FDI technology spillovers, geography, and spatial diffusion

Abstract

This paper investigates the geographic extent of FDI technology spillovers and associated spatial diffusion. By adopting a spatiotemporal autoregressive panel model as the platform of our study, the complex impact resulting from FDI penetration is separated into spatial direct and indirect effect while accounting for feedback loops among regions. A set of spatially partitioned summary measures is produced to identify and to quantify FDI spillovers from different channels with distinct geographic scopes. Empirical results based on data from China document that the direct impacts of FDI presence to a specific location itself are likely to be negative. Domestic firms mainly benefit from FDI presence in their neighboring regions through knowledge spillovers that have wider geographic scope. Negative market stealing effect nevertheless has no spatial boundary. Policy implications of these findings are discussed.

Spatial dependence, foreign investment and productivity spillovers in new EU member states

“…Hence, a change of FDI spillovers in region i affects TFP jt in region j, which in turn affects TFP it in region i via the spatial autoregressive term. These feedback loops arise because region i is considered as a neighbour to its neighbours, so the impact passing through neighbouring regions will create a feedback impact on region i itself (Lin and Kwan, 2016). For this reason, the results in Table 2 and subsequent tables are presented as average direct and indirect effects.…”

“…Because of lower absorptive capacity, it is more likely that horizontal spillovers will be negative due to pecuniary externalities in the form of market stealing. As MNCs operate nationwide, this may explain increasing negative competition effects in wider geographical areas due to highly integrated markets (Lin and Kwan, 2016). The extent of negative horizontal spillovers increases when the range of neighbouring regions expands as the coefficient of interregional spillovers decreases from -0.10 to -0.21 when spatial matrix is not truncated.…”

“…Negative market stealing effects could easily spread to the entire country in highly integrated markets characterised by diverse distribution channels. However, given the limited labour mobility the net effect of horizontal knowledge spillovers may still be limited to narrow geographical scope (Lin and Kwan, 2016). In addition, MNCs may prefer suppliers located in wider geographical areas to benefit from high quality inputs while on the other hand they may develop local linkages to minimize transaction costs and facilitate communication with a domestic supplier or distributor.…”

“…For this reason, the regional average values of above introduced measures enter the model. Such approach, previously applied by Lin and Kwan (2016) and Nguyen et al (2016) inevitably results in a loss of industry-level information but it enables us to properly model the nature of inter-regional spillover effects including spatial feedback loops. In addition, spatial econometric analyses require data to have a balanced structure over time.…”

“…Merlevede and Purice (2016) suggest that distance may have a different effect on FDI spillovers through vertical and horizontal channels. Apart from spatial concentration, there may be important spatial interactions shaped by feedback loops (i.e., from one location to its neighbours, and then neighbours' neighbours, and finally back to the original location indirectly) among indigenous firms and between indigenous firms and MNCs (Lin and Kwan, 2016). This further complicates the identification and interpretation of results.…”