Knowledge Transfer from Multinationals through Labour Mobility: Are There Effects on Productivity, Product Sophistication and Exporting?

ABSTRACT

We investigate whether labour mobility from foreign-owned firms to local firms in the host economy is associated with an increase in productivity, export product complexity and other export indicators of domestic firms. Based on employer-employee level data from Estonia, we confirm that hiring employees with experience from foreign-owned firms is associated with an increase in the total factor productivity (TFP) of the firm and the higher export propensity and breadth of export markets or products. However, on average, these within-firm effects on TFP appear to be not working through increases in the level of the complexity of the export product portfolio of domestically owned firms. One implication of this result is that other channels of upgrading than changes in the Hausmann-Hidalgo export product complexity measure may be more important in this context; for example, such as upgrading the tasks or activities in the global value chain of a product.

KEY WORDS:

• knowledge spillovers
• labour mobility
• multinational enterprise
• product complexity
• product sophistication

JEL CLASSIFICATION:

• F10
• F23
• J62

Acknowledgments

We are grateful to the Statistics Estonia for granting access to the Estonian individual and firm-level datasets and note that all calculations have been made following their confidentiality requirements. The authors are solely responsible for all errors and omissions.

Notes

^1. This idea is presented in these highly cited papers, but the idea itself is of course older than that. For example, related conclusions have been reached concerning the important role of economic structure on growth in Stokey (1988), Young (1991), Grossman and Helpman (1991), among others.

^2. A recent exception is a paper by Varblane and Bormann (2018) based on firm level data from Estonia that investigates the firm level correlation between export product complexity of a firm’s export portfolio and firm’s labour and total factor productivity. Their results are specifically relevant to ours here, as they demonstrate the positive correlation between product complexity of a firm and its productivity, as long as they rely on the cross-sectional variation in the data. The correlation disappears as soon as the firm fixed effects are introduced in their analysis.

^3. The Great Recession hit Estonia heavily with GDP falling about 14% in 2009. Much of the adjustment occurred in the labour market through external flexibility, i.e. reductions in the number of employees, though there were also widespread wage cuts even in the base wages. Already in 2010 a strong recovery was observed (Masso and Krillo 2011).

^4. Typically, the transfer of this knowledge is based on employees’ exposure to training and new knowledge or technology at the foreign owned firm, and then the employee leaving this firm for a local one.

^5. Competition effect is in fact not a knowledge externality effect, therefore should be treated conceptually separately from knowledge externalities such as labour mobility and demonstration effects (see e.g. Bhaumik et al. 2018 for a discussion on this). In empirical analysis, however, it has been often difficult to distinguish between the various channels of the effects of FDI on local firms.

^6. Apart from MNEs, the knowledge transfer effects of labour mobility have been studied in a number of other contexts. For example, this includes the effects of the mobility of R&D workers and researchers (Jaffe, Trajtenberg, and Henderson 1993; Kaiser et al. 2018; Maliranta, Mohnen, and Rouvinen 2009), foreign specialists and migrants (Hiller 2013; Markusen, And, and Trofimenko 2009), expatriates with experience and networks from working abroad (e.g. based on data from China in Filatotchev et al. 2009; Liu et al. 2010).

^7. The key export markets are the neighbouring countries Sweden, Finland and Latvia.

^8. Boone (2008) introduced his competition measure based on the idea that in a more competitive market, firms are punished more harshly for being inefficient. Also, rewards for high efficiency are larger in a more competitive market. There is a more rapid re-allocation of market share and profits to more efficient producers in a more competitive environment. Cost advantages lead to higher profits. The Boone indicator that we use here is an elasticity, a percentage increase in profits due to a 1 percent increase in efficiency, estimated at the 3-digit sector level from a regression of the log of gross profits at firm level on the log of average variable costs and firm fixed effects.

^9. We have further applied a similar IV to the one used in Javorcik, Lo Turco, and Maggioni (2017) in the estimation of the effects of FDI on product complexity. Javorcik, Lo Turco, and Maggioni (2017) used a measure based on FDI presence in Poland in 2-digit NACE industries as an instrumental variable for a measure of Horizontal FDI spillovers in Turkey. We have tried a similar FDI share variable (and its change over time) from Latvia as a IV for labour mobility from MNEs to domestic firms in Estonia’s sectors of manufacturing. Unfortunately, this proved to be a weak instrument.

^10. The rather high average share of managers in the estimation sample (as in Table 1) reflects the large share of micro firms. Note that in each firm at least one employee was defined as a manager.

^11. These results in Annex 1 point to 5–9 per cent higher TFP among firms that have MNE-experienced employees (these figures are calculated in a standard way as exp(β)-1, where β is the estimated parameter, i.e. this interval 5–9 is calculated based on [exp(0.049)-1]; [exp(0.086)-1].

^12. The control variables mostly show the expected results. Exporters among domestically owned firms have significantly higher productivity. The higher cash to assets ratio and share of high-wage employees among the workforce are correlated with the higher productivity of the firm. The share of high-wage employees is included here as an indirect proxy for skill intensity. It is a vital control in estimating the productivity equations. Without accounting for the general high share of high-wage employees at the firm, we could overestimate the gains from having managers and top specialists with MNE experience.

^13. Previous related studies by Balsvik (2011) and Poole (2013) include unit level (plant or individual level) fixed effects, with Balsvik (2011) also using the lagged share of newly hired MNE experienced employees. Balsvik (2011) additionally reports trying the GMM approach to account for the endogeneity of MNE experience. The system GMM estimator uses lags for inputs and a dependent variable as instruments. However, in her analysis the validity of these (internal) instruments was rejected, leaving potential endogeneity issues still in the estimated relationships. Therefore, the GMM results were not reported.

^14. The magnitude of the estimates of the effects is as follows: a ten-percentage point increase in the share of employees from MNEs is associated with 9.4 per cent higher TFP in the domestically owned firm (calculated based on parameter estimate 0.664 in column 1 of Table 3 as 0.1*[exp(0.664)-1]. The estimated effect is significantly larger in the case of the mobility of high-wage employees (the difference is statistically significant at the 1 per cent level). Here, a ten-percentage point increase in their share in the workforce of the domestic firm would increase the TFP of the recipient firm by 26 per cent (calculated based on parameter estimate 1.283 in column 2 of Table 3 as 0.1*[exp(1.283)-1].

^15. We note that other firm level variables are statistically significant in the OLS version of the export complexity models (columns 1 and 2 in Table 4), but not in the fixed effects specifications (columns 3 and 4). The control variables positively correlated in the OLS specifications (columns 1 and 2 in Table 4) with the export complexity are firm size, liquidity (cash to assets), share of managers at the firm (a rather broad proxy of skill intensity at the firm). In addition, most sector dummies at 2-digit NACE sector level are statistically significant in the models in columns 1 and 2 of Table 4, implying that there is a significant sector specific component driving export product complexity.

Additional information

Funding

The authors acknowledge financial support from the Estonian Research Council project IUT20-49 “Structural Change as the Factor of Productivity Growth in the Case of Catching up Economies”. Priit Vahter acknowledges financial support from Östersjöstiftelsen in Sweden (project “The Baltic economies: Catalysts for the internationalization of Swedish SMEs?”) and Jaan Masso from the Ernst Jaakson Memorial Foundation. The authors also acknowledge support for the compilation of the datasets used in the paper from the Estonian Research Infrastructures Roadmap project “Infotechnological Mobility Observatory (IMO)”.