18 June 2012

R&D is not Innovation

During the 1950s and 1960s, advocates for government investments in science and technology (mainly basic research at universities) pulled off a remarkable coup. They successfully integrated conceptions of "basic research" with a linear model of innovation, making R&D a key variable in expectations for what led to national economic competitiveness.

The most recent science policy statement of the US National Academy of Sciences, ominiously titled, Rising Above the Gathering Storm (RAGS), led off with this warning (here in PDF), which illustrates such expectations:
The prosperity the United States enjoys today is due in no small part to investments the nation has made in research and development at universities, corporations, and national laboratories over the last 50 years. Recently, however, corporate, government, and national scientific and technical leaders have expressed concern that pressures on the science and technology enterprise could seriously erode this past success and jeopardize future US prosperity.
But does R&D spending correlate with economic success? Not necessarily says John Bussey in the WSJ:
Asia is spending so much on R&D that this year it will pull ahead of total spending in the Americas for the first time.

Advantage Asia?

Maybe not. In the world of R&D spending, more doesn't necessarily mean better. And R&D may not describe all the innovation that matters.

"I think the numbers are pretty useless," says Michael Schrage, a research fellow at MIT's Sloan School who has studied the subject. "What matters more is the kind of innovator you are. If it were really true that the people who spent the most on R&D were the most successful, we wouldn't be subsidizing General Motors.""There's no statistically significant relationship between how much a company spends on R&D and how they perform over time," adds Barry Jaruzelski of Booz & Co.
Why do we believe that R&D is the wellspring of economic growth? Because we have been seduced by an incorrect reading of economic theory that has distorted Schumpeterian economics and the so-called Solow residual.

Here is how the NAS RAGS report puts it:
Early in the 20th century, Joseph Schumpeter argued that innovation was the most important feature of the capitalist economy. Starting in the 1950s, Robert Solow and others developed methods of accounting for the sources of growth, leading to the observation that technologic change is responsible for over half the observed growth in labor productivity and national income.
Sure, there are caveats to this discussion presented in the RAGS report ... on p. 458. On p. 1 the report says this without such caveats:
Economic studies conducted even before the information-technology revolution have shown that as much as 85% of measured growth in US income per capita was due to technological change.
As I have noted on this blog, "technological change" is an (unfortunate) bit of economic jargon that refers to a change in the so-called production function. It does not refer to science-based technology. But in discussions of innovation policy such as that found in RAGS, "technological change" means only science-based technology, typically supported by government R&D investments.

Benoit Godin, the insightful scholar of the history of innovation, explains the consequences as follows (here in PDF):
The problem is that the academic lobby has successfully claimed a monopoly on the creation of new knowledge, and that policy-makers have been persuaded to confuse the necessary with the sufficient condition that investment in basic research would by itself necessarily lead to successful applications.
Hence, we get arguments for the importance of doubling government S&T funding in this or that area, predicated on a boost to economic competitiveness and keeping up with the Chinese/Indians/Japanese/Europeans/etc. Outside of academia, a broader set of lessons appears to be well-understood:
Booz & Co. in 2007 listed the biggest global corporate spenders of R&D. The top 10 were Toyota, Pfizer, Ford, Johnson & Johnson, DaimlerChrysler, General Motors, Microsoft, GlaxoSmithKline, Siemens and IBM.

Then it drew up a second list, a group of companies it called "high-leverage innovators" that returned the best financial performance for every dollar spent on R&D. Booz screened for companies that, over the five previous years, outperformed industry peers across seven measures—including profit, sales growth, and shareholder return—while also spending less on R&D as a percentage of sales than the median in their industries.

No company from the first list made the second list. (Winners included Adidas, Apple, Exxon, Google, Kobe Steel, Samsung and Tenneco.)

That disconnect essentially hasn't changed, says Mr. Jaruzelski. Winning at innovation "is all about talent, process, execution and strategy," he says. "That's given the U.S. a pretty strong advantage over time."

"Technology," he adds, "is not equal to innovation."
A 2011 Booz and Co. report concluded:
There is no statistically significant relationship between financial performance and innovation spending, in terms of either total R&D dollars or R&D as a percentage of revenues.

Spending more on R&D won’t drive results. The most crucial factors are strategic alignment and a culture that supports innovation.
Obstacles to effective discussions of policies that foster innovation stem from the fact that much of government innovation policy is designed by academics and the S&T lobby. Within that community there is confusion about the role of "technological change" in the economy (and even what that phrase means), and more than a little conflation of self interests with what it means to grow societal wealth.


  1. See also http://www.slate.com/articles/technology/future_tense/2012/06/scientist_shortage_is_a_myth_.html

    As for explanations of the govt role, see http://american.com/archive/2012/june/its-not-a-welfare-state-its-a-special-interest-state

  2. Interesting piece.

    Large bodies of researchers, which is a common approach in Australia with the CSIRO, Germany with the Max Planck Institute, Fraunhofer and so on are perhaps of dubious value. Who knows, by employing significant numbers of smart, able people they may actually be a drain on innovation.

    The myriad of attempts to estable 'Silicon' locations is a litany of failure to create innovation. The 'startup' culture of Silicon Valley is not that much about technology today.

    Facebook, for example, could have been coded by many, many proficient coders the world over. Yet it, like Yahoo and many others, continues to appear in the US. No government would have funded research into something as low tech as Facebook.

    But perhaps innovation is also well developed in supermarkets, franchise restaurants and new ways of doing things. Perhaps Kinkos, Hooters and Walmart are where the US shows really remarkable innovative advantage over the rest of the world.

  3. They have constructed a probabilistic model where innovation corresponds to talent through a diversity (e.g. individuals) function. They assert that increasing diversity will lead to increased innovation. This is an intuitive expectation, but it is has not been proven to be true. If it was, then it would be possible to preemptively classify human potential. Unfortunately, talent does not guarantee conversion of knowledge or skill to a useful product. Worse yet, American universities are increasingly producing technicians rather than engineers or scientists. Whereas the latter can act in the capacity of the former, the reverse is not necessarily possible. Then there is the problem of mutual indoctrination, which further limits the potential of each person as an individual and within a group.

    The value of market economics (i.e. comprehensive, diverse, and adaptive) remains unappreciated.

  4. Then there is the problem of mutual indoctrination, which further limits the potential of each person as an individual and within a group.

    W.H. Whyte had this criticism of the performance you could expect from a scientist in large corporate R&D departments in his book 'The Organization Man'; see Part V: The Organization Scientist.

  5. sien: "But perhaps innovation is also well developed in supermarkets, franchise restaurants and new ways of doing things."

    Well said.

    Xavier Sala-i-Martin addresses 6 common myths in the field of economic growth, in this great TED talk:


    Myth 5 starts at 6:55:
    "The ultimate source of growth is TECHNOLOGICAL PROGRESS. Hence, need R&D/High Tech Sectors/Rich Countries"
    (the myth is in the second sentence.)

    Xavier mentions among others Nespresso and Starbucks as innovations.

    "Only 8% of the ideas com from scientists; 92% is from workers, students, ..., street performers! All sectors and countries."

    #FF @XSalaimartin

    Xavier Sala-i-Martin is the Jerome and Matthew Grossman Professor of Development Economics, Columbia University in New York and a visiting professor at Universitat Pompeu Fabra in Barcelona. He is also the Chief Economic Advisor for the World Economic Forum at Davos.

  6. If you want to understand how Silicon Valley works, I suggest attending pitchfest events there.

    From my personal observation and participation in the process (as an entrepreneur), it is very much the network (I deliberately do not use the word social) effects which I believe is the distinguishing factor. The pitchfests combine entrepreneurs of all manner of categories with angel/VC panelists, but the real value is in allowing the entrepreneurs both to directly see each other in public and the frank, though not mean spirited, feedback between experienced entrepreneurs/investors and startups.

    Between the number of nodes needed for sufficient network effect, plus cultural issues (failure much more forgivable, minds much more open about any possibility, nepotistic/corruption less a factor), this is what makes Silicon Valley stand out.

  7. I wonder if a bioligical evolution model might be more relevant when it comes to innovation where cross polination and competition provide the building blocks and hone the skills needed for success. A linear development model may work for extrapolation but the game changing stuff probably involves people with dissimilar but complimentary backgrounds doing work under the radar that no one approved.

  8. I completely agree with the title 'R&D is not Innovation'. However, you shouldn't confuse government expenditures on R&D (especially regarding basic research) with private expenditures on R&D and associated profits of these companies.

    Public expenditures (and to a lesser extent also private expenditures) have enormous positive externalities.

    Maybe the companies not investing in R&D themselves, use the outcomes of public research in a smarter way (directly by using outcomes of public R&D or indirectly by investing in their university-schooled 'human capital').

    So, the observation that more private expenditures on R&D does not necessarily lead to more innovation, does not imply governments should invest less in public R&D.

    My assertion would be completely in the opposite direction: Less private R&D in the future will lead to higher positive externalities for public R&D and should be accompanied by even higher investments in public R&D!

  9. I wonder how much of the survey participant R & D was primarily D. As a software developer, my take is that an enterprise with a large investment in existing technology must spend a lot on D just to stay current.

  10. Seems that this analysis fails to include the obvious and simple fact that many companies effectively buy R & D by acquiring other companies (that have invested in R & D).

    It also fails to deal with another aspect - only so much growth can be sustained through acquisitions. Companies in the position to acquire other companies - even if the R & D investment by the acquired companies is accounted for - can cherry-pick in ways that seem to maximize the return on R & D expenditure and make it seem that they've "innovated" without R & D expenditures. However, not all companies are in a position to run around acquiring other companies that have managed to maximize their return on R & D very efficiently. It depends on the market sector and the type of business.

    Further - some companies get better return on R & D than others purely by luck, or because they happen to make better decisions (at least for a particular period of time). By aggregating this data you really render it meaningless. Essentially what you are saying is that investment R & D isn't necessarily well-spent, but it could be. Is that somehow supposed to be meaningful or useful? Would that, therefore, mean that companies or governments should stop spending on R & D? What is the point here?

  11. I'm really hoping that I'm missing something, because here's what a summary of this post looks like to me.

    Not all money spent on R&D is well spent. Money that is well spent on R&D brings a better return on investment. Money that isn't well spent on R&D brings less of a return on investment. Companies that spend more on R&D do not necessarily make better R&D investments.

    Here's a question - which companies received a net negative return on their R&D investment?

    Assuming that none did, while it is clear that some companies got a better return than others, all companies should still continue to spend on R&D.

    Assuming that some did, those companies should restructure their R&D investment.

    I know that I am usually quite slow to catch on, but I still don't see what the point of this post was other than to state the obvious.

  12. Innovation follows from the expectation of reward. Remove this incentive and innovation dies.

  13. The difference between the maximum spenders on R&D and the most effective suggests that there is a law of diminishing returns on R&D. This is hardly surprising because there4 is a law of diminishing returns on everything. However if the big spenders believe they are getting their money's worth then they probably are and the fact that smaller companies are getting better than that is not an argument against high R&D spending.

    However when it comes to government encouragement of R&D probably the worst method is what we use now - grants to the established (or often just grants to the politically approved - see climate "science"). The best way is by establishing X-Prizes for specified achievements and letting anybody go for them. When they don't work no prizes are awarded so they are certainly cost effective. When they do they are even moreso. For example the Darpa Road Challenge gave prizes totalling $3 million and achieved results that they said "the Department of Defense has created new technology that would have otherwise cost more than $100 million, and taken a lot longer to perfect."


    If society wants to encourage innovation and thus progress we should heavily fund X-prizes. If politicians just want to distribute patronage they should give grants - which is what they do.

  14. Hey Roger,

    I know I'm late in reading this and joining into the conversation, but I wanted to call into question the source you use that states:

    "A 2011 Booz and Co. report concluded:
    There is no statistically significant relationship between financial performance and innovation spending, in terms of either total R&D dollars or R&D as a percentage of revenues.

    Spending more on R&D won’t drive results. The most crucial factors are strategic alignment and a culture that supports innovation."

    -This ignores all of the work done by B. Hall and all the others working in the Economics of Innovation sphere. ROI on R&D on the LOW end is around 15% when even a basic job of looking at spillovers is included in the analysis; but most studies find social returns on the order of 30%. It's the economies of scale that limit most firms from taking part in the process. Even at the firm level, it is absolutely clear that highly innovative firms (those that invest in R&D) are more profitable, and lots of time/effort have been spent in pinning down the causal effects and sizes (stratifying between industries in many cases.)

    Now: admittedly there are problems with measurement (which I am in the midst of investigating in detail), but given the best data we currently have and the best econometrics we've got, I think that it is clear that there is a positive, causal relationship between basic science (where most of the spillovers occur), applied, development, and new (innovative)salable products. This is in all 4 spheres of innovation: public, private, university, and their intersections.

    The point that I won't contend is the confusion between "economic innovation" and "technological innovation". Economic innovation is based on a production function where some process is employed (either new inputs, or OJT) that make it possible to produce more with less. This is definitely different than what the media has picked up and attributed to Schumpeter as you start off with.

    -Just a couple thoughts to consider.

  15. -14-Unknown

    Thanks ... The Booz study was looking at aggregate R&D spending and relative financial performance. That appears different than the (absolute rather than relative?) metrics you cite. And of course spillovers and social returns are a different variable.

    That said, if you'd like to go from unknown to known, send me an email or just submit something in the comments, and I'd welcome a guest post or specific literature worth reading and summarizing.


  16. I'm not sure why I'm "unknown"... but the recent complete review is actually available in the 2010 handbooks:


    I bought and refer to this pair of books often. It is well rounded, covering most issues surrounding the economics of innovation.


    Justin Hicks