The pharmaceutical industry would like you to believe that they fund the expensive development of new drugs, they get patents in return, and this provides a nice efficient market driven mechanism for developing new drugs. There are many things wrong with this picture:

• In-patent drugs cost several times what an out-of-patent generic costs, but only about 15% or so of this goes into research. This is not an efficient way of funding research.
• Despite this supposed market incentive, governments still have to fund research with grants and tax breaks. They also push pharmaceutical companies to develop particular drugs with one-off incentives such as orphan drug designation. This is not a free market system.
• It is more profitable to market treatments that are not cures. A treatment (that controls symptoms, for example), gets repeat prescriptions for longer. This is, perhaps, why pharmaceutical companies have developed lots of expensive ulcer treatments, but it took Barry Marshall and Robin Warren developed a cure for a common type of ulcer. This is also why they like mental health drugs, even those as dubious as ADHD treatments.
• It is less risky for pharmaceutical companies to develop drugs that are similar to those that already. As a result there are many drugs in families such as statins and proton pump inhibitors.
• It is even less risky to develop variants on existing drugs: almost all pharmaceutical companies keep a brand partly in-patent by developing minor variants, such as extended release versions, close to patent expiry.
• In order to release variants so as to most effectively lengthen the time a company has patents related to a brand, there is an incentive to delay there development, or at least publication.
• An even better, and cheap way, of boosting profits is “disease mongering“, marketing aimed at encouraging people to take drugs for mild or non-existent conditions.
• Many small drug companies have a life cycle that starts with an academic research founder and ends with deals with, or acquisition by, a larger company. The government funding of the academic research means the government underwrites the risk (that the research will not lead to anything), but the private sector takes the profit.
• Third World countries can not afford to buy large volumes of in-patent drugs at high prices, so their markets are not big enough to it to be worth drug companies bothering.
• The system fails to fund new uses for out of patent drugs. This is happening with Dichloroacetate.
• It also fails to fund other unpatentable improvements in medicine, such as new surgical techniques.
• A large part of the cost of drug development are clinical trials, which test safety and effectiveness. These are paid for, and carried out by, pharmaceutical companies. What biases does this introduce?

Added 23 Nov 2009: two more problems created by bad incentives covered in a new post.