The Benefits of Funding Basic Scientific Research

[corky]

I recently heard a talk by a guy from the AAAS (that is, the American Association for the Advancement of Science) about Science and Society. In the Q&A section of the talk (given to astronomers, by the way), a lady asked what she should say when a member of the public asks "What is this good for?"  The speaker's response was simply, "That's for you and your field to figure out."  I thought this was interesting and have started doing some research to answer this question to my satisfaction.  In these (ongoing) notes I will mainly try to answer the question "Why should the public finance basic scientific research?", where "basic sci research" is defined to be that research which is conducted for the sake of the pursuit of knowledge. That is, research not motivated by any immediately apparent application to the private sector.

This is a work in progress and I would be grateful for any input/suggestions/examples you may have.

A lot of the examples and statistics I have gathered from The Simple Economics of Basic Scientific Research, Economics and Social Significance of Science and Engineering Research, and The Economic Benefits of Publicly Funded Basic Research: A Critical Review.  

So why should we fund basic research?  I think the response favored by many doing that research is simply this:  it is a worthwhile goal to understand more about the world around us and it is inherent human nature to seek to do so.  Knowledge for knowledge's sake.  I think this is an entirely satisfactory answer and most people seem to "get" it.  However, this is a very "touchy-feely" kind of response and it is hard to quantify the "value" of knowledge when allocating funds.  For example, the "pursuit of knowledge" and "enrichment of society" angle is probably used by people seeking funding for the arts, but they don't get any where near the funding science gets.  One thing to keep in mind is that allocation of funds is essentially a zero-sum game - one project gets funded at the expense of some other project.  How much do you value "understanding the universe" in comparison to military spending?  or a social safety net?  or domestic infrastructure?  I think it is crucial to always stress the humanistic worth of understanding our universe, but I think we also need to be able to give convincing economic motivations to basic research funding that will allow us to compare its "value" with that of other spending projects.

So what are the economic benefits of funding basic research?  At first glance, this appears to ask "what is the dollar value of knowledge?".  What's the dollar value return on knowing  how stars explode, how black holes merge, or how superconductors work?  Probably not a lot for the first two, maybe a fair deal for the third.  Obviously not all information has equal worth on the market.  However, I would argue that the final scientific result of a study is not the main source of economic benefit from basic research.  Instead, the economic returns from investing in science arise as by-products of the messy process of doing science.  Science produces unexpected results that aid society, develops new instrumentation for a specific purpose that may have wide application, and requires a fairly advanced infrastructure that motivates technological advancement and improvement.

Let's consider some examples of the unexpected results that turn out to be important (most of these taken from the "Simple Economics..." paper).  Pasteur inadvertently stumbled upon the concept of inoculation while studying cholera in chickens.  He accidentally gave a group of chickens a weak batch of cholera and the chickens survived.  Not wanting to waste the chickens, he later reused them and administered to them a stronger (and lethal) batch of cholera.  The chickens again survived, having been inoculated against the disease.  The concept and practice of vaccination has had an enormous effect on the growth and productivity of society.  

What about the mainstay of modern medical diagnostics, x-rays?  They were not discovered because doctors were looking for a way to peer at bones in the human body.  Instead, their diagnostic powers were discovered at around the same time that x-rays were, by Roentgen.  He was conducting experiments at the cutting edge of physics with the new x-ray and happened to find that it allows one to see through the human body (or, more specifically, his wife's hand).  

Or the workhorse of just about every electronic thing you own -- the transistor.  A couple of guys testing some properties of germanium see something neat and that property is exploited to make the component most essential for the economic and technological growth of the twentieth century.

[Aside:  I would really like more examples of this "accidental breakthrough" kind of thing if anyone's got any.]

But it's not just that science is a gamble where rarely something is discovered that makes a huge impact on society.  In fact, most scientific enterprise consistently produces methods and instruments that can be used in the private sector.  Think about electron microscopes, superconducting magnets, and synchrotron radiation sources.  These are created and used out of necessity in science.  Eventually the private sector may adopt them for their own purposes, with the added benefit that they methodologies for using these new technologies have already been developed.  

In addition to making new instruments, the process of doing science often requires the installation of cutting-edge infrastructure.  A bunch of high-energy physicists studying things which they could boast had "no real-world applications" wanted to share electronic versions of their papers across long distances and this sets the groundwork for the Internet.  Particle physicists create a huge accelerator outside Chicago to study matter on the smallest scales, something which should not affect anyone's day to day life.  Currently, Fermilab has a growing number of medical applications for particle accelerators, using neutron beams to kill cancers.  As science continues to become more and more data-intensive, large science projects will motivate the construction of high-speed internet connections and high-powered computers that will certainly affect economic productivity.

I hope that you'll now agree that basic scientific research has had an impact on the economy, but can we quantify this effect at all?  Ideally, we would like to know the marginal social utility of funding science.  That is, if we increase science funding by one unit, how does this translate to the useful output of the research?  Sadly, we don't know this.  However, there have been some studies that have tried to measure the rates of return on basic scientific funding through the private sector.  Work by Mansfield (described in the "Economic Benefits..." paper linked above) tried to do just this.  Mansfield surveyed the private sector to find what percentage of their products could not have been created (ever or without a significant delay) without the use of recent (last 15 years) basic academic research.  He found that roughly 10% of products made during this 15 time period could not have been made without the use of recent basic research results.  From this figure, Mansfield estimated that the rate of return of scientific research is 28%.  This means that an investment in basic research (that is, the allocation of funding) will return 28% in profits each year.  Though the exact figure is based on a set of shaky assumptions, this work shows that funding in science produces a significant return on investment.  

(Though the above results seem to indicate that science investment provides significant returns now, it does not necessarily mean that adding more funds will give the same result.  For this we would need to know the marginal utility for science, which is not known.  However, see the "Simple Economics..." paper listed above for a convincing argument that science funding is currently below where it "should be.")

Alright, so it seems as though investing in basic research provides significant returns.  But wait!, if investment in science produces such great returns, why hasn't the private sector done more for basic research funding?  Let's sit back and let the free market decide.  Well, let's consider this proposition briefly.  A thorough rebuttal of this view can be found in the "Simple Economics..." paper, but I'll outline the gist of it here.  Though I have cited a very high return above, that return is to society as a whole and not to any one particular firm.  If we leave our basic research to individual firms, they may not have enough motivation to invest.  Basic science research provides very important but very rare results which average out to give nice returns.  If we consider a small firm selling a particular type of product, it will simply not have enough capital to invest broadly enough in basic research to get profitable results.  Instead, only very large (and almost monopolistic) firms will have the capital to invest in this enterprise.  Think of Bell Labs or Du Pont chemical company.  They were large enough to absorb the small losses of funding basic research long enough to reap the eventual rewards.  If we advocated that basic research be conducted only in the private sector, this would eventually force large monopolistic firms into existence to make it profitable.  This works to undermine our free-market system!  Stated nicely in the "Simple Economics..." paper, "if we want to maintain our enterprise economy, basic research must be a matter of conscious social policy."

OK, so now it seems evident that funding basic research does have a positive effect on the economy and cannot be funded just through the private sector.  This means that we must invest in basic science research through the government.  So are we investing enough?  I point you to the "Simple Economics..." paper again for a full answer, but in short, no.  The argument goes something like this.  If the private sector is funding any basic research and taking a fairly large risk in doing so, then the public sector is not funding basic research "as much as it should."  This comes largely from the fact that publicly funded research is "free" to the private sector and funding research privately is fairly expensive.  Therefore, current public funding is low enough that (some) private funds have the incentive to take large risks to make up the deficit.  

OK, that makes sense, but what about in these tough economic times?  Surely keeping funding level and not cutting it is as good as you're going to get and we shouldn't be throwing our precious few resources into something as frivolous as science, right?  No way!  We should always keep in mind that keeping a "constant" level of funding is essentially a cut, because it does not take into account inflation.  A true-constant funding level would need to be something like a constant percentage of GDP or something of the like.  Additionally, we should keep in mind the "latency period" for payoff in basic research.  Studies have found that it takes somewhere between 10 and 30 years for basic science results to have an effect on the private sector.  So by cutting (or keeping fixed) our science budgets now,  we are cutting future productivity.  This is not a good thing.  Finally, we should always think of funding science as an investment, both literally and figuratively.  If the above 28% figure is to be trusted, investing in science is one of the biggest returning investments you can make!  (Treasury bonds will get you a couple percent, diversified stocks might get you 10%).  Cutting this funding makes us less productive in the future.

It seems as though an investment in science provides a real tangible economic return that has not yet been fully exploited.  So fund science!

[Yariv Yanay]

Just to play Devil's Advocate for a moment -

I think you're applying the tag of "basic research" a little too widely;
as in, you're using the benefits reaped from semi-practical fields like
condensed matter physics and to justify research in
fundamental-questions science like astrophysics. Both of these can be
called basic research, since they have no immediate, tangible benefits,
but the sort of things done in the basement of Clark are 10-20 years
away from being used in industry while what's going on on the top floors
of Space Science may never have any practical use (and almost certainly
will not in our lifetimes).

That still leaves the developing-tools angle, but I suspect that you're
not getting anywhere near 28% return on your investment there. (unless
you chalk up a large part of the internet to the existence of particle
physics, in which case we may be able to coast just on that for another
fifty years or so)

[corky]

The nanotube people will have a much greater impact (and have greater economic return on their investment) than the gravitational wave people.  I completely agree with that.  They also get much more funding, so their is some "funding distribution function" that tends to fund some projects more than others, probably due to perceived economic returns.  I'm not necessarily advocating changing the shape of this distribution, just multiplying the whole thing by some factor (preferably greater than unity).

In fact, we should still maintain a very broad base of funding.  If we neglect the (valid) philosophical argument of knowledge for knowledge's sake and consider the problem as one of investment, it may seem tempting to just put everything in the (relatively) high-return condensed matter experiments.  This argument fails to take into account two main things:  the very rare very large return discovery that can occur in any field and the lower-return but constant growth associated with doing science.

In the first case, we would want to cast as wide a funding net as possible because almost any field can produce a huge discovery that could have major impacts.  Pure biology studies can produce important medical findings by accident, theoretical physics can produce findings that get realized in products in a few decades (the results of quantum and relativistic physics in the 20s or so started paying dividends by the 80s), and even math can help you win Netflix competitions (I'm sure it's useful for something else too...).  Also, it's important to keep in mind that some of people perceived to be the least economically useful (e.g. high-energy theorists) are also the cheapest.  A salary, laptop, and Mathematica license isn't too hard to fund.

In the second case, I think that the consistent returns from methodology and instrumentation are probably a bit higher than one may think.  Sure, astrophysics (to pick a target) isn't going to be as immediately useful to industry as condensed matter experiment, but it's not going to be negligible.  As far as I know, astronomy and astrophysics consist of telescopes and computers.  Getting better telescopes to explore the universe is a highly significant technological accomplishment.  The Hubble telescope has been up for two decades now and required a lot of work to make something that durable.  The technologies perfected/invented for Hubble (better mirrors, better CCDs, various space technologies) will have an impact on a wide array of non-astronomy industries.  LIGO is another good example.  It's goal is to detect gravitational waves that will have exactly zero effect on the economy.  BUT, to detect these things, they needed to evacuate to vacuum 4-km long arms and create all kinds of new mirrors and sensors to detect something we really have no right to detect.  These new technologies are applicable everywhere.  The computers necessary to process and store all this data push the cutting edge and allow new technologies to get tested and provide a willing consumer whose purchases will help drive down the cost of these technologies for everyone else as the manufacturing process improves.

In essence, all of this science funding is a low-risk investment in which the government can fund the development of new technology and methods.  They don't know what they are going to get, but the odds are they will get something.  In the worst case, we "only" learn something new about the universe.  Compare this to when the government invests in a specific technology (or company).  Their goal is to reduce costs enough for the company to make private motives (profit) coincide with some social good (say, the development of clean energy sources).  This is a good goal, but when such a company fails people throw a hissy fit because they consider it a "lost" investment.  At least with science we learn something.

So I rambled on their for a bit and may not have addressed your actual point.  I guess my response is that you're probably right, but don't underestimate the returns of instrumentation and development even for fundamental-questions science.  The economic returns on the informational content of the CMB may hover around zero USD, but to get that info people had to build really impressive tools like COBE and WMAP.  Those certainly had economic benefits.

[Yariv Yanay]

So let me start with a short preface - the country we live in
(specifically the US but also western Europe) is a very rich society
whose main economical problem seems to be finding a sustainable model
for distributing the wealth it has. Even if I think certain branches of
science today are research of knowledge for knowledge's sake, that
doesn't mean I think we're overspending on science. Sending out another
Planck satellite certainly beats buying another F-35 in terms of benefit
to society. So this is more of a philosophical discussion on our true
worth to society rather than any kind of policy advocacy.

I picked on astrophysics because that's your field, but really I was
thinking of the LHC, whose $10 billion cost I don't think we're going to
recoup in benefits. Certainly mathematicians don't cost a lot of money,
nor high energy theorists, but high energy experimentalists cost a lot
of money, as do, to a lesser degree, space telescopes and things like LIGO.

And these fields, I think, are very unlikely to have one of those
low-probability, high-return discoveries you suggest. (or rather, the
expectation value of such returns is very small) All of the science in
your examples was basically done on a table-top level, while particle
physics now requires billion dollar machines to make the slightest
advances; it seems to me that even if we find a practical use for quarks
(discovered almost fifty years ago now) it's unlikely that the device
required to manipulate them is going to be worth building.

This is not to say that there aren't fields that fall into the
50-100-years-away-from-practical category; those are worth investing in
for the low-probability returns.

That leaves the side-benefits argument, and it's a valid one. My
original point was that we should separate the returns we get that way
from the returns we get from more practical, 20-years-away, science,
that probably gave the 28% annual return figure you quoted in the
previous email. There certainly is a return here, and it may be high
enough to warrant the investment - I don't know.

The flaw in this argument is that one does not have to focus on science
to get side benefits. We could pick any kind of very difficult task,
(I'm partial to the creation of 10-meter tall dinosaur-shaped fighting
robots, myself) hire a lot of smart people to work on it and get the
same benefits. Basic science is convenient in that it feels less
arbitrary, but it is, in the end, arbitrary. So any returns have to be
measured against the other ways we have of investing that money, as a
society.

Yariv