National Academies Press: OpenBook

Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop (2022)

Chapter: 2 Multinational Firms and Global Innovation

« Previous: 1 Introduction
Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×

2

Multinational Firms and Global Innovation

Paper Authors: Anna Gumpert (Ludwig-Maximilians-Universität München [LMU Munich]), Kalina Manova (University College London), Cristina Rujan (Max Planck Institute for Innovation and Competition), and Monika Schnitzer (LMU Munich)

Presenter: Anna Gumpert (LMU Munich)
Moderator: Justin Pierce (Federal Reserve Board)

The workshop paper by Anna Gumpert, Kalina Manova, Cristine Rujan, and Monika Schnitzer explores the role multinational enterprises (MNEs) play in the diffusion of production and innovation across firms and countries. Offshoring of production decisions is a well-known phenomenon, and the authors extend their analysis to the innovation decisions of firms.

Anna Gumpert, assistant professor of economics at Ludwig-Maximilians-Universität München (LMU Munich) and research affiliate at the Centre for Economic Policy Research and at the Center for Economic Studies at the ifo Institute, introduced her presentation by observing how Mercedes-Benz and BMW, both German car manufacturers, can illustrate the process of innovation fragmentation and complex global value chains (GVCs). In 2017, Mercedes-Benz opened a research and development (R&D) laboratory in Seattle to focus on basic design innovation; and in 2018, BMW opened an R&D lab in Shanghai to focus on applied innovation, such as design and autonomous-driving technologies. These examples provide anecdotal evidence of MNE outsourcing of innovation activity to both developed and emerging economies. Gumpert argued that the examples of these firms are evidence of a larger global phenomenon. MNEs are the center for both global technological progress and the global fragmentation of production; they are also responsible for the majority of private R&D expenditures globally. These firms have also increased the fragmentation of their value chains across countries.

Gumpert argued that the contribution of the workshop paper lies in better understanding MNE innovation behavior, a field she said is little studied, outside of the papers discussed in this workshop.

Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×

More specifically, she indicated that their paper’s contribution can be summarized as (1) providing novel facts based on a unique dataset of German MNEs, (2) offering an integrated model of MNE production and innovation, and (3) providing reduced-form empirical evidence consistent with the model assumptions and predictions. (Gumpert noted that this third aspect of the paper is still in progress.)

DATA

The paper uses data from Bureau van Dijk’s Orbis database. These data, covering 15,000 German MNEs and their global affiliate networks from 1999 to 2016, are matched with firm-level data on patents over the same time span from the European Patent Office’s PATSTAT database.1 Gumpert observed that Germany is an ideal context for studying these questions as it is an innovation leader and is home to a large number of MNEs.

Gumpert described how these patent data are rich and offer several channels for inquiry.

  • First, the data are characterized by inventor location, which allows the authors to conduct analysis based on domestic or offshore innovation location, and by whether the offshore innovation was colocated with offshore production. (Innovation and production are colocated if both activities occur within the same country.)
  • Second, the data are characterized by type of patent. Patents are split into three categories: basic innovation, applied product innovation, and applied process innovation. An example of basic innovation is a new chemical reaction. Applied innovation is further from basic science. An example of applied product innovation is a new product based on the new chemical reaction, and an example of applied process innovation is a more efficient process for the new product. In the analysis by Gumpert and her coauthors, distance from science is determined by backward citations to scientific journals, and the different types of applied patents are assigned after a textual analysis of the patent abstract.
  • Finally, the data are characterized by patent quality, which is measured by the number of forward citations of the patent 5 years after application.

___________________

1 The authors originally intended to use data from Bundesbank’s Microdatabase Direct Investment (MiDi). Although they were unable to access MiDi directly because of pandemic restrictions, Gumpert explained that they were able to use Bureau van Dijk’s Orbis database instead to obtain the information.

Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×

STYLIZED FACTS

Gumpert presented three novel, stylized facts uncovered from the match between the firm data and patent data.

  • First, MNEs innovate actively, and frequently abroad. Nearly one-third of German MNEs own at least one patent. Nearly one-third of patent-holding MNEs have a patent with a foreign inventor. Therefore, 15 percent of those in the entire MiDi database are foreign inventors.
  • Second, German MNEs offshore innovation to locations with and without an affiliate present. More than half (56 percent) of German MNE innovative activity is done only at home, and very few MNEs innovate solely abroad whether or not they have a foreign affiliate. At the same time, 14 percent of MNEs innovate in all locations (at home, offshore where they have an affiliate, and offshore where they don’t have an affiliate).
  • Third, the type and intensity of innovation varies across firms, but firms that hold more patents tend to hold patents of significantly higher quality.

MODEL

Gumpert presented an integrated model of MNE production and innovation strategy. This model assumes

  • heterogeneous firms with respect to productivity,
  • three countries (home and two host countries),
  • a constant elasticity of substitution (CES) demand curve with no trade costs, and
  • three types of innovation (basic, applied product, applied process).

Countries differ on multiple dimensions. The production wages are assumed to be highest in the home country and lower in the two hosts, with the hosts having different wages. The authors impose no structure on inventor wages. They assume that fixed costs of innovation are equal in the two host countries and that the fixed cost of innovation is lowest in the home country. The three types of innovation are captured in the model by three cost functions. Basic innovation will increase future profits, applied product innovation decreases the fixed cost of adding a new product, and applied process innovation decreases the marginal cost of production. The cost functions are similar in that endogenous innovation costs increase with innovation quality.

Firms choose the location and intensity of innovation to maximize global profits. The paper authors allow for more generality in the full paper, but for the

Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×

purposes of her presentation Gumpert focused on the case in which production location is given, innovation only occurs in one location, and the only innovation type is basic. Innovation can take place in conjunction with production in host 1 or without production in host 2.

Gumpert then presented the full maximization problem for this simplified model. Firms choose the innovation intensity of basic innovation, q, to maximize their profit function, pi. The production side assumptions are standard (CES demand) and contain variable profits and simple fixed costs. This means that basic innovation either increases future profits or decreases the exit probability of the firm. In the paper, this maximization is done on all types of innovation. When firms innovate, they face innovation costs that are the sum of a fixed innovation cost and the variable cost of innovation that is increasing in quality. The variable costs of innovation also depend on colocation of innovation and production. For applied innovation, the variable costs of innovation are lower if innovation is colocated.

Gumpert noted that, while the paper includes seven testable predictions, she focused on just four of them in her presentation:

  • First, more productive MNEs are more likely to innovate—and innovate more intensively—in each innovation type. This result is due to the structure of the profit function being supermodular with respect to productivity and innovation quality.
  • Second, more productive MNEs are more likely to offshore innovation of each type. This result is driven by higher fixed costs of offshore innovation relative to domestic innovation.
  • Third, applied innovation is more likely than basic innovation to be colocated with production. There are synergies between production and applied innovation that do not exist for basic innovation.
  • Fourth, MNEs are more likely to offshore basic innovation to countries with a comparative advantage in innovation, and applied innovation to countries with a comparative advantage in production. This is driven by the same synergies that drive prediction three.

Gumpert also briefly mentioned the remaining predictions. Fifth, the elasticity of the extensive and intensive margins of innovation with respect to firm productivity may vary across innovation types. Sixth, more productive MNEs are more likely to offshore innovation in house. Lastly, more productive MNEs are more likely to offshore basic innovation at arm’s length to countries with a comparative advantage in innovation, and applied innovation in house to countries with a comparative advantage in production.

EMPIRICAL EVIDENCE

Although she noted that the empirical section of the paper is still in progress, Gumpert presented some of the early results.

Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
  • The authors found significant and positive results suggesting that larger firms are more likely than smaller firms to patent at all, to patent more, and to produce higher-quality innovation. This supports the first prediction.
  • They also found significant and positive results suggesting that larger, or more productive, firms have more foreign innovation by all types of patents. This result is strongest for applied process patents. The results also suggest that these firms have a larger share of foreign innovation relative to smaller firms across all innovation types. This supports their second prediction.

Gumpert and colleagues also tested a corollary of the model, that offshore innovation is of higher quality, and found that in all specifications, the foreign-only patents are of higher quality than the coinvented patents. These results are all weakly significant. They also found that applied innovation is more likely than basic innovation to be colocated with production, the excluded categorical variable in the results table. This supports their third prediction.

Gumpert emphasized that these results are all conditional correlations and are not to be interpreted as implying causal relationships. In order to test for causality, they studied the impact of distance to high-quality universities. They found significant and negative results suggesting that firms that innovate more are physically closer to all universities. However, after controlling for firm size, only the distance to the top university is significant. Gumpert added that she and her colleagues plan to explore this last relationship, as well as the relationship of university quality, in future work.

SUMMARY

Gumpert closed by pointing out key takeaways. First, MNEs innovate intensively and frequently abroad, offshoring innovation to both production and nonproduction locations. Second, more productive MNEs innovate more intensively than less productive MNEs, and they are more likely to offshore innovation. Lastly, offshore applied innovation is more likely than offshore basic innovation to be colocated with production. Gumpert noted that avenues for future work include identifying the optimal innovation and tax policy in developed countries and the optimal foreign direct investment policy for developing countries seeking to attract capital and technology.

DISCUSSION

Discussant: Allison Derrick
(Bureau of Economic Analysis, U.S. Department of Commerce)

Allison Derrick, of the Bureau of Economic Analysis (BEA) in the U.S. Department of Commerce, began her discussion of the paper by briefly

Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×

summarizing the work. The authors’ main objective, she said, is to understand how MNEs produce and innovate. Three types of innovation—basic, applied product, and applied process—are presented and incorporated into a formal model that has several predictions. The model suggests that the different types of innovation have different incentives, and preliminary empirical work using firm-level patent data shows that some predictions of the model appear to be supported.

Derrick indicated that the main contribution of this work is using MNE data to shed light on an open empirical question: how do MNEs innovate? She noted that the dataset used, Orbis, relies on public data and will not be as in-depth or reliable as the confidential MiDi database proposed, and she suggested that the results may or may not hold in the preferred dataset. The paper contributes to the literature by understanding where different stages of innovation take place. This has important implications for developing countries and trade.

Derrick noted that, although it is young, the literature on MNEs and innovation is established and provides several relevant papers. Data from the BEA, which are similar to MiDi data, have been used to study similar questions. The BEA data have also been linked to patent data in a similar fashion to that of Gumpert and her coauthors. For example, Bilir (2014) studied how intellectual property (IP) rights influence MNE production decisions; Berry (2014) found that multicountry patents have grown over time; and Branstetter and colleagues (2019) documented the globalization of R&D and the rise of new R&D hubs, and observed that the increase in foreign R&D among U.S. MNEs is driven by software and information technology industries.

Derrick expressed major concern about the external validity of the results; specifically, whether patenting activity varies across industry and location or the results are driven by the data in the sample. Patenting variation is driven by variable R&D costs and imitation costs, life-cycle length of products (incentive to not apply for a patent until a clear use is identified), and IP protection in the producing/innovating location. Derrick suggested that Gumpert and colleagues should control for patent differences by industry and location. This has the potential to reveal important patterns in the model while also checking to see if one or two industries are driving the results. Without mentioning specifics, she recommended robustness checks for these results.

Derrick suggested that there are several key insights from a paper she is currently working on with her colleague Christopher Steiner. In that work, they looked only at innovation with firm boundaries and have not studied the different stages of innovation. Their work has been motivated by the increases in R&D service imports and exports over the last two decades. Derrick indicated that she is surprised by the large number of German MNEs that collaborate with unaffiliated parties, which is quite different from patterns among U.S. MNEs.

Derrick also mentioned other work relevant to the subject paper. Previous research has shown that the measurement and observation of innovation are difficult (see Corrado et al., 2017; Lev and Gu, 2016). There are three proposed measures of innovation: R&D expenditure and employment, IP-related services, and patents. Derrick and Steiner (in progress) use the first two categories,

Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×

but not patents. They found that 85 percent of R&D is conducted at home and about 60 percent of R&D-intensive MNEs are in manufacturing. Derrick stated that there are two kinds of innovators. First are adapters, those that conduct applied product and process innovation to fulfill production roles. Adapters are common but small in size, and most belong to a manufacturing MNE. Second are R&D centers and IP sellers, those that perform basic or applied innovation and for which R&D may be their primary industry. Relative to adapters, both R&D centers and IP sellers report more sales, have higher R&D expenditures, trade more IP services, are more common among nonmanufacturing MNEs, and are more productive. The major takeaway of this, she said, is that Gumpert and colleagues should look at how industries drive their results because firm innovation behavior is heterogeneous along industry and type.

Another concern Derrick raised is the measure of productivity. The subject paper uses log global sales to measure productivity. A concern with this as a measure is that it directly measures the size of the firm, which may be correlated with firm productivity. However, she said that this concern may be alleviated by access to the MiDi dataset, which has better measures of firm productivity, such as value-added productivity.

Derrick also suggested that patent ownership is likely affected by tax rate differences. As shown in Derrick and Steiner (in progress), the majority of R&D takes place in nonhaven countries, while ownership and royalty payments are shifted to tax-haven countries. In relation to the subject paper, there is the question of how many of the foreign inventors live in tax havens.

Finally, Derrick offered some miscellaneous comments and questions. First, how is “arm’s length” determined? Does this mean there is no affiliate with a German parent or no affiliate within the entire MNE? This is important as the German parent may not be the global owner. Second, for innovation at arm’s length, with whom are the MNEs innovating? This may help shed light on the first causal relationship between innovation and distance to a top university. Lastly, some descriptive statistics on patents should be included.

Justin Pierce, principal economist at the Federal Reserve Board and moderator of the session, commented that the authors should decompose innovation quality in a similar fashion to innovation location.

Anna Gumpert thanked Allison Derrick for a great discussion and the many helpful comments and suggestions. While several suggestions were given, she offered two main responses.

First, in response to the questions Derrick posed about measures of innovation, Gumpert clarified that in their paper, she and her colleagues used only patents owned by German parent firms. This should alleviate some of the concerns raised about tax havens and external validity. R&D expenditure is not a useful measure in this case as it does not allow for analysis of foreign inventors; rather it picks up on the flow of money.

Second, addressing Derrick’s suggestion that the results may not be robust to the MiDi data, Gumpert indicated that this is not a concern because the

Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×

correlations are also present in the MiDi data, although the exact coefficients are different.

During the presentations, one audience member asked about the implications of the use of trade secrets in applied product or process innovation, specifically innovation that isn’t picked up in patents, in firms with foreign affiliates. In response, the authors offered that, while it is impossible to parse this out as they are trade secrets, this would be a larger concern with process innovation because firms are less likely to patent process innovation than product innovation. Another audience member asked where the foreign innovation is taking place and if it is driven by countries like India or China; the authors responded that their work shows that most foreign innovation takes place in countries neighboring Germany.

Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
Page 7
Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
Page 8
Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
Page 9
Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
Page 10
Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
Page 11
Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
Page 12
Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
Page 13
Suggested Citation:"2 Multinational Firms and Global Innovation." National Academies of Sciences, Engineering, and Medicine. 2022. Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/26477.
×
Page 14
Next: 3 Tracing Value Added in the Presence of Multinational Firms with an Application to High-Tech Sectors »
Innovation, Global Value Chains, and Globalization Measurement: Proceedings of a Workshop Get This Book
×
Buy Paperback | $26.00 Buy Ebook | $20.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

In recent decades, production processes of intermediate and final products have been increasingly fragmented across countries in what are called global value chains (GVCs). GVCs may involve companies in one country outsourcing stages of production to unrelated entities in other countries, multinational enterprises (MNEs) offshoring stages of production to units of the MNE overseas, or both. GVCs can also involve completely independent companies merely sourcing their parts from whichever upstream company may be the most competitive, with no control arrangement necessarily involved. The changing global trade environment and the changes in firms' behavior have raised new and more complicated issues for policy makers and have made it difficult for them to understand the extent and operations of GVCs and their spillover effects on national and local economies.

To improve the understanding, measurement, and valuation of GVCs, the Innovation Policy Forum at the National Academies of Sciences, Engineering, and Medicine convened a workshop, "Innovation, Global Value Chains, and Globalization Measurement" May 5-7, 2021. This proceedings has been prepared by the workshop rapporteurs as a factual summary of what occurred at the workshop.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!