140 likes | 307 Views
The effect of human capital on FDI: A meta-regression analysis. Artane Rizvanolli, AAB- Riinvest University Ancona , 21 May 2010. Contents. Introduction: FDI and growth Rationale for MRA Sample MRA model Empirical results Conclusion and further research. Introduction: FDI and growth.
E N D
The effect of human capital on FDI: A meta-regression analysis Artane Rizvanolli, AAB-Riinvest University Ancona, 21 May 2010
Contents • Introduction: FDI and growth • Rationale for MRA • Sample • MRA model • Empirical results • Conclusion and further research
Introduction: FDI and growth • FDI conventionally considered beneficial • Technology and know-how transfer (?) • Spillovers (?) • Hence, overall productivity and growth (?) • Especially important for transition economies • Need for restructuring and modernisation (at firm and economy level) • Limited domestic resources • However, are the benefits automatic?
The rationale for meta-regression analysis (MRA) • Theory: human capital (HC) attracts FDI • Enhancement of productivity, technology adoption and adaption • No consensus in the empirical literature • Negative, positive and insignificant results found • Potential reasons for the diversity of results? • Wide range of specifications, HC measures, countries • Lack of “universal” relationship between HC and FDI: differences in motivation for FDI, sector of economic activity, etc.
The rationale for meta-regression analysis (2) MRA as a means of • Quantifying a survey of empirical literature • Analysing the sensitivity of results to different study characteristics (!) • Identifying and quantifying the “genuine” effect of HC, if present • Identifying publication bias (?) • Informing the specification of further research on the HC-FDI relationship: which measures?
Sample • Around 30 regression analyses identified • EconLit, SSRN, Google Scholar • References in papers • Some excluded • Measures not convincing/comparable • No results reported • Only interaction/squared terms • Preferred regressions only (?)
Sample (2) • 28 studies with a total of 231 regressions • t-stats range -7.8 - 7.7, with a mean of 0.93 • Structure: • Developing, transition, mixed, China, developed • Mostly secondary and tertiary education measures • Majority(static and dynamic) panels
Model • Linear regression: weighted to give each study the same weight, clustered robust (cluster: study), dependent variables divided by SEpcc • Dependent variable: t-statistic of HC variable
Preliminary results • Bi-variate MRA • no publication bias OR “genuine effect” • Multi-variate MRA • Same result as above • Full model mis-specified • Ramsey RESET test : F(3, 205) = 94.52 , Prob > F = 0.0000 • Suffers from multicollinearity
Preliminary results (2) • Testing down: standard procedure in MRA • Improves functional form • Significantly reduces multicollinearity • Some variables highly correlated with INVSEPCC (PERIOD, MEDIANYR, LABCOST, TNOEXPVAR?, EDNOGENITY?, HCSTOCK?)
Conclusion and further research • Heterogeneity in HC-FDI literature can be explained to a very limited extent (!) • Appears to be no genuine effect in the literature: • Models not specified correctly? • Further research: specify model in accordance with theory • human capital variable: level and stock/flow
Thank you! Questions & Comments?