Foreign Direct Investment

The gains from multinational production

This section reviews research that quantifies the aggregate impacts of multinational production. This is often measured by estimating the gains in real income if firms did not face prohibitively costly barriers to producing in foreign countries. On a broader scale, the costs linked to expanding production across borders may encompass fixed costs of operating overseas and obstacles to transferring technology internationally. These challenges include the inherent difficulties of relocating managers from their home country to distant places with diverse cultures and time zones, leading to potential coordination problems among foreign affiliates and headquarters.

MNEs significantly shape a country’s productivity through different avenues, which are considered in the models discussed in this section. This is particularly crucial for developing economies with limited domestic innovation. One major avenue is the direct boost in productivity for foreign affiliates inheriting technology from headquarters, contributing to an overall rise in aggregate productivity (Guadalupe et al. 2012). Other “less direct” channels are also at work in quantitative models. Specifically, these large and highly productive firms influence the host country’s aggregate productivity by increasing competition in product and factor markets, prompting the exit of less productive firms, and reallocating resources from less productive to more efficient enterprises (Alfaro and Chen 2018). However, the models discussed below often don’t account for technological spillovers from multinationals to domestic firms. This includes, for example, the technological upgrading encouraged amongst domestic firms when they provide intermediate inputs to MNEs that require them to meet stringent standards. Those spillovers are discussed in Section 4.

The remaining discussion is organised as follows. First, we explore some of the most important quantitative models of MP, examining their key features and welfare implications. Second, we assess how factors such as sectoral differences, capital-intensive technologies, fixed costs of production, and intra-firm trade can affect the gains from multinational activity. Finally, we conclude with some reflections and suggestions for future research.

Quantifying global gains: How impactful are MNEs?

Recent research has used quantitative analysis to examine the central role of MNEs in generating welfare gains. Ramondo (2014) initiates this investigation by constructing a quantitative multi-country model that assesses the overall impact of multinational production. Within this model, transferring technologies from the home country to a host country is subject to a proportional loss of productivity – the so-called ‘’multinational production frictions". These multinational production frictions refer to the costs of adjusting a foreign technology to a local environment, which can be prohibitively high and therefore inhibit the activity of MNEs in a country.

Analysing data from developed and developing nations, Ramondo (2014) reveals substantial real income per-capita gains from multinational production. Relative to a world with no multinational production, observed levels of multinational activity result in a 5% increase in income for richer countries and a 3.5% increase for poorer countries. Higher-income nations experience higher increases as larger shares of their production are led by foreign companies, highlighting their lower impediments to cross-border investments.

However, the potential for international production is closely tied to trading capabilities. In this context, production abroad can either replace or complement traditional trade. Trade and multinational production are competing ways of accessing foreign markets, as a company may choose a foreign affiliate to serve a market locally instead of exporting. Yet, trade can also complement foreign production in two cases: first, when a foreign affiliate requires inputs from, or sells them to, its parent, and second, when a destination is supplied with exports from a facility overseen by foreign affiliates, commonly referred to as export platforms. A country’s ‘openness’ should account for these two channels, with a country being more open when both trade and international production are less costly.

To deepen our understanding of the gains from openness, Ramondo and Rodriguez-Clare (2013) integrate multinational activity and trade into a multi-country general equilibrium model. Within this framework, the MNE chooses the location with the minimum unit cost for each product to each market, to set up production facilities. In doing so, the firm considers trade costs, labour and capital expenses, and the productivity of foreign affiliates, which mirrors headquarters’ efficiency adjusted by the cost of transferring technology from home to the production facility. Essentially, low-wage countries with convenient access to consumers, both domestically and in third markets, coupled with lower discount rates for using home technology, emerge as the favoured hubs for multinational corporations. Their model, calibrated to nineteen OECD countries, suggests that the average gains from openness are between 15% and 22%. These gains exceed more than twice the average from models with only trade (ranging from 6% to 7%) and roughly double those derived from models focusing solely on multinational production (ranging from 9.1% to 11.6%). This underscores that relying on models with only trade or only multinational production may lead to underestimating welfare gains.

So far, we have discussed two main frictions shaping trade and multinational production patterns: trade barriers, like tariffs and transportation costs, and multinational production barriers linked to technology transfer hurdles. A third friction, highlighted by Head and Mayer (2019), is the difficulty of expanding sales into geographically distant markets from the brand’s headquarters, since translating a brand’s success in home markets into foreign markets is costly. Using comprehensive data on cross-border production and global sales of automobiles, Head and Mayer (2019) demonstrate each friction’s qualitative and quantitative impact on production reallocation due to economic integration. Their findings highlight that consumers are willing to pay twice as much for a domestically designed car model produced abroad compared to a locally assembled car with a foreign design.

How do differences between sectors and capital-intensive technologies shape the gains from multinational production?

Estimates of the gains from receiving MNEs can be severely biased if the fact that MNEs have unequal presence across industries, and that those industries have different degrees of comparative advantages depending on the country, is overlooked. Alviarez (2019) sheds light on this issue, revealing that foreign affiliates are more prevalent in industries where local firms have relatively lower productivity. The intuition for this result is that foreign affiliates are more likely to succeed in sectors where their productivity differential relative to local producers is larger. Since this differential is expected to be larger in developing countries, bigger effects are expected for these economies. This results in a productivity boost across all sectors, with a more pronounced effect in those with a comparative disadvantage, thereby reducing disparities in observed sectoral productivity. In her multi-sector extension of Ramondo and Rodriguez-Clare (2013), Alviarez (2019) demonstrates that lowering frictions to foreign production has dual effects: on the one hand, it diminishes gains from trade, as multinationals erode part of the sector-level Ricardian comparative advantage of domestic firms; on the other hand, simultaneously, it amplifies the dispersion of multinational production shares across sectors, yielding gains from multinational production that are sufficient to offset the lower gains from trade, hence delivering larger gains from openness than one-sector models.

Another important aspect to consider is the close ties between the welfare gains from multinational activity and the intensity with which MNEs utilise factors of production. Typically, these corporate giants employ technologies that lean heavily on capital. Moreover, firms originating from capital-abundant home countries also prefer capital-intensive technologies. Sun (2020) explicitly highlights this bias, emphasising its substantial implications. The liberalisation of multinational production sets in motion a reallocation of factors across firms with differing biases, thereby reshaping the aggregate demand for capital vis-à-vis labour. Sun (2020)’s estimates pinpoint an average reduction in multinational production costs of 8 percentage points from 1996-2001 to 2006-2011. The subsequent increase in multinational activity, facilitated by lower frictions, accounts for about 60% of the average decline in the observed labour share during the same period. This trend is particularly pronounced in countries witnessing a substantial increase in multinational activity. The reduced frictions in multinational operations lead to an average decline in real wages of 2 percentage points, in contrast to gains in the real return to capital of 3.2 percentage points, resulting in a net gain for the average country.

The geographical dispersion of foreign affiliates: The role of fixed costs

As globalisation unfolds, multinational production has drastically shifted the economic landscape, separating the creation of ideas from the manufacturing process. This global split between innovation and production sparks concerns. Nations excelling in production fear a growth slump without innovation, while innovation powerhouses worry about shrinking middle income job opportunities as manufacturing production is located elsewhere. Arkolakis et al. (2018) delved into these issues, proposing a model to quantify the welfare implications of increased specialisation in innovation and production by allowing the development of new varieties carried by innovative new firms. Using their calibrated model, they find that a 5% decline in all multinational production costs, including technology transfer expenses, triggers an increase in global specialisation, boosting average real income by 2%. Contrary to the public’s common worries, production workers benefit everywhere, while innovation workers face challenges in countries facing a contracting innovation sector.

The tractability of this workhorse model is achieved by including fixed costs of innovation but abstracting away from the fixed costs associated with foreign investment, therefore avoiding a preference toward concentration. Incorporating the fixed costs of opening a foreign affiliate substantially complicates the analysis by making the location decisions to serve each market interdependent. For example, minimising marginal costs might entail a firm selling to Chinese customers from a Cambodian affiliate and selling to Japan from an Indonesian affiliate, but the desire to concentrate production in one location due to high fixed costs might imply that Vietnam is the best location, or ‘export platform’ to serve both destination markets (see Antràs and Yeaple 2014).

To address this concern, Tintelnot (2016) develops a quantifiable multi-country general equilibrium model, where multinational firms that engage in export platform sales face fixed costs of investing abroad. Using German firm-level data Tintelnot (2016) estimates fixed costs and finds that those play a large role in the geographic structure of multinational production and are necessary to explain the magnitude of export platform sales in the data. With the calibrated model, Tintelnot (2016) explores the importance of export platforms and third-country effects. He studies this question in the context of a Canada-EU deep trade agreement, which includes foreign investment. In particular, he finds that EU multinationals would divert around 5% of their production from the US to Canada, had the agreement achieved a 20% reduction of variable and fixed production costs. Most of these gains are based on European MNEs’ ability to use Canada as an export platform to serve the US market.

Acknowledging the importance of fixed costs is crucial to understanding the geographic organisation of multinational firms – why a company selects a handful of locations for production to serve dozens of markets. For MNEs to select developing economies as production facilities they must believe they can run long term operations in these locations, and they can also serve nearby markets. In developing countries, MNEs encounter unique fixed costs when establishing production facilities. These costs often revolve around addressing infrastructure deficiencies, navigating regulatory complexities, and managing operational risks distinct from those in developed nations. For instance, substantial investments are needed to boost inadequate infrastructure, including upgrading facilities and enhancing transportation networks. Moreover, regulatory compliance in developing countries can be more intricate, calling for significant investments in learning how to navigate the bureaucratic processes, obtaining permits, and ensuring adherence to evolving standards. Additionally, MNEs may face higher security risks, prompting investments in robust security measures to safeguard personnel and assets. Contrastingly, while developed countries may entail lower infrastructure and regulatory compliance costs, they may present different challenges, such as higher labour costs and stringent environmental regulations. Therefore, in developing countries, effectively managing fixed costs becomes crucial for MNEs to establish a sustainable presence while contributing to local economic development and infrastructure advancement.

How does trade within firms shape the gains from multinationals?

When foreign direct investment is vertical in nature, intra-firm trade arises as a necessary condition that allows the production process to be split into different stages across countries. Multinationals emerge when internalisation proves more profitable than outsourcing from an independent company, affecting the gains from multinational activity.

One motive for cross-border vertical integration lies in the inherent limitations of international contracts, driven by contractual challenges in legal systems, institutional quality, and enforcement ability (Antràs 2003), which are particularly pronounced in developing countries. Additionally, technological disparities and imperfect competition can play a role. Garetto (2013) proposes a model in which outsourcing provides access to superior technology, but firms pay a markup to their suppliers, while integration confines a firm to use in-house technology but saves on labour costs and supplier mark-ups since goods are priced at marginal cost when traded within firm boundaries. Garetto (2013)’s results suggest that a 50% reduction in the barriers to offshoring (through foreign direct investments) could yield a gain of approximately 6% in per capita consumption.

Intra-firm trade also arises when foreign affiliates require inputs from headquarters for embodied knowledge transfer (traded intermediates) rather than disembodied knowledge (direct communication) (Keller and Yeaple 2013). Intra-firm trade from headquarters to affiliates has been used to rationalise why foreign affiliates produce less in more distant locations, also known as the gravity of multinational production (Irarrazabal et al. 2013, Keller and Yeaple 2013). Introducing vertical trade linkages from parents to affiliates, Irarrazabal et al. (2013) show that multinationals may reduce home employment by up to 50% if foreign production is prohibited. This is because reducing barriers to multinational production positively impacts domestic labour since outward multinational activity complements substantial economic activity at home.

Policy implications and next steps

The discussion above highlights previous research’s efforts to capture and quantify key aspects of multinationals’ activities, what drives their location decisions, and implications for welfare. These models have enhanced our understanding of the factors shaping the behaviour of MNEs, including the role played by the transfer of home productivity to foreign facilities, the use of local and home-country factors of production, as well as the role played by barriers, such as trade costs, tariffs, RTAs, domestic infrastructure, governance, and corruption, in attracting FDI.[1]1.See, for instance, Mensah and Traore (2024) on the importance of improving infrastructure quality in attracting FDI in Africa. However, there remain challenges to be addressed both theoretically and quantitatively in order to fully grasp the complexity of MNEs and of their effects on the economy.

So far, the existing literature primarily focuses on either the horizontal or vertical dimensions of multinational operations. However, a company that replicates its production process in a specific foreign location may also find incentives to fragment various stages, including the manufacturing of intermediate inputs across countries and vice versa. Antràs and Yeaple (2014) refer to both forms of multinational production within the firm as “complex strategies”. Future work should focus on making our quantitative models able to handle these inherent complexities.[2]2.Arkolakis et al. (2023) is a recent effort in this direction.

Data availability also poses a common challenge in multinational studies, especially in developing countries. Most country-sector level datasets that track the nationality of foreign producers and the destination of foreign production only focus on OECD and EU countries. Firm-level datasets are even more scarce, with significantly limited coverage for low-income countries. Fortunately, recent years have witnessed a surge in more detailed firm-level datasets, facilitating a more in-depth exploration of multinational behaviour. Closely tracking the activity of multinationals entails knowing information regarding their ownership structure, where and what they produce, as well as their sourcing patterns of intermediate inputs from within and outside the firm, domestically and across borders. Some recent data with some of these characteristics, often available for individual countries, have been assembled by matching different datasets such as customs data, economic censuses, and value-added tax databases often maintained by the country’s tax authorities. This information is useful for estimating models of multinational production, on which we rely to quantify the gains from different forms of economic integration.

Quantifying the gains of multinational production at the country-level is particularly informative for policymakers. The main reason is that this literature incorporates multiple channels that can help governments to better understand the overall consequences of their policies. The lower tariffs achieved by a free trade agreement will have consequences not only on trade flows but also on the geographical structure of multinational production. Recent papers have evaluated the potential consequences of more recent integration agreements, disagreements, and tax policies such as CETA (EU and Canada Trade and Investment Agreement), Brexit, and the tensions between the US and China (e.g. Head and Mayer 2019, Arkolakis et al. 2023, Garetto et al. 2019, Dyrda et al. 2024).

Global R&D and MNEs

So far, we have reviewed the literature that quantifies the global impacts of multinational production. A less discussed, but not less important, aspect of the activities of MNEs is their presence in research and development (R&D) across the globe.

Globalised R&D: Data sources and general patterns

Contrary to the conventional view that when MNEs conduct R&D abroad they do so primarily in developed countries, recent anecdotes suggest a growing trend for these firms to establish R&D facilities in middle-income countries. For instance, according to fDi Intelligence, a Financial Times publication on foreign direct investment (FDI), in the first ten months of 2022, India received more foreign R&D investment projects than the United States, Germany, and Japan combined.

These anecdotes are corroborated by data from three principal sources. The first is surveys of multinationals on their activities at home and abroad, often administered by the headquarters country of these firms. The US Bureau of Economic Analysis (BEA) survey is used often by researchers in this area. According to this survey, the share of R&D expenditures by US manufacturing multinationals that was conducted at foreign affiliates increased from 9% in 1994 to 14% in 2004 (Bilir and Morales 2020). Using data over a longer period, Branstetter et al. (2019) document continued growth in affiliate R&D by US multinationals after 2004. In industries such as Computer and Electronics, over 20% of R&D by these firms is now conducted abroad. These facts hold across a range of other service industries that use Science, Technology, Engineering, and Mathematics (STEM) workers intensively. Importantly, throughout the sample period (the 90s and 2000s), the growth rate of affiliate R&D has consistently been higher in developing countries, such as China and India, than in developed countries, resulting in a shift in the global distribution of R&D by US multinationals.

Based on firm-level surveys at the affiliate level, the OECD compiles a dataset on foreign involvement in business enterprise R&D (as opposed to R&D carried out in universities and research institutions) across OECD countries. Analysis of this dataset suggests that the median country in the sample has around a third of domestic R&D carried out by foreign firms. Moreover, there has been an across-the-board increase in this share in recent decades (Fan 2023). This trend aligns with the general increase in affiliate R&D by US multinationals documented using the BEA data. It also complements this finding by showing that global R&D by foreign firms could be a substantial part of host country R&D even if it is relatively small compared to the overall R&D spending of the parent firms.

Another valuable source of information for understanding global R&D is patent data. When filing for patents, applicants (usually the firm) often provide the address of inventors, which indicates where the R&D behind an invention occurs. By linking patent data to firm-level databases, one can explore the patterns of global R&D by multinationals and the relationship between these patterns and the patterns of multinational production (Hall 2011). This approach has several advantages over using R&D statistics from firm-level surveys: it offers wide accessibility and is less likely to be manipulated by multinationals for profit shifting across jurisdictions. Fan (2023) links virtually the universe of world patents from PATSTAT to the ORBIS firm-level database. His analysis confirms that, first, for country pairs where R&D expenditures are available, multinational R&D measured using patent data generally highly correlates with R&D expenditures. Second, the multinational R&D in tax haven countries constructed from patent data is substantially lower than that from R&D expenditures, highlighting the value of measuring multinational R&D using patent data.

Why do MNEs increase R&D investment in developing countries?

A large literature has examined the rationale behind multinationals’ expanding global R&D. In a survey of the factors influencing multinationals’ choice of R&D locations, Thursby and Thursby (2006) find that the top two factors for location choice among emerging countries are output markets and the quality of R&D personnel. Output market considerations include, for example, the choice of an R&D location either to support local production and sales or for product customisation. Considerations on the quality of R&D personnel mean that firms have incentives to enter host countries with an abundant supply of qualified workers at an affordable wage. These incentives can explain why major firms have been racing to countries with a large number of qualified engineers, such as India, to set up R&D centres.

Patterns of the location of global R&D by multinationals documented in recent studies are consistent with these rationales. Bilir and Morales (2020), for example, find that R&D conducted by affiliates of multinationals tends not to generate spillovers to other affiliates, suggesting that such R&D is primarily tailored to production and customers in the host country. In different settings, Fan (2023) and Fort et al. (2020) document significant colocation between R&D and production within a firm, indicating that R&D locations are often chosen based on their proximity to suitable production locations. Consistent with the importance of qualified R&D personnel for the location choice, Fan (2023) also finds that the intensity of affiliate R&D is positively correlated with the quality of host country talent.

General equilibrium models of R&D

To analyse the aggregate effects of multinational R&D, Fan (2023) develops a general equilibrium model incorporating multinational production, multinational R&D, and trade. In the model, firms jointly determine where to conduct R&D and where to carry out production. The location choice of R&D features the two rationales identified above: first, because R&D relies on qualified researchers, host countries that are relatively abundant in researchers are attractive; second, because separating R&D from production is costly, firms have an incentive to locate R&D where production is likely to occur—locations with high manufacturing productivity, low wages, and/or easy access to final consumers.

In this model, the entry of foreign firms to conduct R&D in the host country has several implications: first, it improves the job prospects of researchers; second, due to the co-location between R&D and production, some new inventions will be produced locally, which increases the demand for production-line workers; lastly, the increase in R&D taking place in the host country also means there will be more varieties catering to the taste of local consumers. On the downside, foreign firms compete with domestic firms both in the product market and in the markets for researchers and production workers.

Fan (2023) finds that, when the model is parameterised to the data, for the host country as a whole, the benefits outweigh the costs. In particular, he finds that each country on average experiences a 2.1% increase in real income from multinational firms’ globalised R&D. The benefits tend to be larger for countries that are home to many multinationals, such as the US, who receive profits from their overseas affiliates. But developing countries also receive positive benefits. For example, China’s real income increases by 0.88%, Brazil’s by 0.68%, and Mexico’s by 0.66%. These benefits interact with trade and multinational production. In particular, for developed countries that already have a comparative advantage in R&D, inward R&D carried out by foreign multinationals tends to reinforce that comparative advantage, facilitating the specialisation of these countries in R&D. As a result, multinational R&D is complementary to globalisation via trade and multinational production for these countries. On the other hand, multinational R&D tends to be a substitute for other forms of globalisation in developing countries.

Policy implications and next steps

The literature that applies a quantitative approach to studying multinational R&D is still small, but it offers several implications relevant to policymaking. First, multinational R&D can be an important channel for countries to benefit from globalisation. Even for developing countries, the gains are not trivial. Second, in terms of attracting foreign firms’ R&D, given the important co-location between R&D and production, a well-developed manufacturing sector can make a host country more attractive for foreign R&D. The talent acquisition motive also implies that cultivating a suitable workforce in targeted industries can be effective.

A few important mechanisms through which multinational R&D can affect the host country have not been incorporated into structural models. For example, a compelling case for encouraging foreign firms to carry out R&D in a host country is that they can bring knowledge spillovers to domestic workers. Supporting this mechanism, Kerr and Kerr (2018) find that many of the inventions by multinationals in foreign locations are co-invented by local inventors and inventors from the headquarters. These ’co-invented’ patents are more influential than patents invented only by investors in the host country, suggesting potential knowledge spillovers from the headquarters to host countries. Future work should incorporate such spillovers to assess their importance relative to other channels for spillovers analysed in detail in Section 4.

Global value chains and the global transmission of shocks

The complex global supply chains of MNEs create links between countries that can facilitate the transmission of foreign shocks. For instance, the 2011 Tohoku Earthquake and Tsunami in Japan was transmitted to the US largely through the inelastic supply chains of Japanese MNE affiliates. US manufacturing employment fell by 1% as a result (Boehm et al. 2019).

While global supply chains increasingly span countries across the world at all stages of development, the role of MNE supply chains within the larger global value chain is still not well understood. Previous research studying the transmission of shocks through global value chains has largely focused on sectoral data, due to the ready availability of multi-country input-output tables such as the World Input Output Database (WIOD) or the OECD Inter-Country Input Output tables.[3]3.See, among others, Johnson (2014) or Huo et al. (2019), who show that such I-O data is sufficient to calibrate a large multi-country multi-sector business cycle model of shock propagation in global value chains, or Baqaee and Farhi (2024) who study the response to sectoral shocks in an international trade model with production networks. While these papers have shown that the transmission of shocks through global value chains is important to understand aggregate fluctuations and international co-movement, data limitations have meant that the specific contribution of MNE supply chains to fluctuations worldwide — and to developing countries specifically — is less studied. The sample of countries included in these studies has also been limited by the countries present in the I-O tables. Bonadio et al. (2021) considered the role of global value chains in transmitting the COVID-19 lockdown shock across 63 countries, and found that around a quarter of GDP declines observed on average could be attributed to transmission through global value chains. The countries considered included a number of middle-income and developing countries. However, the distinguishing feature of whether or not countries were more exposed to the shocks through their value chain was not their income level. Rather, their openness to trade and participation in value chains combined with their policy-desired lockdowns in response to the shocks determined whether they were more or less affected by transmission through value chains. 

Countries with policy preferences for very strict lockdowns, which spanned the income distribution, (e.g. Peru, Singapore) mitigated the effect of their domestic lockdowns on their economy by participating in the global value chain, as their imported inputs were relatively less disrupted than their domestic inputs. Also in the context of the COVID-19 pandemic, Çakmaklı et al. (2021) use a multi-country multi-sector model to study how low vaccination rates and high infection rates in developing countries transmit to output losses in developed countries through higher imported intermediate input prices and low demand for developed country exports. They show that with high complementarity of inputs, as estimated by Boehm et al. (2019), the transmission of low vaccination rates could be economically large.

While the research using multi-country multi-sector quantitative models of global value chains is rapidly developing, a related literature has studied the role of MNEs in propagating shocks across countries using firm-level data. Cravino and Levchenko (2016) use firm-level data from ORBIS and find an important role for MNEs in shock transmission. They find that, for a typical country in their data, the impact of foreign shocks transmitted through multinationals can account for about 10% of aggregate productivity shocks. However, bilateral multinational production shares are small, and so the multinational headquarter-affiliate links create minimal interdependence between individual country-pairs. Data limitations in the ORBIS dataset mean that the focus of this research is primarily developed countries, and while multinationals and affiliates can be identified, a more detailed view of the role of multinational supply chains (which could span inputs sourced within and across firms by the multinational parent) is not feasible. In related work, Kleinert et al. (2015) use French data to show that regions in France with affiliates of foreign multinationals co-move with the country of origin of the affiliates’ parents. More recently, Biermann and Huber (2023) show that the internal capital markets of MNEs can transmit shocks across countries using data from Germany.

Policy implications and next steps

Summing up, there is empirical and theoretical work that highlights the role of global value chains in the transmission of shocks across countries. While MNEs are likely the largest players in the global value chain, data constraints have limited studies of the specific role of MNEs’ supply chains for the transmission of shocks across developed and developing countries. Studies of MNEs and their affiliates, without identifying the full supply chain of these firms, already find evidence on their importance for international business cycle synchronisation. A more comprehensive study of the role of these firms’ complex global supply chains in the transmission of shocks remains a promising avenue for future research.

References

Alfaro, L and M X Chen (2018), "Selection and Market Reallocation: Productivity Gains from Multinational Production", American Economic Journal: Economic Policy 10(2): 1–38.

Alviarez, V (2019), "Multinational Production and Comparative Advantage", Journal of International Economics 119: 1–54.

Antràs, P (2003), "Firms, Contracts, and Trade Structure", The Quarterly Journal of Economics 118(4): 1375–1418.

Antràs, P and S R Yeaple (2014), "Multinational Firms and the Structure of International Trade" in Gopinath, G, E Helpman, and K Rogoff, eds, Handbook of International Economics 4. Elsevier.

Arkolakis, C, F Eckert, and R Shi (2023), "Combinatorial discrete choice: A quantitative model of multinational location decisions", NBER Working Paper No. 31877.

Arkolakis, C, N Ramondo, A Rodríguez-Clare, and S Yeaple (2018), "Innovation and production in the global economy", American Economic Review 108(8): 2128–2173.

Baqaee, D and E Farhi (2024), "Networks, Barriers, and Trade", Econometrica 92(2).

Biermann, M and K Huber (2023), "Tracing the international transmission of a crisis through multinational firms", NBER Working Paper No. w31061.

Bilir, L K and E Morales (2020), "Innovation in the global firm", Journal of Political Economy 128(4): 1566–1625.

Boehm, C E, A Flaaen, and N Pandalai-Nayar (2019), "Input Linkages and the Transmission of Shocks: Firm-Level Evidence from the 2011 Tohoku Earthquake", Review of Economics and Statistics 101(1): 60–75.

Bonadio, B, Z Huo, A A Levchenko, and N Pandalai-Nayar (2021), "Global supply chains in the pandemic", Journal of International Economics 133: 103534.

Branstetter, L G, B Glennon, and J B Jensen (2019), "The IT revolution and the globalization of R&D", Innovation Policy and the Economy 19(1): 1–37.

Çakmaklı, C, S Demiralp, Ṣ Kalemli-Özcan, S Yesiltas, and M A Yıldırım (2021), "The economic case for global vaccinations: An epidemiological model with international production networks", NBER Working Paper No. w28395.

Cravino, J and A A Levchenko (2016), "Multinational Firms and International Business Cycle Transmission", The Quarterly Journal of Economics 132(2): 921–962.

Dyrda, S, G Hong, and J B Steinberg (2024), "Optimal taxation of multinational enterprises: A Ramsey approach", Journal of Monetary Economics 141: 74–97.

Fan, J (2023), "Talent, geography, and offshore R&D", Working Paper.

Fort, T C, W Keller, P K Schott, S Yeaple, and N Zolas (2020), "Colocation of production and innovation: Evidence from the United States", Working Paper.

Garetto, S (2013), "Input sourcing and multinational production", American Economic Journal: Macroeconomics 5(2): 118–151.

Garetto, S, L Oldenski, and N Ramondo (2019), "Multinational expansion in time and space", NBER Working Paper No. w25804.

Guadalupe, M, O Kuzmina, and C Thomas (2012), "Innovation and foreign ownership", American Economic Review 102(7): 3594–3627.

Hall, B H (2011), "The internationalization of R&D", SSRN Working Paper 2179941.

Head, K and T Mayer (2019), "Brands in motion: How frictions shape multinational production", American Economic Review 109(9): 3073–3124.

Huo, Z, A A Levchenko, and N Pandalai-Nayar (2019), "International comovement in the global production network", NBER Working Paper No. w25978.

Irarrazabal, A, A Moxnes, and L D Opromolla (2013), "The margins of multinational production and the role of intrafirm trade", Journal of Political Economy 121(1): 74–126.

Johnson, R C (2014), "Trade in intermediate inputs and business cycle comovement", American Economic Journal: Macroeconomics 6(4): 39–83.

Keller, W and S R Yeaple (2013), "The gravity of knowledge", American Economic Review 103(4): 1414–1444.

Kerr, S P and W R Kerr (2018), "Global collaborative patents", The Economic Journal 128(612): F235–F272.

Kleinert, J, J Martin, and F Toubal (2015), "The few leading the many: Foreign affiliates and business cycle comovement", American Economic Journal: Macroeconomics 7(4): 134–159.

Mensah, J T and N Traore (2024), "Infrastructure Quality and FDI Inflows: Evidence from the Arrival of High-Speed Internet in Africa", The World Bank Economic Review 38(1): 1-23.

Ramondo, N (2014), "A quantitative approach to multinational production", Journal of International Economics 93(1): 108–122.

Ramondo, N and A Rodriguez-Clare (2013), "Trade, multinational production, and the gains from openness", Journal of Political Economy 121(2): 273–322.

Sun, C (2020), "Multinational production with non-neutral technologies", Journal of International Economics 123: 103294.

Thursby, M and J Thursby (2006), Here or There?: A Survey of Factors in Multinational R&D Location–Report to the Government-University-Industry Research Roundtable, Washington D.C: National Academies Press.

Tintelnot, F (2016), "Global Production with Export Platforms", The Quarterly Journal of Economics 132(1): 157–209.