Wednesday 3 July 2013

Can governments use procurement to boost innovation? The US story.


Many have argued that governments can improve growth by procuring “well”, that is to say, in such a fashion that will promote innovation.  In Europe there is a lot of romance about some US examples like DARPA or the SBIF or Apollo/Manhattan programs.   But what are the facts?  David Mowery gave a wonderful presentation today on lessons of procurement from the US. Here is my take on it.

1.     1.  US public procurement is separated in the data from R&D. Federal procurement is actually 63% military.  The non-defense part of procurement is mostly energy, federal housing and veterans admin (mostly drugs and hospital equipment).
2.      2. Note that DARPA and SBIR are not, in the data, procuring.  They are under separate accounts, but besides the accounting point, the important point is that there is an R&D and a procurement program i.e. two instruments.  The former might be used for R&D and the latter for buying the cheapest regardless of innovation.  Or the latter might be used to complement the former.   But with two instruments its not necessarily the case that not using procurement for innovation is missing a trick.
3.      3. The Apollo and Manhattan projects were designed deliberately not to be adopted in the private sector.  So that is very different to e.g. using procurement to trying to find a solution to a grand challenge, like health.   
4.      4. Procurement can take many firms
a.      Paying just for the prototype.   That might encourage innovation just up to an early stage.
b.      Requiring, as in the case of semiconductors, more than one supplier when procuring.  That helps competition and information sharing between the two suppliers.
c.       Paying for the whole project but not promising to then buy the final product.
d.      Paying for the whole project and also buying the final product perhaps from one supplier

5.      some practical details.
a.      Most procurement can be thought of as demand pull.  It procures a product for a particular government missions.
b.      Less common are policies to specifically encourage particular new technologies e.g. deliberately buying innovative energy products to push a green agenda.

6.      How might benefits to innovation from procurement come via the demand pull mechanism?
a.      A lead customer who buys the product might helps firms learn, give them scale economies etc.
b.      Procurement contracts might require new suppliers  or might legitimize standards.

7.      Perhaps most interesting are some cases.  
a.      The big success is IT.  In the 1950s the US military procured  a lot of e.g. software and semiconductors.  This helped Texas Instruments for example to develop semiconductors and seems to have crowded in other suppliers.  Note that one provision of this program was that more than one supplier was needed and this made sharing of information and common standards in the industry a common practice. But note that by the late 1960s, the military was buying a very small share of IT products, so exit by the public sector was important.  In addition, the projects were, in fact, highly complementary to private sector needs.  Indeed, they are the victims of their own success: the military now buys software from the private sector since it’s more advanced.   An attempt by the military in the 1980s to have its own software was a failure.  So ultimately the success was to develop something that was very complementary to private sector needs and the private sector just took over the lead.
b.      Perhaps less well known are some failures.  For example, in the 1970s/80s the USAF tired to develop CNC machine tools but they were a failure: the Japanese did much better.  Another example was civilian nuclear power.  This program focused on PWRs since they were going to be used by nuclear subs.  They were a failure, unsafe and became unused etc.

8.      so what are the lessons?
a.      If the project is aligned with civilian needs, then more chance to be widely adopted.  Very hard to say ex ante: indeed totally impossible.   Components will have more chance rather than a whole system perhaps.   Or maybe the link is endogenous, maybe procurement can try to encourage sharing etc.
b.      There’s a natural tension between wanting to get the lowest price and wanting to support innovative products which might be expensive.   But when there are two targets you need two instruments and you have this: an R&D and a procurement budget.
c.       The Apollo and Manhattan projects highlight that the purpose of the project has to be clear: they were never intended for public use, so aren’t good examples of encouraginginnovation.  Indeed, that is case where the procurement project was a different instrument to the R&D budget.
d.      Supporting prototypes might be good since they don’t mean lock into a whole project. But prototypes are impossible in some technologies e.g. nuclear submarines.

Finally, in the discussion, David Mowery made some other very interesting points.  My interpretation of them are:
1.              the romance with SBIR is misplaced.  There are very few rigorous studies of its effectiveness and what studies there are show very mixed results
2.              Much of venture capital in the US does not just supply money.  It also provides managers, help with business plans etc. that’s why some of them are very local, since they are tailor to local labor market needs.   So it may not be something that the EU can imitate by just providing money if the EU cannot also supply the other skills and advice.
3.              Regarding tech transfer, everyone seems to concentrate on the gross return that universities make from ownership of IP.  The net return, aside from some very few top universities, is very modest, because it costs a lot of money to run the office and do the work.  In particular, it costs a lot for patent attorneys (although these costs might be reduced by universities sharing facilities).