North-South FDI and Bilateral Investment Treaties (BITs) Neil Foster-McGregor

Transcription

1 North-South FDI and Bilateral Investment Treaties (BITs) Neil Foster-McGregor

2 Introduction Around 3,000 BITs currently in force globally Aim is to encourage FDI inflows from developed to developing countries by providing certain guarantees to developed country investors Expected benefits: For developing countries; increased capital inflows, technology, competition, knowledge spillovers,... For developed country firms; MFN treatment, equitable treatment for foreign investors, free transfer/repatriation of capital/profits, dispute settlement Costs to developing country relate to reduced sovereignty (e.g. dispute mechanisms) Expected benefits to developing countries rely on increased FDI flows in response to BIT formation An empirical issue

3 Introduction Empirical evidence is limited (and varied) Existing studies consider both bilateral and total FDI flows(stocks) Do BITs impact on bilateral FDI only or, by providing a signal, on inward FDI more generally? Studies of bilateral FDI tend to use a variant of the gravity equation Wide range of estimated effects in the literature (Bellack, 2013) Majority are positive and significant Numerous estimated semi-elasticities >100% or <-50% Little attention to paid endogeneity (self-selection) Bergstrand and Egger (2013) find that economic fundamentals that determine FDI flows also determine BIT membership Are BITs substitutes or complements for institutional quality? Hallward-Driemier (2003) find evidence of complementary relationship, while Neumayer and Spess (2005) find evidence of substitutability

4 This Paper Estimates the causal impact of BIT formation on bilateral FDI flows (and stocks) Uses matched difference-in-difference methods to estimate causal effects Considers FDI flows (and stocks) over the period from 20 OECD countries to a sample of up to 119 other countries Allow for effects along the intensive and extensive margins Considers heterogeneous effects of BIT formation on FDI flows (stocks)

5 Method

6 Method

7 Method

8 FDI Data Intensive and Extensive Margins

9 Probit Results Coefficients (1) (2) (3) (4) (5) (6) 0.811*** 1.035*** 1.427*** 3.257*** 4.610*** 5.604*** (0.115) (0.132) (0.158) (0.192) (0.505) (0.481) 0.233*** 0.248*** 0.245*** 3.392*** 1.690*** 1.638*** (0.0224) (0.0235) (0.0241) (0.0448) (0.101) (0.0989) *** *** *** *** *** *** (0.0345) (0.0368) (0.0385) (0.0873) (0.242) (0.179) *** *** ** *** *** (0.320) (0.325) (0.309) (1.081) (1.152) (0.722) *** 2.279*** *** (0.105) (0.109) (0.116) (0.190) (0.330) (0.436) 0.602*** 0.647*** 0.469*** *** 11.44*** 5.894*** (0.142) (0.149) (0.158) (0.267) (0.544) (0.515) *** *** *** 3.139*** *** *** (0.107) (0.125) (0.150) (0.187) (0.439) (0.436) 0.545** 0.946*** 1.473*** * 4.305*** (0.268) (0.295) (0.323) (0.514) (1.269) (1.273) 0.251** 0.275** 0.300*** 1.214*** 0.911*** 0.910*** (0.103) (0.109) (0.110) (0.128) (0.312) (0.308) 0.376*** 0.203*** 0.175** 3.614*** 1.912*** 1.805*** (0.0655) (0.0678) (0.0736) (0.172) (0.271) (0.286) Year FE No Yes Yes Yes Yes Yes MR Controls No No Yes No No Yes Random-effects Probit No No No Yes Yes Yes Percent correctly predicted

10 Standard Gravity Estimates of BITs (1) (2) (3) (4) 1.087*** 0.886*** 0.776*** 0.762*** (0.124) (0.123) (0.116) (0.115) 0.265** 0.303*** 0.337*** 0.313*** (0.103) (0.103) (0.0908) (0.0890) 0.363*** 0.308*** * * (0.0357) (0.0357) (0.0490) (0.0486) Year FE Yes Yes Yes Yes MR Controls No Yes No Yes Country-Pair FE No No Yes Yes

11 Matching Estimates ATT ATE ATT ATE ATT ATE Mahalanobis Distance Metric One to One Matching (0.2224) (0.1236) (0.2202) (0.1791) (0.0226) * (0.0180) Five Nearest Neighbour Matching *** (0.1783) *** (0.0896) (0.1673) (0.1631) * (0.0176) *** (0.0121) Propensity Score Matching One to One Matching (0.2227) Five Nearest Neighbour Matching *** (0.1758) Radius Matching (radius = 0.01) *** (0.1617) Kernel Matching *** (0.1623) * (0.2039) ** (0.2141) *** (0.1022) *** (0.1610) (0.1940) (0.1581) * (0.1524) (0.1534) ** (0.1361) ** (0.1486) (0.1522) * (0.1188) ** (0.0224) *** (0.0175) *** (0.0158) *** (0.0158) * (0.0160) ** (0.0212) *** (0.0178) *** (0.0241) BIT formation is estimated to increase FDI flows by between 33 and 170% (mean 80%) Little evidence of a significant impact of BIT formation on the intensive margin BIT formation is estimated to increase the probability of a new FDI flow (extensive margin) by between 2.34 and 10.53%

12 Heterogeneous Effects (Selected Estimates) NN 5NN NN 5NN NN 5NN Main effect of new BIT (0.173) (0.177) (0.284) (0.286) (0.0171) (0.0174) Interaction terms with: ** ** ** *** (0.180) (0.192) (0.182) (0.183) (0.0178) (0.0189) 0.358*** 0.552*** *** *** (0.102) (0.104) (0.122) (0.115) (0.0101) (0.0102) 3.982*** 3.405*** 3.200*** 5.657*** (1.264) (1.285) (0.828) (0.932) (0.126) (0.126) 1.125** 1.366** ** ** (0.547) (0.612) (0.754) (0.631) (0.0543) (0.0600) 1.313* 1.538** *** 0.308*** (0.692) (0.711) (0.733) (0.705) (0.0687) (0.0698) * *** *** (0.452) (0.467) (0.422) (0.416) (0.0449) (0.0459)

13 Conclusion BIT formation impacts positively upon bilateral FDI flows Using the matched DID approach tends to result in smaller estimates The impact of BITs works primarily on the formation of new FDI relationships rather than the expansion of existing ones i.e. The extensive rather than the intensive margin Impact of BITs is heterogeneous Being smaller for more distant partners and partners with an existing PTA Being bigger for more partners sharing a common language and border and having more similar income level Increased institutional quality is complementary to BIT formation, an effect tht works along the extensive margin

14 Descriptive Statistics Variable Observations Mean Std. Dev. Min Max 29, , , , , , , , Variable Observations Frequency Percentage (%) 29, , , , , , ,