This paper tackles the usual figure of 1.3 bil USD and brings it down to about 40 million (!).
Source: The London School of Economics and Political Science, 2011
There has been a reply from DiMasi, and a reply to the reply by Light. They are quite fun and interesting to read. See below.
DiMasi (Tufts) response [PDF]
Light response to DiMasi’s response (hopefully your organisation provides you with access. Otherwise, email us and we’ll see what we can do)
…BUT ULTIMATELY, the area is very grey due to the undisclosed nature of R&D costs. See this Evidence.