May 24, 2013

Open innovation meets strategy in Atlanta

At the Atlanta Competitive Advantage Conference Thursday, the final conference session was a panel entitled “Open Innovation: Multidisciplinary Perspectives and Opportunities for Future Research.”

The economics discipline was represented on the panel by Ashish Arora of Duke, a Carnegie-Mellon grad. (I’m told Ashish is not a regular at ACAC, but CMU was well represented). Org theory was represented by Todd Zenger of Wash. U St. Louis. I guess I represented innovation studies (if that’s a discipline). We spoke in alphabetical order and then engaged in a discussion with the audience, moderated by Henry Sauermann of Georgia Tech.

1. Overview of Open Innovation

I presented an overview of open innovation entitled “Open Innovation: The First Decade.” (My slides are up on SlideShare). Although I started from a previous talk, here I made few assumptions about people’s familiarity with OI. As usual, I started with the official Chesbrough (2006) definition of open innovation, mentioned my own shorthand definition en passim, showed the familiar Chesbrough “funnel” (also from Chesbrough, 2006), and enumerate the citation counts for the various Chesbrough books, including 6,161 for the original book, and 1,230 for our 2006 book.

Then, given the audience and panel, I briefly summarized how OI built upon and extended other innovation studies (adapting the former from an unpublished paper and the latter from Table 1.1 of Chesbrough 2006). Clearly I missed an opportunity to tell a new audience that almost anything they needed to know about open innovation can be gained by reading Chesbrough (2006).

Again, to set the stage, I summarized a comparison of OI and UI from chapter I'm doing with Frank Piller from the next Chesbrough-Vanhaverbeke-West open innovation book, Open Innovation: New Research and Developments. To complete the introduction, I should have shown that not all open source software is open innovation (Figure 5.1 from the 2006 book), to address the questions and comments I heard in the hallway during the conference.

The rest of the talk focused on the three OI modes ("core processes") defined by Gassmann & Enkel (2004) — inbound, outbound and coupled. Most of the remaining talk summarized the forthcoming West & Bogers review of inbound (and coupled) open innovation. For the presentation, I converted Table 3 of our paper (summarizing our corpus of 165 papers) into a Venn Diagram, which was much easier even for me to make sense of. The rest of the inbound discussion briefly summarized that paper.

For outbound, it was much harder. The majority of the 12 retracted Lichtenthaler papers are about outbound IP licensing or other similar processes, leaving a major void. (Perhaps some doctoral students will fill this void). As for coupled, there’s also a void, but one I’m hoping to help fill soon, both through the chapter with Frank Piller and another chapter under preparation for journal submission.

Henry Sauermann, Todd Zenger, Joel West and Ashish Arora.
Photo by Phillip C. Anderson (U. Illinois)
2. Historical Context

In Ashish Arora’s presentation on his work related to open innovation, one of the major themes was the historical context. We found ourselves agreeing (and presumably with Dick Langlois as well) that the vertically integrated Chandlerian model may have just been a temporary blip of the 20th century.

Similarly, Todd Zenger laid out three phases of organizational research. In the 20th century, the M-form was prevalent for the first 75 years, while an interest in networks (such as the Japanese production networks) arose in the last 20 years. In the 21st century, scholars have developed an interest in open innovation, user innovation, contests, communities and other related phenomena.

To estimate the magnitude of licensing revenues in the US, Ashish pointed to the research of Carol Robbins, of the U.S. Bureau of Economic Analysis. I believe he was referring to Robbins (2009), and plan to study that further.

In response to a question, Ashish suggested that the motivation of (inbound) OI was the failure of the internal R&D business model — starting in the 1980s — when firms discovered that their industrial R&D labs were not producing the desired returns. In agreement, I noted the recent tendency of firms to use open innovation as an excuse (or fig leaf) for cutting their internal research labs.

Todd and I also both argued that there’s an organizational innovation that was adopted by early firms and copied by other firms. For example, IP-based business models 20 years ago were exemplified by Dolby, but now we have Arm, Gore and Qualcomm.

3. When Does OI Work?

Both Ashish and Todd have been involved in our special issue of Research Policy on open innovation. Ashish is the lead editor monitoring the process for RP — and thus has seen all the submissions — while Todd was an active participant at the special issue conference that we held last summer at Imperial College London.

Both voiced similar criticisms of the open innovation research to date. Both said we don’t need more research that says open innovation is “cool”, “good” or “common” research. Ashish is also getting tired of open source research.

Some areas were more contentious. Ashish thinks the substitution of internal and external research is a settled issue, but not everyone in the audience agrees. (As Marcel & I noted in our review, we think there’s at least one more U-shape OI model to be tested — use of inbound OI can reduce the need for internal R&D capabilities but only up to the point of losing absorptive capacity).

Both Ashish and Todd wanted to know both under what conditions do firms adopt OI, and under what conditions is it a better choice. Testing the performance of OI strategies doesn’t address the endogeneity of which firms chose OI: under some conditions, OI is probably suboptimal but we won’t see that if we only test OI under conditions where firms thought it would be optimal. As an example, Ashish mentioned Mueller 1962, which showed that that the profitability of Dupont’s internally developed and externally sourced technologies was identical, a result that puzzled Ken Arrow.

We had a very interesting comment from one participant (who I was later told was Gordon Walker) who described his work with Xerox more than a decade ago. The company had a very realistic, nuanced and theoretically sound strategy for sourcing software, but it still failed. Both Todd and I wondered whether that was just because it was Xerox (who had trouble doing anything right during this period), and I also noted that most hardware companies during this period (cf. Motorola, Nokia) failed to recognize that software was a core capability rather than something that should be outsourced.

In my talk, I quoted Kathy Eisenhardt’s luncheon speech Wednesday about the need for higher level abstractions, urging scholars to do more to abstract what we know about open innovation to broader theories. I noted I am perhaps more guilty than of staying at the level of the phenomenon.

As it turns out, such abstraction was the focus of Todd Zenger’s talk, which followed mine.

4. Problem Perspective

Not surprisingly, the major emphasis of Todd’s talk built upon his efforts over the past 10 years to develop a problem-based view of the firm (e.g., Nickerson and Zenger, 2004).

He argued that when firms making a sourcing decision, the level of the firm is too general, the level of the innovation is unknowable a priori (until the innovation is selected), and thus the proper unit of analysis is the problem. Todd then presented an updated version of last summer's paper on applying problem-based view to the firm choice of open innovation strategies.

In the latest paper, Teppo Fellin and Zenger (2012) use a model of communication cost and problem complexity predict that the external sources will be used when a large amount of the knowledge that a firm needs is both outside the firm and not easily identified. In such cases, the “hidden” knowledge is self-revealed by those who have it.

Todd presented Table 3 from the paper, and below is my interpretation of that table with the inbound OI strategies shown in bold:

Problem Complexity
High Moderate Low
Hidden Knowledge High (User) Community Partnership or Alliance Contests or Platforms
Low Consensus-based hierarchy Authority-based hierarchy Contract via market

In the follow up, Joe Mahoney noted the link of the problem-solving view to Herb Simon-style decomposability. At the same time, he argued that the Zenger formulation should (as advised by Milgrom & Roberts) treat separately the coordination and incentive problems. (Zenger expressed doubts that the two are distinct).

At the same time, Ashish expressed skepticism firms could switch their sourcing strategies for each problem, due to the inertia [presumably associated with their structure and capabilities]. Instead, he suggested: “I would argue that they would pick the problems to match their capabilities.”

5. New Research

A doctoral student asked whether OI could be studied in the context of disruptive innovation. I believe it’s an open empirical question whether OI is more or less common for disruptive, radical or incremental innovations — something worth studying. However, my suspicion is that over the past 20 years (perhaps since the R&D cutbacks identified by Ashish) the more radical innovations have been coming from outside. For example, Google bought Android from outside — and, as Ashish noted, also YouTube and Google Glass.

For me, the most novel* OI insight was one slide in Ashish’s talk, about the research of his doctoral student, Luis Rios. The slide summarized Rios’ analysis of the sourcing and IP strategies for a highly visible radical innovation, Google Glass. In response to an email, Rios summarized the story as follow.
Google Glass incorporates quite possibly the entire spectrum of inbound technology, including acquired IP, in-licensed IP, outsourced IP, as well as internally developed technologies. But at the hub of this revolutionary and complex technology seems to be a figure that we will likely hear more about in the coming years: Neven Hartmut, who came to Google (along with his patents) as part of the deal when he sold his eponymous Neven Vision, Inc in 2006.
I look forward to reading Rios’ work in progress, “On the genealogy of knowledge,” when it becomes available for wider distribution. It could provide important evidence about make and buy (rather than make vs. buy) strategies being used to create complex assembled products.

* I should note that I wrote my own speech, heard parts of Ashish’s talk at UC Berkeley in March, and an earlier version of Zenger’s talk last July


The 75-minute format, the audience (about 45 people) and the panel made for a lively discussion. I heard from a number of people that they better understood the issues of open innovation from my intro and the discussion. I certainly gained additional insights, both from fellow panelists and the sorts of questions and concerns of the audience.


Felin, Teppo and Zenger, Todd R. 2012. Open and Closed Innovation, Problem Solving and Governance Choice (October 4). Available at SSRN:

Mueller, Willard F. 1962. "The Origins of the Basic Inventions Underlying Du Pont," in the Rate and Direction of Inventive Activity: Economic and Social Factors, p. 323 - 358, URL:

Nickerson, Jack A. and Zenger, Todd R. 2004. “A Knowledge-based Theory of the Firm - A Problem-solving Perspective” Organization Science, 15 (6): 617-632, DOI: 10.1287/orsc.1040.0093

Robbins, Carol A. 2009. in “Measuring Payments for the Supply and Use of Intellectual Property,” in Marshall Reinsdorf and Matthew J. Slaughter, editors, International Trade in Services and Intangibles in the Era of Globalization, University of Chicago Press, URL:

No comments: