#expbio ASPET Blogging: Stabilizing a productive but structurally unstable research system

Figure 1: Alarm - boom - bust cycles 

destabilize the scientific research system 

in the US.

The scientific system in the United States is highly productive -- predominant globally -- and yet is showing many signs of instability. Flat funding levels, declining success rates for federal grant proposals, and a disconnect between supply and demand of trainees all contribute to the system’s instability. Dr. Michael Teitelbaum, a demographer with the Labor and Worklife Program at Harvard Law School, explained why these problems have arisen and why they are not being corrected.

Contrary to what popular media would have you believe, the United States is not currently on the brink of falling behind other countries in STEM fields. This fear of falling behind is not new though. Dr. Teitelbaum explained that there have been five cycles of alarm -> boom -> bust since World War II.

1) Booms in physics, especially nuclear (until 1957),

2) post-Sputnik (1957 – early 1970s),

3) Reagan era (1981 - early 1990s),

4) High-tech “bubbles” [internet, IT, telecom, biotech (~1995 - 2005)], and

5) NIH budget doubling followed by “funding crisis” (1998 - 2008).

In each case, the government responded to alarms that the US was falling behind in science and engineering, either by pumping more money into the system (increases that later ended) or by expanding temporary visa programs. In the most recent cycle relevant to biomedical research, the NIH budget was doubled in only five years (1998-2003) because grant success rates had been declining (see Figures 2 and 3), and arguably, also to spur technical advances that would stimulate economic growth. During this time, success rates did improve and labs prospered: hiring more staff, postdocs, and bringing in more graduate students. But the budget increases of 14% per year had been promised to last only five years, and from 2003 they suddenly went flat and have fallen well behind inflation since then… leaving labs and trainees in the lurch (Figure 2).

Figure 2: The NIH budget increased overall from 1960 to 2003 before falling off. The doubling (from $13.6B (1998) to $27.3B (2003) is portrayed within the red rectangle.

Positive feedback

The current biomedical research and training system has strong positive feedback loops that make it prone to instability, while it is deficient in stabilizing negative feedback that would cause it to self-correct. Sources of positive feedback include the fact that 78% of NIH-supported PhD students and postdocs are being financed from research grants rather than under training programs or fellowships, so increased research budgets produce more funded slots for PhD students and postdocs. (About 86% of NSF-supported graduate students are supported by research funds.) These increases in slots occur if research budgets rise, even if there is no increase in demand in the labor market for recent PhDs and postdocs, leading to disconnects between rising supply and non-rising demand.

Booms and busts in research funding also lead to ups and downs in success rates for federal research proposals. More research funding begets an increased number of proposals, leading with a delay to decreased success rate for those proposals (Figure 3, line graph).

Figure 3: Increased funding led to decreased success rates.

Other elements of the current system incentivize research institutions to expand their research facilities and staff assuming future increases in research funding that may not appear. There is no limit to the percentage of researchers’ salaries that can funded from NIH research funds (though there is an overall dollar limit per investigator), incentivizing an increasing number of institutions to encourage or require faculty to secure their own salaries with NIH grant funding (aka “soft money”). Some institutions require >80% of a faculty members’ salary to be paid for by grants. In contrast, NSF will only pay 2/9^th of a 9 mo salary, which means fewer incentives to expand research faculty when soft money is abundant.

Negative feedback

In a closed system with full information, these positive feedback loops would be balanced by negative feedback.

If funding were to decrease, such negative feedback loops would cause the system to “adjust” as PhD/postdoc stipends declined in terms of inflation-adjusted dollars, and as talented US students realized that career prospects were deteriorating. Institutions would be expected to respond by improving their offers to a smaller number of the very best PhD prospects, while other would-be PhD students would choose to pursue other careers. But the current system is neither closed, nor does it provide full information. With rapidly increasing numbers of science graduates and PhDs in lower-income countries, PIs can recruit increasing numbers of PhD students and postdocs so long as they can fund them from federal research grants. The (surprisingly weak) data on postdocs suggests that a majority of this expanding pool now are international postdocs. Meanwhile prospective PhD students and postdocs, both US and international, are able to obtain only limited information from US universities about the career experiences of their recent graduates. Whether the current PhD/postdoc system is really “broken” or not depends on your point-of-view. Is it a US system, or a global system? 

A final source of instability is due to the very different time horizons of basic research and research funding. Basic research is very much long-term, while federal budgets are  adjusted annually, and with a history of erratic increases and decreases leading to the boom-bust cycles. Stable funding would help to stabilize the system, but would not resolve its structural problems.

How to fix the system

• A longer-term horizon for basic research funding, with steady increases rather than booms and busts.

• Consideration by NIH of limits on the percentage of salary that can be paid from research grants; many have suggested moving gradually to a cap set at 50%. This would be stabilizing because universities would have fewer incentives to encourage their researchers to fund larger fractions of their salaries from soft money. 

• Reduce the number of trainees that can be financed by research grants. At least 50% of trainees on traineeships would be better for several reasons: encourage better graduate education due to requirements built in to such training grants; limit the positive feedback from research funding increases to expanding PhD and postdoc numbers; and could improve connections between supply and demand if enhanced reporting were required on career outcomes.

• Increase transparency to students and trainees at all levels regarding career outcomes.

• Clarify goals of current policies that fund large and increasing percentages of international students and postdocs.

Dr. Teitelbaum said these changes need to come from those within the system: biomedical scientists who are concerned about instabilities in the system  and are willing and able to encourage discussion and suggest interventions. Such discussion would best be led by biomedical research leaders and by early-career scientists, who together have a comparative advantage in identifying the best ways to stabilize the system. For further reading, check out Dr. Teitelbaum’s book, Falling Behind: Boom, Bust, and the Global Race for Scientific Talent. (Princeton University Press, 2014)

Katie would like to thank Dr. Teitelbaum for his input in the writing of this post. 

Figure sources: alarm by Pearson Scott Foresman, stacks of money by Allureme, empty pockets from ISME. Design by Katie Collette.