Technology spillovers: Kenya and Malaysia compared

Abstract

This paper aimed at investigating the existence of productivity spillovers and their transmission channels in both Kenya and Malaysia firm-level panel data from the manufacturing sector for the period 2000–2005. Both countries have a long history of relying on FDI in industrial development. The existing literature on productivity spillovers suggests that productivity spillovers may be one of the most important effects that foreign MNEs impart to local firms in developing countries. Yet still, few studies exist in both countries on productivity spillovers and their transmission channels. Three spillover channels were examined: demonstration, competition, and information. In addition, the backward linkage channel was examined for the case of Malaysia. The results reveal that there is limited evidence of negative productivity spillovers from foreign firms to domestic firms through the competition effects in Kenya. In Malaysia, there is evidence of positive spillovers from foreign-owned firms to domestic firms through the demonstration effects. In addition, there is evidence of negative spillovers through the competition effects as well as backward linkages. There is also evidence of positive productivity spillovers from domestic firms to foreign-owned firms through backward linkages. Productivity spillovers are found to be dependent on the technology gap.

Keywords:

• foreign direct investment
• manufacturing sector
• multinational firms
• panel data
• productivity
• spillovers

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. See Lall (1995).

2. GOK (1994).

3. This definition was adopted because the Kenyan national authorities use the same benchmark. This definition follows that of OECD and UNCTAD.

4. This approach was followed by Gachino (2006), except that the export values were deflated using export price indices for manufactured goods. Bigsten, Kimuyu, and Lindvall (2004) used firm-specific deflators based on export share-weighted averages of the domestic and international prices to control for variations in the exchange rates.

5. There is no way of establishing the sampling frame used, and therefore the register may or may not be representative of firms. Besides, to the best of my knowledge, the number of firm establishments in Kenya is unknown.

6. L and K are each divided by their mean value. The translog production function can be regarded as a second-order approximation of arbitrary function at one point. When estimating the translog production function, it is common to use the approximation at point (L; K) = (1;1). Takii (2005) used the approximation at point (mean of L; and mean of K) because it appeared to yield better results.

7. However, there are cases where MNEs can conduct R&D at home or in third countries. Hence, it is included in the products (or processes) that the MNE brings into the host country. Hence, demonstration effects can also take place even if the MNE does not run the R&D activities in the host country. In fact, many R&D activities are concentrated in a few locations where the required talent/knowledge can be found (Castellani, Jiménez, and Zanfei 2013) and later MNEs transfer the outcomes/innovations to other subsidiaries around the world.