8
Keynote Address: Foreign Direct Investments and Superstar Spillovers: Evidence from Firm-to-Firm Transactions
Paper Authors: Mary Amiti (Federal Reserve Bank of New York and Centre for Economic Policy Research [CEPR]), Cedric Duprez (National Bank of Belgium), Jozef Konings (Nazarbayev University and Katholieke Universiteit Leuven), and John Van Reenen (London School of Economics, Massachusetts Institute of Technology, British Academy, Econometric Society, National Bureau of Economic Research, CEPR, and Society of Labor Economists)
Presenter: John Van Reenen (London School of Economics)
Moderator: Wolfgang Keller (University of Colorado Boulder)
John Van Reenen, Ronald Coase School professor at the London School of Economics; digital fellow in the Initiative for the Digital Economy at the Massachusetts Institute of Technology; and a fellow of the British Academy, Econometric Society, National Bureau of Economic Research, Centre for Economic Policy Research, and Society of Labor Economists, began the keynote presentation by noting that he will be discussing his and his coauthors’ new work in this area. Governments often encourage multinational enterprises (MNEs) to locate activities in their countries. MNEs are more productive and can offer higher wages than domestic firms, and their location in a country is often accompanied by new technologies and management practices, making them very attractive targets. MNEs offer spillovers to domestic firms as these new technologies and practices are introduced.
Van Reenen explained that the average quality of management practices of foreign MNEs is higher than that of domestic firms across the income distribution. The trend in domestic management practice roughly follows productivity distribution, with the United States, Japan, and Germany having the highest-quality management, while low-income countries rank lower in management quality. The distribution of average quality of management for affiliates of foreign MNEs is higher for every country in the sample and flatter across the host countries.
The rationale for government policies intended to attract MNEs is deeper than simply targeting high-productivity firms, Van Reenen said, as governments assume that MNE activity will have spillover benefits for domestic firms. Evidence from firm-level case studies (see Iacovone et al., 2015 on Wal-Mex; see also Bloom et al., 2013; Sutton, 2004) indicates that foreign MNEs can work with domestic firms in the supply chain to upgrade technology and practices.
However, as Van Reenen shared, the econometric evidence on MNE spillovers is mixed. Aitken and Harrison (1999) found negative effects from horizontal foreign direct investment (FDI), and Javorcik (2004) found positive evidence from downstream FDI. An issue with the econometric literature is the use of industry-level data on MNE exposure. If spillovers are due to direct relationships, then industry-level analysis is unlikely to identify the effect. Alfaro-Urena and colleagues (2019) exploited variation in firm-to-firm sales data in Costa Rica and found positive effects to total factor productivity (TFP) for firms that enter the supply chain of an MNE, using an event study approach.
Van Reenen and his coauthors wondered if the results from Costa Rica, an emerging economy, would generalize to richer countries. Furthermore, are the spillovers due to activity in the supply chain of foreign MNEs only, or could the spillovers be due to activity in the supply chain of any superstar firm, including firms that are either a large exporter or very large domestic firm? Lastly, if the answer to the prior questions is affirmative, then what is the mechanism to explain the behavior?
Finally, Van Reenen went on to describe the work of Greenstone and colleagues (2010), who studied the productivity gains from FDI in the United States. They compared counties that won a large “million-dollar plant” to those who just lost out. They did this by scraping news stories in Site Selection magazine, which contains information about winning and runner-up counties of large-scale infrastructure projects. Greenstone and colleagues (2010) found that incumbent plants in winning counties have higher productivity growth than incumbent plants in runner-up counties. Bloom and colleagues (2019) updated and supplemented the dataset with news coverage from other sources and added management quality as an additional outcome variable. Their report shows a positive overall treatment effect to management quality following entry by the MNE, domestic or foreign, using an event-study framework. The treatment effect is heterogeneous across industries with management spillovers, and larger in industries in which the flow of managerial information is likely to be larger. This suggests that managerial know-how is a significant mechanism for explaining spillovers between MNEs and firms in their supply chain.
FDI AND SUPERSTAR SPILLOVERS: EVIDENCE FROM FIRM-TO-FIRM TRANSACTIONS
Van Reenen moved to discussing his and his colleagues’ new work in this area. Using firm-to-firm panel data consisting of the universe of Belgian firms from 2002 to 2014 and event study analysis, the authors found about 10 percent
positive productivity effects after 5 years for firms that enter the supply chain of an MNE. These supplying firms also see an increase in outputs, inputs (labor, capital, and intermediate goods), and exports. The authors also found a positive effect on TFP for entering the supply chain of very large firms, even if they are not globally engaged or heavy exporters. They found no effect from entering the supply chain of nonsuperstar firms.
Van Reenen stated that two mechanisms have been proposed to explain these spillovers: (1) technology transfer, in which treatment effects are larger when the superstar is in an industry that is research and development (R&D) intensive, information technology related, or human capital intensive; and (2) match making, in which the number of buyers increases, particularly within the superstar’s network. These results suggest that the benefits of high-productivity firms extend beyond MNEs.
Van Reenan described this work as fitting into five strands of literature:
- MNE spillovers: Aitken and Harrison (1999), Smarzynska Javorcik (2004), Alfaro-Urena et al. (2019), Alvarez and López (2008), Keller and Yeaple (2009), and Keller (2021)
- Higher productivity of MNEs: Bloom et al. (2012), Helpman et al. (2004), Chaney (2014), Antràs and Chor (2013), Eaton et al. (2011), Antràs et al. (2017), Lim (2018), and Dhyne et al. (2021)
- The impact of large firm entry: Greenstone et al. (2010) and Bloom et al. (2019)
- Production networks: Acemoglu et al. (2012, 2017), Liu (2019), Acemoglu and Azar (2020), Atalay et al. (2011), and Iyoha (2021)
- The rise of superstar firms: Furman and Orszag (2018), Autor et al. (2017, 2020), Bajgar et al. (2019), Gutiérrez and Philippon (2019), and De Loecker et al. (2020)
DATA
Van Reenen went on to discuss data sources, sharing that the main data source for this work is the National Bank of Belgium (NBB) business-to-business transaction dataset, which contains the value of sales, all transactions worth more than 250 euros, in all buyer–seller relationships in Belgium based on value-added tax declarations. This is merged with company accounts data from the NBB Central Balance Sheet office. The company-accounts data cover all incorporated firms and include information on sales, labor, capital, and intermediate inputs. FDI information comes from the NBB FDI survey. Data on trade within the European Union were sourced from the Intrastat trade survey, and data on trade outside of the European Union come from customs data. The baseline productivity measure follows the Wooldridge (2009) approach, but this is compared with more recent methods, such as Collard-Wexler and De Loecker (2020), and a simple
measure of value added. The authors are working on implementation of the Iyoha (2021) method, but it is not likely there will be significant differences in results.
ECONOMETRIC STRATEGY
Van Reenen said he and his coauthors used an event-study differences-in-differences approach. Superstar firm j is defined in three ways:
- MNE, more than 10 percent foreign-owned and inward FDI;
- exporter, nonwholesaler with more than 10 percent of sales exported; and
- whether it is a large firm, the top 0.1 percent of the sales distribution.
Supplying firm i enters the supply chain at time t. The authors focused on “serious relationships,” in which the supplying firm sells more than 10 percent of its total sales to the superstar. The authors also included firm, industry, and year fixed effects.
BASELINE RESULTS
Van Reenen presented two figures displaying his and his coauthors’ results, starting with the increase in TFP for firms in the supply chain of an MNE. Five years after entering the supply chain of an MNE, the supplying firm’s TFP increases by about 9 percent. The increase in TFP is slightly lagged, taking 2–3 years to significantly impact the supplying firm (see Figure 8-1).
Next, the authors examined several other outcome variables of interest: total sales, total inputs, total fixed assets, number of buyers, and number of other buyers (those outside the firm’s network). For all outcomes listed, there is a significant and positive effect on entering the supply chain of an MNE (see Figure 8-2).
While the analysis above focuses on supplying to MNEs, Van Reenen said that he and his coauthors also studied the effects of supplying to exporting firms. Using the same outcome variables of interest as above, the results are all significant and positive. Lastly, the authors studied the impact of supplying very large firms. Again looking at the same outcome variables as above, there is a significant and positive effect of similar magnitude when a firm enters the supply chain of very large firms.
Van Reenen explained that these results imply that what matters is the relationship between very large firms and firms in their supply chain. However, there is an overlap between global firms and very large firms in a country, so the authors split their analysis into large global firms and large nonglobal firms. The TFP gains for supplying firms are positive and significant for both groups and larger for large nonglobal firms. This indicates that the results are not driven by global engagement, but rather by selling to large firms.
MECHANISMS
Van Reenen restated that he and his coauthors explored two mechanisms: technology transfer and match making. To understand the technology transfer motive, firms are split into R&D-intensive industries, information and computing industries, and human capital–intensive industries. Regardless of the type of firm (MNE, exporter, or large firm) the results are positive and significant, suggesting that spillovers are larger for high-tech and high-skill industries and that technology transfer is an important mechanism.
The second motive, Van Reenen explained, is match making or the “dating agency” effect. The outcome variables of interest are the number of buyers in the network and the number of buyers outside the network. Across all indicators of firm types, the results are positive and significant. This suggests that the impact on buyers within the superstar’s network is strong, although the precise reason for this impact is unclear. It could be that there is a signaling effect when a firm starts selling to a superstar firm, indicating that the supplying firm is of high quality.
SUMMARY
Van Reenen concluded his presentation by stating that firms that enter the supply chain of a superstar firm see positive, nontrivial effects across a range of outcome variables. The main proposed mechanism is the transfer of technology and know-how; match making also plays a role. The authors did not rule out more general spillovers, as this variation is removed with the inclusion of industry by year fixed effects. A main finding is that the firm does not need to be an MNE or globally engaged; the data show that local superstar firms have a similar impact on suppliers. The policy implications of the results are that barriers to firm growth due to misallocation could be costly (see Aghion et al., 2021), supporting government policies that promote the location of MNEs in their country. The next steps in the project include using instrumental variable analysis to understand superstar partnerships and further quantify the results.
DISCUSSION
During the keynote presentation, an audience member asked if there is selection into the MNE/superstar supply chain. In other words, are the results endogenous based on the firm from which the MNE/superstar chooses to purchase? Van Reenen responded that the match-making mechanism is this exact effect. Entering into a relationship with a superstar is beneficial because a firm then gains better access to their network. This is true across all types of firms, and it may be beneficial to investigate whether supplying to MNEs would provide access to foreign networks, in addition to the domestic network. Mary Amati, economist and vice president of the Microeconomic Studies Function at the
Federal Reserve Bank of New York, an associate of the Centre for Economic Research, and a coauthor of the presented work, answered the question by mentioning the breakdown of large nonglobal and large global firms, suggesting that, according to the results, there is no difference between the two.
This page intentionally left blank.