INTRODUCTION
Everyone thinks big pharmaceutical research and development organizations are dinosaurs.
We don’t. We think they have the potential to be far more productive than small biotech firms, if they are structured right.
Big pharma R&D operations are unproductive because they are set up like the centrally planned economy of the former Soviet Union, with similar bureaucracies, inefficiencies and disincentives. That’s tremendously frustrating for both managers and scientists.
There’s an alternative, however, and one that we’re all familiar with: the decentralized free market economy, which rewards innovation and efficiency.
Of course, big pharma R&D exists within a free market now. Yet because of big pharma’s centralized structure and unique challenges -- especially the time lag from the genesis of a drug to FDA approval – it is unable to take advantage of the vital and rapid feedback the free market offers.
Our solution: Transform the internal structure of big pharma R&D departments to resemble decentralized, free markets, with projects that can operate as nimbly as small start-up firms. Add the advantages bigger pharmaceutical firms already enjoy – especially abundant financial resources and insulation from public disclosure during the early stages of development -- and you have a formula for success.
Tinkering won’t do the trick, though. That’s been amply proven by experiments the biggest firms have undertaken in recent years on the advice of consultants and innovation-minded managers.
Those experiments have failed because the changes have addressed only one aspect of the problem – usually centralized planning -- without addressing other vital elements such as positive incentives, rapid feedback, and autonomy for scientists and support staff to work on projects they believe in.
For transformation to be successful, change must be radical and account for the interplay of many factors and groups. This is where the science of economics can provide a critical, “big picture” perspective.
First, we can use economics to understand all the elements of the problem and how they contribute to the drag on productivity. Then we can use economics to show how key changes in organization and incentives can eliminate that drag and stimulate development – without micromanagement. We also can use economics to figure out how best to make the transition.
But first, let’s dissect the problem.
FREE MARKETS 101, OR TOMATOES FOR MOSCOW
There are three key elements to the success of a free market and companies within that market: prices, incentives, and decentralized decision-making. In centralized economies, these elements are absent or distorted in such a way that producers cannot respond quickly to market demand and have no reason to do so.
Let’s use the hypothetical example of agriculture in the former Soviet Union. A planner in the Ministry of Agriculture in Leningrad tells a farmer in the Ukraine to grow 10,000 bushels of salad tomatoes this year and take a certain number each Friday in August and September to the local shipping depot. Truckers will carry the tomatoes to Moscow, where they will be stored in a warehouse, then distributed to government-run food stores.
The Ministry of Agriculture pays the farmer a fixed price – 50,000 rubles – for all the tomatoes he grows. Through the local agricultural collective, the government provides each farmer with seeds, fertilizer, pesticide, and the loan of machinery for plowing, planting, fertilizing and spraying. The farmer also is allotted a certain number of pickers each Thursday during harvest season.
In Moscow, prices for tomatoes are set by the government at 2 rubles apiece. There are about 30 salad tomatoes in a bushel. If everything works perfectly, this price gives the government enough money to pay the farmer, the trucker, the pickers, the people in the warehouse and the cost of seeds, tractors, trucks, fuel, etc. Meanwhile, Muscovites get all the fresh tomatoes they want, when they want them.
But farming is a risky business, and rarely does everything go perfectly. Some risks are inherent, such as the weather, which neither the farmer nor the government can control. Other problems are exacerbated by the centrally planned economy.
If the Ministry of Agriculture planner incorrectly estimates the demand for tomatoes in Moscow this year, there will be a shortage or a glut. Either way, the government will be unable to sell enough tomatoes to cover the cost of producing and shipping them. The same holds true if the planner overestimates or underestimates the number of tomatoes each farmer can grow.
If several farmers in one area have been assigned to grow salad tomatoes, they will all want to plow, plant, fertilize and harvest around the same time, leading to a shortage of equipment and workers at some times and idling these resources at others. Then there’s the weather: If the picking and trucking schedules cannot adapt to early or late harvests, spoilage will increase.
Now let’s look at this problem through the lens of the three factors discussed earlier: pricing, incentives and decentralized decision-making.
First, centralized planning is inefficient because the planner in Leningrad is distant from the farmers and the shopkeepers. The farmer knows more than the planner about which varieties of tomato grow well on his land, what he needs to grow them, how many he can produce during an average season and when they will be ready to ship. The shopkeeper in Moscow also knows more than the planner about what kinds of tomatoes her customers want and how much they are willing to pay for them.
But the farmer’s and shopkeeper’s expertise and knowledge of local conditions are not easily communicated to the central planner. If he hears from every farmer and shopkeeper, he will be overwhelmed with data he is unable to integrate. So he sets up a chain of communication that creates a bottleneck, with only the most important information getting through. This is extremely frustrating for the farmers and shopkeepers, and after a while they are likely to give up advocating for changes that would make them more productive or keep their customers happy. They become chronically demoralized.
Meanwhile, the central planner has worked out a detailed plan for this year’s tomato supply based on data from last year’s harvest and sales. But even the best planner cannot account for every happenstance. Eventually, word might trickle up to him that cold and rainy weather in the Ukraine will delay and decrease the tomato harvest, or that Muscovites want more tomatoes and residents of Leningrad want fewer.
But by the time he has figured out how to adapt his master plan and communicated the necessary changes back down the bureaucratic chain, it may be too late to deal with the problem effectively. And during that lag time, something else may have changed that, once again, throws all his careful calculations into disarray. This is a recipe for planner/manager failure and burnout.
Now imagine that the planner’s boss must integrate the tomato plan with the plans for wheat, spinach, peppers and melons. She must weigh and balance many more factors. With this greater complexity comes a greater likelihood of error and a greater potential for damage from even one miscalculation. This higher-level planner also is even further away from the farmers and shopkeepers. Information is less likely to get to her in a timely way, thus decreasing her chances of being able to respond effectively to changing circumstances.
Second, in this centrally planned economy, there are no real incentives for individuals to change their behavior, even if the tomato harvest is poor. The farmer who produces a bigger harvest by working harder or acquiring greater expertise gets paid no more than the farmer who fails to meet his quota through laziness or inefficiency. The farmer who knows that spinach grows well in cool, rainy weather has no ability or incentive to shift part of his acreage from tomatoes to spinach.
The shopkeeper who works hard to procure good tomatoes for her customers cannot raise her prices to compensate her for her efforts. The truckers, warehouse workers and pickers get paid to show up, whether or not they have work to do or goods to carry. There are no incentives for any of them to adapt, innovate, work harder, be more efficient, or help the central planner. In fact, there may be many disincentives for rocking the boat, not least of which is the central planner’s fear that suggested changes to his plan will hurt his standing with his boss.
Pricing, the third factor, is the most important, because it connects decision-making with incentives. Prices do two things in a free market. First, they provide rapid, concise feedback to everyone from the farmer to the shopkeeper about fluctuations in supply and demand. Looking at the range of prices for a single stock or commodity over several years also allows everyone to see long-term trends, such as a growing demand for tomatoes or for particular varieties of tomato.
Second, changes in price provide incentives for people to respond quickly, efficiently and innovatively. For example, if the tomato harvest is later and smaller than usual, shopkeepers who want to attract customers will pay a higher price per bushel. If there’s a shortage of tomatoes in Moscow, warehouse managers will pay a higher price per truckload, raising the average wholesale price enough to compensate most farmers adequately for their costs. In addition, those farmers who can harvest more or earlier will earn a premium, giving all farmers an incentive to diversify their crops, try new varieties and improve their growing techniques. Pickers who get paid by the bushel will go where there are tomatoes to pick. Trucking companies that get paid by how much product they deliver on time have an incentive to deploy their fleets efficiently.
Best of all, the central planner/manager does not need to be all-seeing and all-knowing. He does not need to micromanage everybody or solve everyone’s problems. He simply needs to ensure that everyone gets accurate, timely price information, that they are free to act on it as innovatively and efficiently as possible, and that everyone plays by the same rules (no bribes change hands for inefficient behavior, no one is allowed to fix prices, etc.).
Of course, free markets are not infallible and central planning is not a complete failure. There are plenty of uncontrolled risks in the free market, and people in the free market make mistakes. Also, an excellent central planner dealing with a small number of people and projects may do a good job. But when changes in conditions occur in a free market, people find out quickly through pricing and have incentives and the ability to respond appropriately, without the delay inherent in waiting for a central manager to process the information, make a decision, and communicate changes down the chain of command.
Some of the parallels between tomato farming and pharmaceutical R&D should be obvious by now. But pharmaceutical companies face unique challenges, and it is important to understand these before applying our analysis to this highly complex industry.
THE NATURE OF THE BEAST: FROM FARMING TO PHARMA
Unique characteristics of drug research, whether at a small biotech startup or a large traditional company, make it difficult for pharmaceutical companies to benefit from the pricing information and incentives offered by the free market.
First and foremost among these is the long timeline – 10 years on average -- between the genesis of a drug and its approval by the U.S. Food and Drug Administration or other regulatory agency. This goes to the heart of the productivity problem, because managers and scientists who try new approaches must wait a decade or more for feedback in the form of market success and stock price. Few managers stay with a company that long. And few companies are willing to make big changes, then stay the course for a decade or more before seeing results.
At a small company with one or two drugs, eventually revenues and the stock price will have a clear and precise relationship to the drug’s success or failure, but in the meantime, investors may go on a roller-coaster ride of intermediary achievements and setbacks, with no guaranteed outcome. And even at small companies, the long development timeline makes it difficult to use revenues or stock price to judge the relative contributions of each scientist, team, manager or decision. This imprecision increases exponentially with the size of the company and the number of drugs it has in development or on the market, making stock price nearly useless as an internal feedback mechanism for the largest companies.
Second, the odds are very small that a given molecule will make it all the way to approval. Drug discovery is inherently risky because it is still very difficult to predict how drugs will act in the body until human trials. A molecule that seems safe all the way through animal trials might turn out to have unacceptable side effects in humans; other molecules that closely resemble toxic chemicals may turn out to be surprisingly safe and effective in clinical trials. Yet at every step of the development process, managers must make critical decisions about which drugs to pursue and which projects to drop.
The largest pharmaceutical companies are doing very well if they win approval for 3 or 4 new drugs a year, while start-ups may labor for years to bring a single drug to market. This means companies spend millions or billions of dollars and deploy thousands of scientists and support staff every year on projects that are destined to fail. Again, the larger the company, the harder it is for top managers to make rapid, informed decisions about allocating resources and for scientists to communicate their concerns about a project or switch to a different one.
Third, pharmaceutical research has entered an era of immense complexity. The rapid development of computers, combinatorial and computational chemistry techniques, and assay technology have led to an explosion of data. So have major scientific discoveries, such as the decoding of the human genome. This overwhelming complexity has been accompanied by increased scientific specialization. Two or three decades ago, one scientist could master most of the available information about a particular disease or biological pathway. Now managers must assemble numerous specialists and support staff in well-integrated research teams. The increase in the numbers and types of players and the greater complexity of R&D organizations increases the potential for mistakes _ and the potential ripple effects of each mistake. Again, all these difficulties increase with the size of the R&D organization.
Fourth, the nature of scientific discovery is very different from manufacturing, for example. Strong central management, strict standards and quality control and accountability at every level may work very well for manufacturers. But scientific research is a creative process that cannot be easily regulated or held to a strict timetable. Good scientists need enough independence, time, and support to play with ideas, experiment, learn from their mistakes and change their approaches. Often, the greatest ideas appear counter-intuitive initially, and these are often squashed or abandoned under a strong central authority. The more scientists are micromanaged, the fewer opportunities they have for the kind of creative problem-solving that leads to treatment breakthroughs. Yet as companies grow larger, managers often create more layers of bureaucracy and impose more accountability requirements as they try to get their arms around an increasingly unwieldy organization.
The problem has reached a crisis point for big pharma in the past few years, as serial mergers have created sprawling giants. Change has been rapid and much manager energy has been diverted from encouraging excellent research to reorganizing and meshing differing corporate cultures. Meanwhile, many scientists spend more time worrying about their jobs and personnel changes than pursuing ideas. Ironically, the faith in economies of scale that drove these mergers, while it works for pharmaceutical marketing, has created a greater drag on R&D productivity for all the reasons mentioned above.
Some managers at these large firms have given up entirely, scaling back their in-house R&D operations and focusing on acquiring or forming partnerships with smaller biotech firms. Others follow the latest management trends, such as reorganizing large R&D divisions into disease groups or biological systems groups. Some managers stick with what they know, relying on strategies that worked when they ran a smaller company. New managers brought in with “a mandate for change” are expected to make bold decisions, but may make cosmetic changes that don’t address the fundamental problems – or make rash decisions in the full expectation that they will be long gone before the consequences are clear.
Virtually all managers believe, privately or publicly, that small biotech firms are much more productive – and right now, they’re correct. Many promising young scientists are voting with their feet, shunning big pharma and taking lower-paying jobs at small firms. So what do the small firms have that big pharma doesn’t?
WHY SMALL FIRMS ARE BETTER, OR DAVID VS. GOLIATH
Let’s look at small vs. large firms through the lens of economics, focusing on the three key elements of a free market: pricing, appropriate incentives and decentralized decision-making.
First, small start-ups depend on funding from venture capitalists, hedge funds, big pharmaceutical companies or some other source. In this respect, they get much more frequent, immediate feedback from the market on how well or poorly they are doing than the Goliaths. When they get big enough to go public, they also get better feedback from stock prices. As we saw earlier, if a company has only one or two drugs in development, investor commitment and stock price will bear a much closer relationship to the success or failure of each drug than at a Goliath.
The free market, whether in the form of investment dollars or stock price, is a fairly neutral evaluator of companies and a disciplined allocator of capital. Investors who don’t make good bets most of the time go out of business. They rarely risk their capital or their reputations on personal relationships. They make it their business to know what’s going on at the companies they invest in and expect to see clear benchmarks of progress. When these are not met, they very quickly reduce or terminate funding, or sell their stock. Companies with projects that aren’t going anywhere quickly lose funding. Companies that can show progress draw more investment.
This means incentives for companies to work efficiently are appropriate. Good management and science are rewarded; poor management and science quickly lose support. Most start-ups and small firms also create appropriate internal incentives for good performance. Generally, they pay lower salaries than the Goliaths, but award bonuses based on the company’s ability to show progress and/or make stock options a significant part of compensation. That means everyone has incentives to help the company do well and no one has good reason to pursue research that looks like a dead end. Specialists and support staff have incentives to cooperate with each other and slackers are poorly tolerated.
Of course, money is not the primary motivating factor for some scientists. Many researchers are motivated by a strong desire to help people: They want to discover a better treatment or find a cure for a serious disease. Some are motivated by a desire to work in an environment where they can pursue their passions with a minimum of interference: They want the freedom to pursue new ideas, to be “in flow,” to create and discover without being micromanaged. Some want to work with a person or group that inspires them: They want to learn from a genius or mentor young scientists, they want to work in a collegial environment, or they want to collaborate with people whose talents complement their own. Some want to work for a company that tolerates unconventional personalities or methods: They do their best work at night, or solo, or with rock and roll blasting in the background. Small companies often give these scientists the autonomy to do what they want, how they want – as long as they are productive. That’s the bottom line.
Likewise, many managers are motivated by the desire to be a key contributor to a company’s success. Many prefer a challenging environment to a comfortable, predictable one. Some want to take risks, trying a new management strategy or approach. Some pride themselves on their ability to keep a diverse, temperamental group of geniuses working well together and communicating. Some scientist-managers want to feel like part of a team or keep close tabs on what’s going on in the lab. Many value the freedom to manage efficiently and change course quickly when a strategy isn’t working. Again, small companies offer these people opportunities they are not likely to find at a Goliath, even though Goliath might pay more.
As far as decentralized decision-making goes, small companies tend to be more collegial and less hierarchical, with ideas and concerns discussed informally among people at all levels. Often people wear several hats: a manager may also be a scientist who adds his ideas at meetings or works in the lab, and a scientist may take on some management duties, including explaining a project’s progress to investors. Even when small companies appear to have highly centralized decision-making, the lines of communication are much shorter. Scientists don’t have to wonder for long whether the CEO is a good listener: They’re probably talking with him often, so they can quickly judge whether their needs are being met and their concerns heard. Likewise, it’s hard for scientists to hide key information or a lack of progress from the CEO. This leads to “information symmetry,” another way of saying that in a 50-person firm, everyone knows everyone else and knows what they’re doing. Also, investors in start-ups may want to hear from the scientists, not just the managers – and that’s a great equalizer, even in nominally hierarchical firms.
Now let’s look at the Goliaths through the same lenses: prices, incentives and de-centralized decision-making.
We have already discussed the disjunction between stock price and the success or failure of individual drugs, decisions and people at the big firms. Big firms don’t have to disclose their progress until a drug reaches the human trials phase, and even then, public access to data may be poor. FDA approval is the next big public milestone, followed by the actual market performance of the drug, and savvy investors watch these closely. Still, when a company has dozens of drugs on the market and dozens more in development, the failure of one pharmaceutical in human trials is pretty much a blip on the stock market’s radar screen. Stock price trends over several years may tell investors a company is well-managed or in a slump, but that’s about it – it tells them little about particular individuals in management, except possibly the CEO. Meanwhile, the relative dearth of public information about these companies means their stock prices are more easily buffeted by rumors, speculation, and other factors beyond their control.
Internally, these companies usually pay very handsome salaries, but meager bonuses and no stock options. Inotherwords, salaries are not tied to company or individual performance. The scientist who discovers a blockbuster drug will certainly gain the esteem of his peers, but he is unlikely to get a bonus topping $25,000 or be holding stock options worth millions of dollars. He may get a salary increase, but his pay will remain only incrementally higher than that of other scientists with the same job description and experience. His reputation will soar within the company, but outside it he will probably remain obscure.
For some scientists, this may be enough – especially if their status wins them greater freedom to pursue new projects of their choosing. Without prices, however, companies do not have the tools they need to create appropriate financial incentives for scientists or managers.
This problem is aggravated by the “information asymmetry” that results from centralized decision-making in a big company. The head of R&D cannot possibly keep close tabs on the thousands of scientists and hundreds of projects she oversees. Some have recognized this and have tried decentralizing their organizations, breaking them up into divisions by disease or body system: There may be divisions for heart disease and cancer research, or divisions for drugs affecting the circulatory system and those acting on the nervous system.
By Dave Webster, with Katharine Webster
November 2006.
