Reversing the downshift in federal R&D spending: My long-read Q&A with John Van Reenen

is factory productivity growth slowing in the US, and how can we reverse this
trend? What does public R&D spending accomplish that private spending doesn’t?
And how close should we come to an “industrial policy” for innovation? John Van
Reenen joined the Political Economy podcast to discuss these questions and more.

John is the Gordan Y Billard Professor and Management and Economics and Professor of Applied Economics at the MIT Sloan School of Management and MIT Department of Economics. In June 2020, he published “Innovation Policies to Boost Productivity” as part of the Hamilton Project at the Brookings Institution.

What follows is a lightly edited transcript of our conversation, including brief portions that were cut from the original podcast. You can download the episode here, and don’t forget to subscribe to my podcast on iTunes or Stitcher. Tell your friends, leave a review.

Pethokoukis: All right, let’s start out very simply. What
is the problem that you are trying to solve here with your idea?

Van Reenen: Well, there are twin problems, I guess. The
big picture problem is that even before the COVID pandemic, America was
suffering from a productivity growth problem. Productivity growth had slowed
down since the Great Recession. And productivity growth, in the long run, is
the key to having sustainable wage growth. So if you want to get better wages,
you need better productivity. That’s the first problem.

The second problem is that there’s a group of challenges —
innovation challenges, big challenges — facing the US and all major countries
around the world related to (obviously) health and the pandemic, but also
related to the environment, like climate change. So these are twin problems. The
problem of low productivity growth and the problem of environmental and health challenges
are the problems that I think I want to tackle here with my proposal.

What is your explanation for our historically weak
productivity growth? If 1973 was the end of a productivity boom, why did we see
this downshift?

Yeah. Well, I don’t think we know for sure, exactly, all
the reasons why this has happened. I mean, there are a number of different
possibilities out there that people have discussed. For example, there was a
lot of rebuilding after the Second World War, which gave a big productivity
boost, and that kind of ran out of steam. In the mid-’70s, you had the oil shocks,
which was also a negative effect. I think there’s a long-run problem, if you
look at the productivity growth in the US, that we’ve increased our investments
in R&D, which is one of the factors we’ll talk about as a key factor in
increasing productivity growth. But the return that we’re getting for the
R&D doesn’t seem to be as great as it was in the past.

We still spend a lot of money. We’re spending a lot of
money every year on R&D, right?

We are. We spend about 2.8 percent of our GDP, or
something, on R&D. And of course as GDP goes up, that amount goes up. But
if you look at the composition of that R&D and where it goes, then there’s
been a big shift. Federal R&D, the amount of money which the federal
government spends, has decreased a lot. In the mid-’60s, we were running at
about 1.9 percent. Today that’s down to about 0.7 percent of GDP.

Via Twenty20

So what’s happened is that business R&D — private
sector R&D — has increased, which is good. But the concern is that a lot of
federal R&D is going for the basic type of R&D, the kind of R&D
which creates a lot of what economist call “spillovers” to the rest
of the economy. Some of the business R&D is more near-market, so it has
value for the business itself but may not create as many spillovers to the rest
of the economy. So one of the concerns is that the fall of federal R&D, of
basic R&D, is one of the reasons why we’re not getting so much bang for the
overall R&D buck.

We recently had Dietrich Vollrath on the podcast, and he was completely unconcerned about weaker productivity growth. He thinks it’s happening because of a shift in the economy from manufacturing to services, and it’s just harder to improve the efficiency of a service than it is to produce a product. I assume you think slow productivity growth is a problem, so what is wrong or incomplete about this view?

He could be saying different things. One is, he could be
saying that while real productivity growth is increasing, real GDP is
increasing, but we’re not measuring it properly. So because the economy has
shifted into services or other things where it’s hard to measure, we’re getting
this real productivity, but it’s not actually being measured properly. That’s
one argument.

The other argument — that we shouldn’t care if the real
consumption or real growth of the economy is going up — seems a bit weird. I
mean, I think you could say that we’re not measuring things properly. Maybe
quality has improved faster than we do. There’re lots of free goods. So that’s
a completely serious argument. I think that’s probably where he’s coming from,
and there’s a lot of truth in that.

Yeah, well, it’s just that we have the shift into service
sector work, which inherently tends to have lower productivity growth. If more
of the economy is that, then it shouldn’t be shocking that there’s lower
productivity growth overall. It’s not necessarily we’re doing anything wrong.
In fact, we might be doing something right, if that’s the kind of economy we

Well, that would be an economy in which you had slower
wage growth, slower increases in people’s incomes, less money to spend on
health. I mean, you might say, “Well, we’re in decline, but it’s
inevitable. There’s nothing we can do about it. That’s just life.” That’s
a view.

I don’t share that view. I don’t think we should be so
pessimistic as to say, “Well, there’s nothing that we can do about those
things.” If you look at another part of the service economy — if you look
at Amazon or Google or Walmart — they’re all parts of the service economy. They
have actually had periods where they’ve had very strong productivity growth.

A truck with the logo of Amazon Prime Delivery is seen at the Amazon logistics center in Lauwin-Planque, northern France, April 22, 2020 after Amazon extended the closure of its French warehouses until April 25 included, following dispute with unions over health protection measures amid the coronavirus disease (COVID-19) outbreak. Via REUTERS/Pascal

In the late ’90s, there was a lot of productivity growth
happening in the retail sector as they were using information technology very
intensively. So I don’t think it’s inevitable that we have to have low
productivity growth in parts of the service sector. I think the problem is
measuring it properly. We may not be measuring the innovations that are created
in parts of the service sector as well as we do in the manufacturing sector.
Yeah, as I said, there’s some truth in that. But even when you try and correct
for all those measurement problems, you still see some decline of productivity
growth, especially since the Great Recession. It’s hard to believe that those
measurement problems got so much worse around 2008.

For sure.

So I think there’s some element of that, but I don’t think
that’s the whole story. We could sit back and be relaxed and say there’s
nothing we can do about it. But I, personally, think that’s too pessimistic. I
think that there are things we can do, and I think there are ways that we can
reinvigorate the economy so we can get decent wage growth again.

You were saying that spending on R&D is not all the
same. So what does the mix of government spending on research look like, and
how do you want to change it, if at all?

Well, I think my main proposal is that we should have an
ambition to increase the amounts of resources that we give towards innovation
and towards R&D. Not to go all the way back to where things were in the
mid-’60s, but to have an ambition to raise it by about a half a percentage
point of GDP to about $100 billion.

About $100 billion a year?

Yeah. I don’t think that all that should be spent in the
same way. I think that the way to think about this is to look at the evidence
of what types of innovation policies are most effective in raising innovation.
In my proposal, I give some indications about the types of policies we know have
had an impact. Part of this may actually be changing the tax system in order to
provide bigger incentives to innovate. The US, under Ronald Reagan, introduced
a research and experimentation tax credit in the early 1980s, and there’s a lot
of evidence that this was successful in raising R&D and, indeed, innovation
for many firms.

So I think that that’s one pot of money which could be
improved and expanded. There are also direct grants which go to different
companies. For example, the Small Business Innovation Research Fund. That’s a
federal program focusing on smaller businesses, and there’s some great evidence
out there that it has been effective in raising R&D and crowding and
venture capital funding for many small enterprises — in the energy department,
for example.

Some of these grants would go to companies, would go to

Yeah, absolutely.

Would Google get a grant? Or Amazon or Intel? Would big
companies get grants too?

Yeah, absolutely. In fact, a lot of the R&D tax credit
does go towards large companies.

But would they get these direct grants?

There’s a lot of detail about how you decide how the
grants are allocated. I’m not going to say that company X should get the grant
and company Y shouldn’t get the grant. Part of this is setting up a system for
allocating grants based on different criteria. If a company who applies fulfills
that criteria, then, by all means, that company should get the grant.

And the grants would be to do what? Would they say,
“We want to do very basic research,” or, “We have an idea that’s
not really basic, but it’s a little further along and we want money for
that.” What would be funded?

We want to try and think about funding types of research
which create these spillovers, which are things which benefit not just the firm
itself but, potentially, other firms. We should also think about these
challenges which the economy faces as a whole. We face health challenges around
pandemics and viruses. We face challenges around environmental things like
pollution and climate change. We face a number of other different strategic and
military challenges, like thinking about what’s happening with China.

So there’s a set of priorities which are set by people and
politicians, and then there’s a question of who can best meet those challenges.
Companies can, like they do, bid for different types of funds. And then there
has to be a way of allocating. The principles of allocation should be that the
allocation is based on having agencies which are not just controlled by
politicians who are trying to give out the money to people in their local area,
but people who are more independent from the government.

And they should be evaluated very rigorously to make sure
that we’re getting good value for money for those things. There are lots of
programs which try to do that in a robust way. Nothing is perfect, but I think
there are ways of doing that to reduce some of the risks. The obvious risk is
that these become things which become captured by the people who are getting
the money or captured by the politicians. You have to do a lot of things to try
and shield that. And you think about the way the NIH or different types of
agencies use the grant money.

Dr Francis Collins, Director of the National Institutes of Health (NIH), holds up a model of SARS-CoV-2, known as the novel coronavirus, during a U.S. Senate Appropriations Subcommittee Hearing on the plan to research, manufacture and distribute a coronavirus vaccine, known as Operation Warp Speed on Capitol Hill in Washington, D.C, U.S., July 2, 2020. Saul Loeb/Pool via REUTERS

Would the NIH and the National Science Foundation be
getting this money? Would they be directing where this money would go? What
would be the roles in the existing infrastructure?
Or is it a new agency? I don’t

Well, I think there are two things. One is that you would have more funds for some of these existing agencies. But I think if we really went big on this, we could think about setting up a larger, new agency to deliver some of these funds, built on some of the principles that exist in some of the current agencies. So there’s a difference. I don’t want to say exactly how to do this, but there is a set of proposals out there by, for example, my colleagues at MIT, Simon Johnson and John Gruber.

Professor Gruber has been on the podcast.

Oh, well, you know what he’s proposing, then. He’s
actually setting up a new agency to try and deliver some of the money, and one
of the things that he emphasizes with Simon is that, in terms of who gets the
money, we should think about the places
which get the money. So there’s a risk that if you have a lot of money, it just
goes to the same old subject — it goes to MIT, to Stanford, to the coasts.

I personally think that it would be a mistake to have all
the money being directed into those kind of places. One: Because they are
expensive. And two: Because in order to make this thing politically
sustainable, you have to actually also get buy-in from many of the places who
are not currently major hubs of technology but potentially could be hubs,
places which have a good educational basis. So there’s a possibility of using
people or educators that have some possibility of using this money well but
also maybe are not as expensive as some of the places and the coasts where the
main funds often go to.

It’s that part that worries me. It seems like we’re
accepting some inefficiency and wasted spending by spreading this money, just
so we can get the political support for spending any more money on R&D.

Yeah. In my proposal I’m not advocating this place-based
approach that Simon and John are doing, because I share some of your worries, to
be honest.

Which seems amazingly bureaucratic, because they have a 93-point
criteria plan to figure out which communities. I’m exaggerating, of course.

They have two basic principles, I think. One is that you
need to have a certain level of education. If there’s not enough people who
have college degrees, then it’s going to be difficult to really get something
going. But they also have the criteria that they don’t want it to be too
expensive, so the housing prices have to be below a certain level. Now, those
are the two main criteria. But I agree, you’ve got to have a balance between
not having it too bureaucratic but having something that is politically

I feel most comfortable with the economics, and I think
you probably do as well. We have these pressing needs; there’s good evidence
that some innovation policies work. The private sector, left to itself, is not
going to produce enough of it. So that creates a good economic argument. Then
we get into the nitty-gritty detail of how we’re actually going to do it. And
do we need to make some compromises in order to make it politically
sustainable? As an economist, I also get nervous, but I think we’ve got to be
realistic about this. If we’re going to spend large amounts of money, then we
have to do it in a way which makes it not just economically sensible, but
politically sensible.

Well, no, I get it. It’s a balance between what is your optimal, ideal policy versus giving some focus on it actually happening and the political implications. I know you’ve done other work on innovation policy toolkits. I have a chart in front of me, a chart I’ve blogged about many times, in which you list 12 policies. I think this is from a paper you did with Nick Bloom.

Oh, yes.You’re talking about the light bulb table.

Yeah, yeah.

I love the light bulb table!

So on this table, mission-oriented policies are rated one
out of three lightbulbs for net benefit, with low quality of evidence and low
conclusiveness of evidence. So it seems like this is not an optimal policy, but
you’re including it because we need this research in order to handle pressing
problems like climate change and public health concerns?

Well, yeah. I mean, the productivity stuff we discussed is
pretty clear, but there are other things, like environmental challenges and
health challenges, which we’d want to deal with anyway.

A researcher works inside a laboratory of Chulalongkorn University during the development of an mRNA type vaccine candidate for the coronavirus disease (COVID-19) in Bangkok, Thailand, May 25, 2020. REUTERS/Athit Perawongmetha

You can think about climate change. There is an issue, and
it’s not going to be solved, as much as a million other economists would love
it to be the case, by just having the right level of carbon tax or the right
level of regulation. We need innovation to deal with that problem. We need
technological change. So if we need that technological change to deal with
those type of problems, then that is something we need, period. So how do we
then get that? Well, there are these other policies which we know work in other
circumstances that we can leverage.

The way I see this is that we decide what we need. And
maybe that’s beating the Chinese at artificial intelligence, maybe it’s not
climate change. That’s a need we have from other reasons. How do we make that
happen? What kind of innovation policies are going to make that happen? The
other innovation policies are the other ones in the light bulb table: the
R&D tax credits, the direct grants, the trying to get more people to study
STEM, the trying to deal with the fact that we have many potential smart people
and kids that are not becoming inventors, who could become inventors if they
got the opportunity. So those types of policies, I think, are ways we could
actually improve the innovation pipeline to deal with the real missions that we
need to tackle.

The one area of government innovation research that gets a
lot of praise is DARPA, The Defense Advanced Research Projects Agency. Are we
trying to create a super-DARPA here, that goes beyond just defense and finds
big missions? Is what we’re creating a super-DARPA, and maybe one mission would
be climate change?

I think that’s a good way to frame it. There are a lot of
lessons we can take from DARPA. DARPA is super successful. If you look at its history
in generating a lot of innovations, I think that many of the lessons from DARPA
are relevant towards thinking of what we’re doing, and they have certainly
inspired a lot of our thinking here.

One of the principles is, for example, that you’ve got to allow
mistakes. You can’t expect everything to work. You’ve got to be tolerant of
some degree of failure in innovation, because there’s a lot of experimentation
going on. But then when you find something which works, you then try to go for
it big time. So it’s all about allowing a lot of experimentation and being
tolerant of many of those things failing. But for the things which do work, you
really put resources behind them.

It’s also allowing there to be some creativity and
autonomy of the people who are doing the research. If you try and direct
everything too strongly from the center, then that’s also something which often
doesn’t work. It’s not going to be identical to what DARPA is, but there are a
lot of lessons that we can draw from DARPA. I like your word, “super-DARPA.”

Are you proposing a form of industrial policy, where the
government plays favorites with certain technologies and companies? I know that
phrase is coming back into vogue, but I don’t like it — it worries me.

Yes, economists hear the word
industrial policy and, let’s say, reach for their revolver or something.We
have an instinctive reaction against it.

There is an element of industrial policy which is, in my
view, a total failure. And we can see that from what’s happened in my own home
country. I was born in Britain. In the 1970s, there was an attempt to create
these national champions like British Leyland, the car company, which resulted
in total failures. You see this in France, and you see it in many other
countries. The reason that those failed was that we tried to erect lots of
reductions of competition and barriers to protect those companies. Again, if you
look at our light bulb table, a good innovation policy is actually to increase

It’s a maximum three light bulb policy.

Yeah, we had three light bulbs for that. The traditional
industrial policy, which is about picking winners and protection, I think, is
discredited. But I think there are some elements of having an idea that you can
have missions where there are these things that we have a pressing need for and
we can try and create a role for the states in trying to meet those needs. It
does have some credibility to it. So if we can identify where the market
failures are — like there’s a market failure for innovation — we can look at
the evidence over what works and what doesn’t work. And we can set up
institutions which can try and direct the money to where it could be useful,
without saying, “We’re all going to be picking the winners. We set the criteria
for what we want firms or actors to do.” That, I think, is part of a strategy
which can be beneficial for innovation and growth in the future.

If you want to call that innovation strategy, then fine.
The idea really is to say, “Well, there are parts of the traditional industrial
strategy which are clearly wrong.” Which is to say, there is a role for the
government in trying to improve the economy, if you can define those things
well and get the institutions right.

Let’s say you had a company, and one company is just doing
the research but thinks a brand new 5G technology looks promising. Then you
have company two, which is not just research — they’ve actually got something
they can show you, and it’s advanced but not quite ready for market. But they
look promising. And the third company is a company that has a new technology…
but it would be great if they could sell it at a super cheap price, and it’d be
great if the government could somehow give tax credits to states, localities, and
companies to buy this technology.

So you have three different companies. Should they all get
money from your innovation fund?

Well, I think the third one has already developed the

Well, it could help subsidize a purchase.

I mean, you could do that, but it’s not like the early
stage research and development where we think there’s a market failure. I mean,
there may be other reasons.

All right. So number three is out. What about the others? I’m trying to differentiate
between basic research and more applied research, I guess. This is my
rudimentary way of trying to get at that.

Well, basic research has the biggest market failure. So
that’s the one where we think there is a strong reason for putting the money
in. But if you had a mission — so if there are particular things like a
military need to develop face recognition or artificial intelligence to deal
with the fact the Chinese are moving ahead of us in drone technology — then
that might be a reason for doing more near-market research stuff, because then
there’s an additional pressing need. But the pure economics is that you want to
do the early market stuff where we think there’s the biggest market failure. So
that would be the criteria.

Right. Because I’m sure a lot of people listening to this
are going to be like, “This sounds like the Solyndra debacle during the
Obama years. We’re giving companies money.” I think that whole scenario
has been overplayed, but I think a lot of people are like, “Oh, sure. For
basic research, fine, but anything beyond that, government should not be

The headquarters of bankrupt Solyndra LLC is shown in Fremont, California September 20, 2011. Via REUTERS/Robert Galbraith

Correct me if I’m wrong, but you’re saying that we should
be funding basic research more, but there are also these other things which you
also have to fund more?

Yeah. The line between basic research and applied research
is not like the Grand Canyon. They kind of blend together, and you have to make
a decision, based on a set of criteria, about which one you fund or not. I
mean, I would say, “Yes, you want to push it towards basic.” But if
there are these other missions, there’s going to be some other criteria, and
that might mean you push a little bit more on the applied science.

But basic research would get more funding under your plan,


Why did the US let this downshift in research funding
happen to begin with, given how important it is?

That’s a good question. I don’t know for sure. I think we
got complacent. The US was so dominant that a certain amount of complacency
probably set in, and we thought growth was going to continue. I think, also,
there was maybe a view that the private sector would step in and do the basic
research which the government started to move out of. But if you look at what’s
happened — and people have looked at private sector research — if anything,
it’s become even more near-market. So it’s become even less basic, and that’s
not surprising, right? Because the incentives are not really there. So I think
that a certain amount of complacency came in, especially with the end of the
Cold War and the defeat of the Soviet Union. America was clearly number one
and, in many respects, still is. I think that now is a good time. Now we can
see the rise of China and that the Chinese are challenging us.

It shouldn’t be that hard of a political sell, I think,
given that we have slow growth and we’re supposed to have slow growth going
forward. Even before the pandemic we had this rising competitor, which seems to
be throwing gobs of money at all these cutting-edge technologies. If you can’t
sell the idea now, then I think it just can’t be sold.

Well, Eric Schmidt made a good point. I mean, we all got
complacent, in a sense, because we kept saying, “China is great at copying
stuff, but they’re not really innovating, they’re not pushing the frontier like
America or the West does.” And then, I don’t know, at some point we’ve woken up
to the fact — I have it in my paper there — that their R&D intensity has
overtaken France and the UK. They’re going gung-ho with many of the advanced
technologies like artificial intelligence.

So I think we got complacent. We thought that many of
these countries are good at copying but not good at innovating — good at fusion
but not pushing the frontier. That’s not true. They’re not as advanced as we
are, on average, but they’ve caught up. And in many places they’re overtaking
us. I mean, I guess with 5G we’re now seeing this as well.

A researcher uses a pipette to develop assay to detect specific gene of corn at a lab in Syngenta Biotech Center in Beijing, China, February 19, 2016. Picture taken February 19, 2016. REUTERS/Kim Kyung-Hoon

So now is the right time. Now is the moment, I think. Just
think of this like a Marshall Plan. Now we’re thinking about what we need to do
after the pandemic to get back to growth. We should think not about the
short-term things.

A Marshall Plan for ourselves, not for other people.

A Marshall Plan for America. Exactly.

And to finish up, do you have any concerns about the
ability of the United States to implement this kind of plan, especially given
our response to the pandemic?

I do. I mean, I think that there are serious political
issues with the rise of polarization and the ability of the state to do things.
But I don’t think we should be too pessimistic. The way I think of it is the
following: It’s often at times of crisis that you can get really big changes,
and those changes can be bad as well as good. After the First World War, the
whole developed world retreated behind trade barriers and there was the rise of

But after the Second World War, the opposite happened. The
crisis actually galvanized people to realize, “Well, we do have serious
problems, a lot of independencies in the world,” and it became a moment to
seize the initiative to do things differently. So, yeah, I do think there are
huge problems, and I don’t underestimate the problems of state capacity, but we
should never bet against America. We did this before, and we can do it again.

It’s going to take a lot of effort. It’s going to take a
lot of popular will to do that, but I think we might be at a moment where
people realize there is a challenge and realize that we can pull together to
make a difference and hold our politicians to account. It’s not going to be
easy, but I do think it’s possible.

My guest today has been John Van Reenen. John, thanks for
coming on the podcast.

Thank you.