Multinational companies are at the heart of two key globalisation trends: the fragmentation of production and the internationalisation of innovation. This column studies the global innovation activity of German multinationals, and finds that larger multinationals file more patents, of higher quality, at home and in more foreign countries both with and without affiliates. Multinationals follow countries’ comparative advantages in innovation, and more frequently pursue basic (science-based) innovation than applied (non-science-based) innovation abroad, but co-locate it with production less frequently. The findings inform the scope for multilateral deep agreements that span tariff reductions, investment promotion and intellectual property rights protection.
Multinational companies (MNCs) are at the heart of two key globalisation trends: the fragmentation of production and the internationalisation of innovation. MNCs manage complex production networks across both affiliated and independent parties in multiple countries. They also conduct the vast majority of frontier R&D and enable cross-border technology transfer. This raises first-order policy questions in the current climate of trade disintegration on the one hand (e.g. the China-US trade war, Brexit), and regional deep integration on the other (e.g. the Regional Comprehensive Economic Partnership, or RCEP). For developed economies, how does offshoring production affect MNCs’ innovation prowess, and does R&D abroad come at the expense of R&D at home? For developing countries, what attracts MNC production and innovation to fuel capital and knowledge exchange? From a global perspective, how does the world R&D map shape global growth? And what are the implications for trade, FDI, and innovation policy in rich nations, in emerging markets, and in multilateral deep agreements?
While much is known about the global organisation of MNC production (Antràs and Chor 2022), the global organisation of MNC innovation is not well understood. Traditional priors suggest that MNCs headquartered in rich countries offshore lower-skill manufacturing to poorer countries, retain skill-intensive R&D at home, and deploy home-grown technology advances across their affiliates (Bilir and Morales 2020). Yet, anecdotal evidence points to increasing MNC innovation abroad, in both advanced and emerging economies. For example, while leading German car producer BMW has for a long time sourced auto components from China, in 2018 it also unveiled a large R&D centre in Shanghai to specialise in digital services, autonomous driving, and automotive design. In 2017, Mercedez-Benz opened its sixth R&D lab in Seattle, primed as a digital hub for cloud computing.
In recent work (Gumpert et al. 2025), we study the global innovation activity of German multinationals to inform these questions. Germany provides an ideal economic context, being the third biggest exporter, a top MNC origin, and a world innovation leader. We establish new facts based on detailed data on MNCs’ network of affiliates and patent development. We distinguish between basic (science-based) and applied (not science-based) patents, and introduce a new patent measure of countries’ revealed comparative advantage (RCA) in innovation across technology areas. The facts motivate a model of heterogeneous firms that simultaneously choose the locations and scale of manufacturing, applied innovation that lowers marginal costs, and basic innovation that improves future profitability. Consistent with the model, we find that bigger MNCs file more patents, of higher quality, at home and in more foreign countries both with and without affiliates. MNCs follow countries’ RCA in innovation, and more frequently pursue basic innovation than applied innovation abroad, but co-locate it with production less frequently. MNC innovation complementarity across countries lends support to multilateral deep agreements that span trade, investment, and innovation policy.
Key innovation takeaways from MNC patent activity
We exploit rich firm data from Microdatabase Direct investment (MiDi) of Deutsche Bundesbank, matched to patent data from PATSTAT Global of European Patent Office. Compared to typically incomplete R&D data that reflect the cost of in-house research activities, the comprehensive records of patent counts and citations provide informative metrics for the quantity and quality of successful research. They also make it possible to distinguish between basic and applied innovation, as well as to capture MNCs’ innovation in countries both with and without affiliates.
For each German parent company, we use the location of its subsidiaries and patent filers to identify patents developed at home, abroad in a country with an affiliate, and abroad in a country without an affiliate. We distinguish between basic R&D that advances fundamental knowledge and applied R&D that adapts fundamental knowledge to production uses, according to each patent’s distance to science proxied by backward citations to scientific articles (Ahmadpoor and Jones 2017). We quantify patent quality with the number of forward citations by subsequent patent applications (Hall et al 2005).
Our findings highlight interdependencies in MNCs’ production and innovation, with important policy implications. In particular, there is complementarity between offshore production and innovation, between basic and applied innovation, and in innovation across countries. This supports national package reforms and multilateral deep integration that span trade, investment, and innovation policy to reduce costs to both production fragmentation and cross-border investment, while also promoting R&D. This alleviates concerns in developed countries about offshore innovation displacing domestic innovation, and reinforces incentives to attract FDI to developing countries.
Our results also recognise the potential for cutting-edge research in both advanced and emerging economies to contribute to global growth, where the former have comparative advantage in basic innovation and the latter in manufacturing and synergistic applied innovation. Intuitively, a German pharmaceutical company may pursue fundamental scientific discovery (e.g. new chemical reaction) in technologically advanced nations like the US or Switzerland, but conduct applied product or process innovation (e.g. designing a new combo-vitamin pack or reducing gas dissipation in operations) alongside production in developing countries with favourable manufacturing costs like China, to benefit from two-way communication between scientists and production/sales managers on site.
Multinational companies innovate intensively across the globe
Multinationals are highly innovative, and pursue R&D across the globe. Thirty percent of all German MNCs file at least one patent during 1999-2016; of those, 43% develop patent(s) outside of Germany. Moreover, MNCs innovate in multiple home and foreign locations, including countries both with and without affiliates. As Figure 1 illustrates, 56% of all patenting multinationals conducted the underlying R&D exclusively at home. Some 8% innovated both at home and in another country where they maintain an affiliate, while 17% did so both at home and in another country with no subsidiary. Fully 15% of MNCs undertook patented research in all three location types, while innovating exclusively abroad is extremely rare.
Figure 1 Location of global MNC innovation
Notes: This Venn diagram summarises the global organisation of German MNC patent activity in 1999-2016. Each segment indicates the share of firms that file EP patents with inventors residing at home in Germany, offshore in a country with an MNC affiliate, and/or offshore in a country with no MNC affiliate. N = 2,030 MNCs.
Data sources: Research Data and Service Center of the Deutsche Bundesbank (MiDi) and PATSTAT, authors’ calculations.
Figure 2 plots the number of German MNC patents emanating from an inventor country over the 1999-2016 period against the country’s GDP per capita. Offshore R&D is concentrated in rich, developed, and often proximate economies at the technological frontier, with the top hubs being the US (19%), France (8%), Austria (7%), and Switzerland (6%). Nevertheless, numerous other countries across the income distribution attract non-trivial offshore innovation activity.
Figure 2 Number of MNC-filed patents
Notes: The figure plots the log total number of patents invented in a given country against the country’s average log GDP per capita in 1999-2016. Patents with inventors from multiple locations are assigned to each country using equal fractions. The sample comprises 39 countries hosting innovation by at least 10 MNCs.
Data sources: Research Data and Service Center of the Deutsche Bundesbank (MiDi), PATSTAT and World Bank National Accounts, authors’ calculations.
MNCs optimise complementary production, basic, and applied innovation across countries
To better understand MNCs’ global production and innovation strategies, we theoretically model heterogeneous firms that jointly choose the location and scale of their production, basic innovation, and applied innovation to maximise total profits. Firms may pursue each R&D type in multiple countries, either alongside a production affiliate or by subcontracting innovation to another party. The returns to innovation are additive across countries, with applied innovation raising profits immediately by lowering marginal production costs and basic innovation increasing future expected profits (e.g. by facilitating future applied innovation). Site-specific innovation costs rise with innovation intensity and inventor wages, but applied R&D is less costly when co-located with production.
This model delivers testable, policy-relevant predictions. First, more productive multinationals are more likely to innovate and to do so more intensively. Second, more productive MNCs are more likely to conduct R&D both at home and abroad in more countries, both with and without a subsidiary. Third, MNCs innovate more intensively in countries with lower inventor wages and, if these vary across sectors, to follow countries’ comparative advantage for innovation in different technology areas. Finally, MNCs are more likely to co-locate production with applied innovation, compared to basic innovation.
In our empirical analysis, we establish model-consistent correlations between joint outcomes of the MNC firm problem (global sales and innovation performance) and exploit exogenous variation in RCA across countries and technology areas within firms.
Bigger MNCs develop more patents, at higher quality, and in more locations
The data reveal that German MNCs with higher global sales are more likely to file patents. Conditional on patent activity, bigger MNCs generate more patents that receive more citations on average. We illustrate this key finding from our rigorous regression analysis with bin-scatter plots for the raw data in Figure 3.
Figure 3 MNC size and innovation intensity and quality
a) Innovation intensity
b) Innovation quality
Notes: The binscatters plot the log average annual number of EP patents per firm in 1999-2016 and the average number of 5-year forward citations per EP patent per firm in 1999-2011, by firm size bin. German MNCs are assigned to ten bins each year according to their annual global sales. Year fixed effects are absorbed.
Data sources: Research Data and Service Center of the Deutsche Bundesbank, MiDi, 1999-2016, combined with PATSTAT, own calculations.
Larger firms are also more likely to innovate both at home and offshore. Along the extensive margin, bigger MNCs have a higher probability of innovating in at least one foreign country, innovate in more countries on average, and file patents both in locations with and without an affiliate. Along the intensive margin, bigger MNCs develop a greater share of their patented technologies abroad.
MNCs follow countries’ comparative advantage in innovation across technology areas
German MNCs also respond to cross-country differences in comparative advantage in innovation. We construct a novel measure of innovation RCA based on the number of patents filed in each country and technology class (by non-German firms). Patent activity strongly follows comparative advantage across countries and technology fields within firms. For example, MNCs are more likely to research energy and transportation technology in Germany, computer and telecommunication technologies in the US, measurement and medical technologies in Switzerland, and machine tools and metallurgy in Austria.
MNCs see greater synergies in locating production with applied innovation, relative to basic innovation
All of the patterns above hold when we distinguish between MNCs’ basic and applied innovation activity. At the same time, multinationals are more likely to co-locate applied R&D with production, compared to basic R&D. This provides rationale for the simultaneous rise in applied patents filed in less developed countries with attractive manufacturing wages and in basic patents in more advanced economies with unique scientific talent.
Policy implications
Our findings shed new light on the design of global trade, investment, and innovation policy. They inform the scope for multilateral deep agreements that span tariff reductions, investment promotion and intellectual property rights protection, especially as developed and developing countries occupy different segments of global value chains and engage differently in technological innovation and adoption. Our work points to complementarity rather than substitutability in innovation activity across countries, which may alleviate concerns about the impact of offshore innovation on sending economies. Moreover, improved opportunities for production offshoring can incentivise innovation both at headquarters and abroad, enabling countries at different stages of technological development to engage with R&D. Finally, MNCs’ global operations may shape the impact of technological leaps such as automation on the global distribution of production, innovation, and adoption, and thereby on economic growth across countries.
Source : VOXeu
