There are many reasons to avoid taking the King's shilling if you can. Chief among them is that politicians are extremely fickle taskmasters. This makes it risky to do any business with the government that requires substantial up-front investment, like a new factory or the next eight years of your life. This is a problem with corporations too, obviously--indeed, the requirement for co-specialized assets is one of the few times when a merger definitely makes good economic sense. But companies don't routinely turn over their management every two-to-four years.
This has recently become a problem for healthcare research, thanks to a funding boom and bust, and the government's poor system for fund allocation:
let's go back in time to the golden era between fiscal years 1998 and 2003, when a bipartisan effort succeeded in nearly doubling the NIH budget. Paylines loosened up, and the spigot of money government money and its associated indirect costs seemed never-ending. Truly, times were good for a brief time. Unfortunately, two problems in addition to the flat NIH budget from 2004 (courtesy of the Bush administration and the war) on set the stage for our present crisis. The first was that there appeared to be little planning for how to cover the "out year" commitments of all these new grants being awarded. What many don't realize is that a five year R01 grant is in reality five one-year grants. The money isn't in a bank to cover the five years, each "out" year of the grant is covered from the NIH budget of each fiscal year the grant encompasses. If a grant is awarded in fiscal year 2008, for example, then it will in reality be five grants in FY 2008, 2009, 2010, 2011, and 2012. Even before the five year doubling was well under way, there was concern about preventing a "hard landing" in FY 2004, the first year after the doubling of the budget occurred and all those extra grants funded in the heady days between FY 1998 and 2003 were still on the books and had to be funded all the way through FY 2008. The concern, as experience shows, was justified. When the budget went flat and, adjusted for inflation, even started to decrease slightly, the result was predictable: The out year commitments from grants funded during the doubling started to squeeze out funds for new grants. Moreover, the NIH couldn't save any of this windfall for a rainy day; by statute it has to spend its budget each year, as described in Science and Inside Higher Ed. The NIH did a number of things to try to keep the percentage of success from falling too far (cutting 23% off of the budget of my grant and those of many others right off the top and then cutting 2-3% per year), but that only kept the situation from becoming even more dire than it is now.
The second issue is the universities themselves. The doubling of the NIH budget led to a recruitment and building boom at research institutions all over the U.S., as universities sought to capitalize on the increase in the NIH budget.
A steady, slow increase, he says, would have been better than a doubling followed by a flatline; now the old grants are crowding out new research, and possibly crippling the careers of young researchers who can't get onto a project. This suggests that the public should have a preference for politicians with modest promises, rather than radical new plans. But in the case of things like scientific research, this is emphatically not the case.