Sougiannis (2001), Journal of Finance. Weber (2018), Journal of Financial Economics. Dechow, Sloan & Soliman (2004), Review of Accounting Studies. Cohen, Diether & Malloy (2013), Review of Financial Studies.
- Thaler (1988), Journal of Economic Perspectives. Roll (1986), Journal of Business. Morck, Shleifer & Vishny (1990), Harford & Li (2007), Journal of Finance. Grinstein & Hribar (2004), Yim (2013), Malmendier & Tate (2008), Journal of Financial Economics. Gu & Lev (2011), The Accounting Review.
- Jones & Lamont (2002), Geczy, Musto & Reed (2002), Lerner (1994), Journal of Financial Economics. Boyer, Mitton & Vorkink (2010), Review of Financial Studies. Tversky & Kahneman (1992), Journal of Risk and Uncertainty. Shiller (1981), Marsh & Merton (1986), American Economic Review. Harrison & Kreps (1978), Quarterly Journal of Economics. Kleidon (1986), Journal of Political Economy.
- Gormsen & Lazarus (2026), NBER working paper 34814. Adam & Nagel (2022), NBER working paper 29977. Haddad, He, Huebner, Kondor & Loualiche (2026), NBER working paper 35413. Knox & Vissing-Jorgensen (2026), Federal Reserve FEDS working paper 2026-023. van der Beck (2022), Swiss Finance Institute research paper 22-67. Fuchs, Fukuda & Neuhann (2025), working paper.