Why Would FDA Approval Cause a Stock to Increase Ten-Fold?

In the news:

Vanda Pharmaceuticals won U.S. approval for its first product, a drug to treat schizophrenia, the Food and Drug Administration said. ... Vanda, which closed at $1.08 in regular trading yesterday, soared to $9.98 in extended Nasdaq trading.

It would appear, going by the jump in price, that prior to the FDA's decision, investors gave the drug at most a 1 in 10 chance of being approved. If Vanda's case is typical, then, even moments before the actual decision is made, it is hard to predict what drugs the FDA will approve.

I can understand it being difficult to predict whether a new chemical will turn out to be a useful drug when all the testing is done five or ten years down the road. But once the testing has been done, at the very least the results of those tests are fully known. There is no uncertainty about the results that have already been observed. And the FDA decision is, necessarily, based entirely on the results available at the time the decision is made.

If the FDA's decisionmaking process is

a) not arbitrary, and
b) based on the available results,

then the FDA's decision should be highly replicable, and therefore highly predictable, by any independent entity with access to the same results. And yet the FDA's decision is, apparently, hard to predict. Two possible alternative explanations are:

1) The results that the FDA bases its decision on are extremely well-guarded right up until the very moment that the FDA makes its decision. I doubt this is possible.
2) The FDA's decisionmaking process is highly arbitrary. This is my tentative conclusion.

A few anticipated objections and responses.

Objection: Vanda's case is not typical.

Answer: Could be. However, this seems not all that atypical. When I read the story it didn't really stand out as atypical.

Objection: The typical investor doesn't know how to interpret the data, doesn't know what the data is, etc.

Answer: This is true of most investors in most publicly traded companies. If it were a significant problem the efficient market hypothesis (EMH) would be not only wrong, but wildly wrong, all the time, and Vanda's case would provide a model for disproving the EMH.

Objection: Pharmaceutical companies really keep a tight lid on their results.

Answer: I have a hard time believing that. A tremendous number of people are involved in any study that gets to this stage. Even partial information should give a sense of how well a drug is working and what its side-effects are.

Objection: The cause of the unpredictability isn't that the FDA is arbitrary, but that the drug is borderline useful, and even the most predictable decisionmaking process will be unpredictable when it comes to borderline cases.

Answer: But surely the typical drug is not borderline.

Objection: The FDA can hardly be blamed because it is fundamentally hard to judge whether a drug is useful or not. It is unclear and/or subjective whether a given drug is useful or not.

Answer: Then why is the FDA making a decision for all of us?

Share this

Very nice post. There's a

Very nice post. There's a case you don't consider, a combination of two explanation, Vanda's case is atypical in that it is marginally useful.

Another one. Analyzing the result is costly and prediction of the FDA approval may not allow a profit large enough to cover the cost of analysis. Especially if you assume other people may have already done that job and the price might be priced accurately.

(1 - Probability that the data has been analyzed and is priced in) * (Payoff from superior information) < Cost of Analysis

It'd be nice to have the distribution of returns of pharmaceutical companies following FDA announcements.

thats simply how the stock market works

I don't think its that complicated. The approval of the medicine caused an event causing recognition of a company and product that most investors had not heard of previously. This news story is probably the first that any potential investors heard of the pharmaceutical company (especially seeing this is the first drug they ever produced).

For instance, there is another type of "event" that causes stock prices to go up, its the pump-and-dump, i.e. microcap stock fraud. In this case, the event isn't a legitimate news article, it is emailed spam causing fervor over a particular stock that causes its value to escalate, and that's when the fraudsters dump their stocks.

If you're really good in the stock market, you do your research, keep a close eye on upcoming pharmaceutical companies, you research the drug they're intending to produce, you estimate exactly when that drug is coming out - congratulations you took somewhat of a gamble and got that company the early funds it needed to see the product come to fruition, and made a healthy profit for it. Everyone else that drove the price to $9.98 - merely represent people who had no clue about the product or company prior to the news event.

I see your article as simply how the whole stock market works, nothing extraordinary or remarkable. Those doing research and being the first to invest in something before it is hyped profit far more than those who respond the day after a newsworthy event takes place that would manipulate a stock.

The conclusion that the total sum of investors (lets say 100 investors) 5 of them decided to invest early and 95 decided they would rather wait until after approval is probably not an accurate way to look at it. Its more like 5 did the investigative work and invested, and 95 didn't know about it until the newsworthy event which caused a huge influx of demand.

Contradicts the EMH

I understand where you're coming from. I see investors and investment advisers talk this way all the time. But it contradicts the EMH. While I don't expect the market to conform with infinite perfection to the EMH, the systematic large departure that you are claiming exists strikes me as being implausible.

The EMH can be refined by

The EMH can be refined by including the cost of obtaining information.
It's costly to look for a stock no one has heard of, it's costly to analyze it's value. This cost has to be balanced against the potential profit if a mis pricing is uncovered. If market actors do not know which stocks are being monitored, it can be rational not to look into stocks, assuming it's very likely someone has done the job and the price is now correct.

You already said this

But this is an exception to the EMH. I haven't denied that there can be large exceptions here and there. What I have trouble with is the suggestion that this is the rule rather than the exception. Jack mentions "pump and dump", but this is exceptional. This does not happen to the majority of stocks. But then Jack says, "I see your article as simply how the whole stock market works, nothing extraordinary or remarkable." The whole stock market! That is what I object to. I am not objecting to the possibility that occasionally available information is not incorporated because the computational cost of incorporating it is too high and (as you also add) there may be uncertainty about whether or not somebody has already done the computation (such a situation, in which nobody computes it because everybody mistakenly thinks someone else has probably computed it already, is surely a market failure, and therefore an exception to the EMH). I accept that market failures are possible. What I object to is the notion that the whole stock market is a market failure.

What I describe is a general

What I describe is a general rule, not an exception. There is no reason to believe the EMH to be correct once one has agreed that getting information isn't costless.

It's not about correct/incorrect

The claim that the EMH is never precisely correct is not what I was objecting to. What I found implausible was that "large [departures from EMH]" are "the rule rather than the exception". Emphasis on "large" along with "rule". In particular, I find your speculation about calculation cost as it would be applied to the question of why the large jump in stock price only logically possible, not very probable. By "your speculation" I do not mean your claim that the cost is non-zero. I mean the speculation that it may sufficiently high that, combined with worries that it has already been performed, nobody performs it. If you want when I have time I can crunch some numbers to explain why I find it extremely unlikely that this is a significant factor in the price jump.

I'll give a rough argument. First, I don't think it's tremendously costly to estimate the value of the drug once approved. Therefore, using this value, it is inexpensive to discover that the current value of the stock is a small fraction of the value it would be if the FDA approved it. And meanwhile, it is (since not so close to zero) presumably well above the value it would be if the FDA rejected it. Therefore it is clear - it is inexpensive to discover - that it is not the case that:

a) the FDA's decision is predictable and

b) someone has predicted it and brought the price up or down to the corresponding value.

Clearly, that has not happened, so clearly, either (a)-and-not-(b), or else not-(a). Since this was clear all along and yet no one went ahead and did the calculation at any point, we can probably rule out the first, leaving not-(a).

The cost of predicting the FDA's decision is (since it merely involves performing the same calculation that the FDA itself performs) comparable to the cost that the FDA itself incurs when making the decision. I think it is unlikely that the FDA's cost of making the decision comes anywhere near the value of the drug. Or anywhere near 1/10 or 1/20 or 1/1000 the value of the drug. And the probability (going in) of FDA approval is not super-remote. It's not 1/1000. The stock had a certain price prior to approval, and going by this price, the market did not estimate the drug's chance of approval as super-remote.

Mostly Arbitrary

It seems the needed information on drug trails isn't very expensive:


From quickly doing a few google searches it seems stock prices of many biotech companies are heavily influenced by FDA approval. Also smaller companies struggle to raise the capital required to pass FDA approval (it's taking more than a decade to get products to the market since the FDA gained greater powers in 1962). Not to mention the fact that the number of new products every year has been halved.

Some good reading:


From anecdotal evidence it seems that not even the companies themselves know whether their product will be approved or not. Which is obviously rather strange since the companies are doing all the tests and know exactly what the FDA's requirements are. The trails and testing span several years with constant FDA supervision. One would think that the scientists smart enough to develop the drugs in the first place would also be smart enough to know whether or not the drugs meet the FDA's regulations.

So if FDA approval / denial appears arbitrary to even the manufacturers themselves clearly there's no way traders will know any better.

I believe in a situation like this even the semi strong EMH should hold since traders do have access to the fundamentals, ie. drug trail results etc. So unless FDA approval is arbitrary the stock price should have increased to $10 during the course of the drug trials. Even if only 5 traders bothered to read the drug trail reports they would have bid up the price to $10. Uninformed traders wouldn't take the risk of shorting the stock (and even if they did the other 5 would buy those cheaper stocks) so only those traders that did their research would be pushing the price.

I believe it's quite clear that the fundamentals aren't driving the FDA's decision making process and traders were giving approval a 10% chance irrespective of drug trail results.

And if it isn't enough that the FDA is making our markets less efficient they also happen to kill more people every year then they save. That's right, the FDA's nett result is dead people.

Just to clarify the EMH

The EMH doesn't state that a stock's (or whatever is being traded) current value represents actual future earnings, only that it represents the best probability of future earnings with information currently available.

This case actually substantiates the EMH. It shows that the market is using known information, ie. the FDA's actions in the past, to determine the current price. So even though traders knew drug trails showed the drug to be safe, they also knew the FDA's decision is not based on drug trail results. Thus all information available to all traders was giving FDA approval only a %10 probability.

Furthermore the EMH doesn't require all (or even many) traders to be biding on a stock for the EMH to hold. Only those traders with information regarding a stock's future earnings. So even if other traders hear about a specific company for the first time and start trading it, as long as they don't have any new information, the stock price won't be effected. This might seem counter intuitive to your real world experience, but remember that the reason companies appear in the news is usually because there is some new information regarding their future earnings. This leads us to think that simply mentioning a company's name in the news drives its stock price. (Which it arguably might since good marketing leads to good earnings (-: )

EMH is a pretty good theory and I'm yet to see a good attempt at disproving it. Of course markets aren't perfect and they're constantly being made more efficient.

what about the risk to capital?

Before the first approval for a small company like this, there are a lot of risks associated with the company. They could run out of cash and go out of business before being able to sell their product, or they could be incompetently managed, or their underlying drug might not work. There is a great deal of uncertainty in the value of their stock. Most investors hate uncertainty, so they are reluctant to invest - and so to attract capital, the company has to offer a major discount, which manifests itself in the form of a stock price much lower - even by a factor of ten, say - than what a purely rational analysis would indicate.

Once the approval drops, however, the uncertainty is greatly reduced. Now the much larger community of investors is willing to consider the stock, since the biggest form of uncertainty (that the FDA would say no) has been eliminated. The stock is genuinely much more valuable now, because the chance that the company will vaporize is now much smaller.

See cputter's comments

FYI, cputter's comments, particularly the second one ("just to clarify the EMH"), add to my argument and explain some things which I didn't make explicit, particularly the point about the EMH being robust even when the stock is relatively unknown. This addresses Jack's point that "the approval of the medicine caused an event causing recognition of a company and product that most investors had not heard of previously" and Robert's similar point that "now the much larger community of investors is willing to consider the stock."

Robert makes the point that "the biggest form of uncertainty [is] that the FDA would say no." Exactly. Robert talks about the discounting of the stock price in the face of uncertainty - but that's what I've been talking about, i.e., uncertainty about FDA approval causing the stock price to be 1/10 of what it will be once the FDA approves. Robert talks about "a stock price [being] much lower ... than what a purely rational analysis would indicate", which suggests that maybe Robert has in mind that there will be an additional discount of the stock price over and above the rational discount in the face of uncertainty. If that were the case, it would be a significant violation of the EMH. But I don't think that excessive caution plays as large role as one might think, because investors can shield themselves from risk by diversifying.

One of the ways that investors can overcome the human tendency to be excessively cautious in the face of uncertainty (as Robert suggests they tend to be) is to diversify their portfolio. This allows an investor to invest in high-risk stocks without exposing himself to very much risk at all. Once the investor is no longer himself exposed to high risk, the tendency to be excessively cautious no longer kicks in. Of course diversification fails if the individual risks are not independent. An example of non-independent risk is the risk of a general economy-wide recession. But the risk of recession affects both low-risk and high-risk stocks.

Risk, uncertianty, probability...

I like this topic. More posts regarding the amazingly wonderful behaviour that emerges from millions of self interested people cooperating will be great.

We've been throwing about several different terms like risk, uncertainty and probability which all refer to the same thing in this context: future events unknown to mere mortals...

Sticking with the topic du jour let's assume the future event we're considering is Vanda receiving a letter from the FDA proclaiming that they won't be jailed for improving the quality of life of thousands of Americans. The future occurrence of this event is assigned a probability p. If p = 1 we are certain that this will happen, p = 0 implies this will never happen and p = 0.5 means we might as well flip a coin since we have no reason to suspect it's outcome either way. The closer p is to 0.5 the more uncertainty we face regarding this future event.

If Vanda does not get approval we know it's future earnings will decrease. Thus we risk the loss of future earnings if they do not receive approval. The probability of this happening is 1 - p, exactly the opposite probability of them being approved. So risk refers to the high probability of events occurring in the future that will impact our earnings negatively, or conversely the low probability of positive events.

Our assessment of the probability of future events is referred to as information. 'Risk' and 'uncertainty' are merely used to classify our information. In the stock market information is the name of the game. Millions of people are acting on their information, driving stock prices. EMH says that the current price is the most accurate representation of all available information regarding future earnings, ie. it represents all the probabilities of all future events.

For all intents and purposes the market functions like a time machine, it brings the future to the present.

So investors don't discount stocks prices due to risk or uncertainty. Both are merely more information regarding future events that gets added to all the other information already out there. Every tiny little thing is incorporated into the price, the manager's past performance, his expected future performance, the companies cash flow, quarterly reports, asset values, R&D, industry performance, etc., etc... All known information, past and future.

The 'pump-and-dump' example is pretty interesting and points out the market's major weak point: a lack of information. Fraudsters exploit companies that have little public information available by hyping them via spam e-mails. Investors receive the supposedly positive information regarding the company but can't verify it independently. Some of them react on this which drives the stock up, the fraudsters cash out the same day. This again substantiates the EMH by showing how sensitive the market is to information (remember, EMH doesn't require the information to be correct).


Good point about diversification reducing risk. Again it's all about the probability of future events. Your reducing the probability of losing all your investments. If the probability of 5 investments going bust is 0.5 each, and they're independent, the probability of all 5 going bust is 0.5^5 = 0.03. So if you have $1000 to invest, instead of a %50 chance of losing everything in one investment, you can diversify and face only a %3 chance of losing everything. Much better odds :-)

also have to consider dividends

Another thing to consider, a company that has uncertainty about FDA approval, still in research and development mode, most likely does not pay dividends yet.

A pharmaceutical company that just had FDA approval on a medicine is in a position to get into the "black" and to offer good dividends to its stockholders.

The extra income producing ability of the stock at this point may make up for a majority of the stock's increase in value - because this would offset the cost.

You could abstract this to anything. I will pay $10 for anything that will produce ____ amount of expected dividends and at least retain its value of ____ or more. It could be a rental property, for instance. A rental property that is NOT rentable - but might be rentable in 2 years (maybe, maybe never) ... ok I'll pay $19,800 for that. A unit that I can rent starting tomorrow for $1000 a month, ok I'll pay $198,000 for that. Why the 10 times increase? Because that $1000 a month will pay more than a bank CD and cover my initial loan for the property to begin with. An investment for $19,800 in the "maybe" property is very risky - and in today's economic climate where people have lost half or more of their nesteggs, there are very few willing to take that kind of risk.

Math's off

The increase from $1.08 to $9.98 is closer to 9-fold than 10-fold. A 9-fold increase is $9.72 while a 10-fold is $10.80.

Oddly enough

By a curious coincidence, my first draft was "nine-fold", though not for your good reason - I simply rounded both values down before taking their ratio, i.e. 1 versus 9. I switched it to "ten-fold" for the reason that you might imagine - I rounded the two numbers to the nearest whole increment before taking their ratio, i.e. 1 versus 10.