Technology transfers, foreign investment and productivity spillovers: Evidence from Vietnam John Rand University of Copenhagen Presentation based on work done in collaboration with Carol Newman, Theo Talbot and Finn Tarp

Motivation Attracting FDI is a policy priority in many developing countries, including Vietnam. Aside from providing jobs and capital, FDI firms also bring new technology and knowledge. Argument is that FDI firms are likely to be technologically superior to domestic firms. Through their interactions, knowledge/new technology can be transferred to domestic sector leading to productivity improvements. This can happen through many different mechanisms but disentangling these empirically have been challenging. While the topic has received a lot of attention in the literature there is conflicting empirical evidence on the nature of spillovers and limited evidence on the underlying mechanisms.

Interactions – What do we mean? Recent newspaper example Taiwanese company Binh Duong chosen, because other Taiwanese companies located here + good business environment. Produce lighting products for exports to the EU and US. Machinery and intermediate inputs imported from China. Interactions??? Vertical spillovers??? 3

What we do …. Use a unique data source to analyze various mechanisms for spillovers from foreign- invested firms to the domestic enterprises in Vietnam We examine whether horizontal, forward and backward spillovers exist in this context We disentangle contractual technology transfers from externalities associated with FDI using a measure gathered from specially designed survey data We consider the extent to which competition effects dominate positive externalities from FDI. We compare spillovers from joint-venture FDI firms and wholly-foreign owned firms.

Conceptual framework (1) Horizontal or intra-sector spillovers (Caves, 1996): FDI firm has firm-specific asset with a public good characteristic (e.g. knowledge or superior technology) Cannot prevent it from being transferred to competing firms E.g. through worker mobility, business or other networks, etc. Vertical or inter-sector spillovers (Rodriguez-Clare 1996): Through the supply chain Backward: from foreign firms to domestic input suppliers by increasing demand for specialized inputs. Forward: from foreign intermediate input suppliers to domestic producers by increasing the production of more complex inputs. To illustrate…..

Conceptual framework (2)

Conceptual framework (3) Backward spillovers: Positive: Deliberate knowledge transfer e.g. technical assistance, management experience, quality assurance (Moran 2001). Incentives for suppliers to improve quality of inputs (Javorcik 2004). Scale economies. Negative: Asymmetric bargaining power (Girma et al. 2008). Domestic firms not suited to producing input varieties demanded by foreign firms (Rodriguez-Clare 1996). Increased competition from other foreign firms supplying inputs (Aitken and Harrison 1999) or from imported inputs.

Conceptual framework (4) Forward spillovers: Forward spillovers have been very little attention in the literature. Positive: Embodied technologies (Girma et al 2008) Accompanying services (Javorcik 2004) Competition effects Negative: ‘Lock-in’ to using inputs purchased from FDI firms Asymmetric bargaining power possible if FDI firms gain dominant position upstream Cultural factors

Empirical Evidence Horizontal spillovers: Very little empirical evidence that they exist Foreign-invested firms compete with domestic firms in the same sector – incentive to prevent their technology from leaking (Javorcik 2004) Barrios et al. (2011), Blalock and Gertler (2008), Bwalya (2006), Damijan et al. (2008), Javorcik (2004) and Kugler (2006) - none find evidence for horizontal spillovers Backward spillovers: Javorcik (2004)- Lithuania Blalock and Gertler (2008) – Indonesia Kugler (2006) - Columbia Forward spillovers: No evidence (as far as we know)

Related issues Characteristics of foreign and domestic firms may matter: Javorcik (2004) – backward spillovers only evident from partially-owned foreign firms. Giroud et al (2012), Marin and Bell (2006) – spillovers more likely from firms that are technologically/knowledge intensive. Crespo and Fontoura (2007) – absorptive capacity of domestic firms matters Blomstrom and Sjoholm (1999) – export status of firm Aitken and Harrison (1999) – firm size Marin and Bell (2006) – investments in technology and training Distinction between externalities and actual technology transfers: Giroud et al. (2012) and Zanfei (2012) critique literature on this point Smeets (2008) – technology transfers and spillovers are distinct concepts that should be considered as such in empirical analysis This is one of our key points of departure…..

Measurement of spillovers (Javorcik, 2004) Horizontal spillovers: the proportion of total revenue, R, within each 4-digit sector, j, accounted for by k foreign-owned firms (firms denoted with subscript i and time with t ). Common Empirical Approach (1)

Forward spillovers: the proportion of total revenue in upstream sectors accounted for by foreign-owned firms  ut is the proportion of inputs into sector j that are purchased from sector u in time t and H ut is the proportion of foreign-owned firms in upstream sector u. Common Empirical Approach (2)

Backward spillovers: the proportion of total revenue in downstream sectors accounted for by foreign-owned firms  dt is the proportion of output from sector j that is sold to sector d in time t and H dt is the proportion of foreign-owned firms in downstream sector d. Common Empirical Approach (3)

Baseline model (Javorcik, 2004): detecting spillovers Y : value added L : total labor input K : capital inputs  i : firm fixed effects s j : 4-digit sector fixed effects  t : time fixed effects  How is productivity of firm related with foreign dominance within sectors (H), in upstream sectors (F) and in downstream sectors (B)? Common Empirical Approach (4)

Detecting technology transfers: tech_back : firm received a technology transfer from a downstream firm tech_for : firm received a technology transfer from an upstream firm  B : backward FDI spillovers due to direct technology transfers  F : forward FDI spillovers due to direct technology transfers  B : backward FDI spillovers due to externalities  F : forward FDI spillovers due to externalities Two Marginal Effects of interest: Our Empirical Approach (1)

Our Empirical Approach (2) OLS estimation biased. Standard “endogenuos” OP approach using Wooldridge’s (2009) one step GMM estimator. Allows us to simultaneously address the problem of measurement error in the capital input which will place further downward bias on the estimate of capital. Identification challenge: many potential confounding factors that impact on the change in the amount of FDI into a sector and the change in the productivity of the firm. We try to address most of the concerns. 16

Data Technology and Competitiveness Survey (TCS) 2009 and onwards Sample of more than 7,500 firms Vietnamese Enterprise Survey (various years) Population of all registered enterprises in Vietnam with 30 employees or more and representative sample of smaller firms TCS implemented by GSO as part of Vietnam Enterprise Survey and so data can be combined. Supply Use Tables (SUT) for Vietnam in 2007 to measure proportion of inputs/outputs traded between sectors. Export and import data at 4-digit level taken from COMTRADE – control variables. 17

Summary Statistics (1) 18 MeanSt. Dev. Tech Transfers: Tech Transfer Forwards Tech Transfer Backwards Absorptive capacity: New Machinery New ICT Process Innovation Quality Innovation Expand Variety Expand Product Switch Sector Tech Adaptation R&D

Summary Statistics (2) 19 MeanSt. Dev. FDI Spillovers: Horizontal Forwards Forwards 100% Forwards JV Backwards Backwards 100% Backwards JV Note: Time-varying sector level controls also included: concentration, imports and exports.

Positive forward spillovers - Negative backward spillovers - No horizontal spillovers Consistent across all models There is a clear distinction between externalities and direct technology transfers even after controlling for technology transfers a large part of FDI spillovers remains unexplained. But we find that: Forward spillovers: OLS: JVs create productivity externalities that filter along the supply chain. Wholly foreign-owned projects only enhance the productivity of domestic customers where there is a contractual obligation to transfer knowledge. IV: Cannot conclude anything about whether upstream spillovers come from JVs or 100% foreign owned FDIs. Forward linkages generally positive through the externality effect. Results (1)

Results (2) Backward spillovers: Negative spillovers are found and are due to wholly foreign- owned firms. Part of this is explained by negative competition (crowding out) effects. Controlling for sector concentration we also find negative spillovers from JVs in competitive sectors. Raises a key question: Are there any knowledge spillover benefits to domestic Vietnamese firms from being directly linked with FDI firms? 21

Are there any direct benefits from engaging with FDIs? No evidence that direct supply chain linkages are productivity enhancing. No evidence that direct technology transfers are productivity enhancing. Some evidence of negative productivity effects associated with having FDI customers once interaction with tech transfers is controlled for. 22

Conclusion (1) Contrary to other empirical studies we find for Vietnam that forward linkages lead to positive productivity spillovers, but the main part of these are unexplained (through externalities). Backward linkages negatively impact the productivity of domestic firms, and increased competition from imports explains most of the negative backward spillover from downstream FDI firms. Absorptive capacity can maybe cushion firms from negative backward spillovers Additional supportive evidence.

Zooming in on the Direct Tech Transfers Purpose sampling using a methodological triangulation approach 7 countries including Vietnam Data collected based on an identical semi-structured interview guide. The sample of firms were selected as follows (purpose and sequential/snowball sampling): Semi-structured interviews with country IPAs. Should lead to the identification of the 15 most influential MNCs/FDIs with majority foreign ownership. FDIs/MNCs that produce intermediates for the domestic market (if present) should have high priority. Semi-structured interviews with identified MNCs/FDIs. The interview should lead to the following identification : (i) three domestically owned industrial firms which are customers of the MNC/FDI. (ii) three domestically owned industrial firms which are suppliers to the MNC/FDI. (iii) three in-country direct competitors to the MNC/FDI. The interviews above could lead to the identification of relevant (i) competitors, (ii) domestically owned industrial suppliers of FDIs/MNCs and (iii) domestically owned industrial customers of FDIs/MNCs. Semi-structured interviews carried out.

Identifying Links: Vietnam 25

Identifying Links: Kenya 26

Conclusion (2) Preconditions for vertical spillovers (within country) are relatively weak in the African countries considered due to less developed customer/supplier chains and networks (Economic Complexity). However, for a given business network structure direct spillovers (especially forward linkages) are more likely to occur in the African sample. 27