What Is Real and What Is Not in the Global FDI Network? Jannick - PowerPoint PPT Presentation

What Is Real and What Is Not in the Global FDI Network? Jannick Damgaard (Danmarks Nationalbank) Thomas Elkjær (International Monetary Fund) Niels Johannesen (University of Copenhagen) December 2, 2019 The views are expressed here are those of the authors; they do not necessarily reflect the views of the institutions with which they are affiliated

Introduction FDI is important dimension of economic integration - enters common indexes of globalization - widely believed to boost productivity and growth However, FDI is not only about productive investment Example: Luxembourg has inward FDI of $4 trillion same as the U.S. or the five major EU economies combined What is going on? Many tax avoidance strategies involve large nominal investments in offshore financial centers

Offshore financial centers play an outsized role in the global FDI network 5000 4000 Inward FDI ($ billions) 3000 2000 1000 0 Netherlands United States Luxembourg United Kingdom China Hong Kong SAR Switzerland Singapore Germany Ireland British Virgin Islands France Bermuda Cayman Islands Australia Canada Brazil Spain Belgium Mexico

This paper Define ”Phantom FDI” as investment into empty shells almost no employees and no substantial links to the local economy Our goal : Estimate global FDI network while separating ”Phantom FDI” and ”Real FDI” Data sources : - Coordinated Direct Investment Survey (CDIS) standard FDI statistics - new OECD statistics on FDI into Special Purpose Entities (”SPEs”) - new OECD statistics on FDI by economy of ultimate investors - ORBIS: global database with firm-level data

A stylized corporate group structure U.S. COMPANY $100 investment LUX COMPANY $100 investment GERMAN COMPANY

Standard FDI statistics confound ”Real” and ”Phantom” FDI Standard FDI Statistics: U.S. $100 FDI in Lux from US COMPANY $100 FDI in Germany from Lux  $200 FDI globally  No US-Germany FDI link $100 investment LUX COMPANY $100 investment GERMAN COMPANY

Our goal is to measure ”Phantom” FDI and real investment links Standard FDI Statistics: U.S. $100 FDI in Lux from US COMPANY $100 FDI in Germany from Lux  $200 FDI globally  No US-Germany FDI link $100 investment Our New FDI Database: LUX $100 ”Real FDI” in Germany from US COMPANY $100 ”Phantom FDI” in Lux  $100 ”Real FDI” globally $100  $100 ”Phantom FDI” globally investment GERMAN COMPANY

Our database also yields measures of exposure to tax avoidance opportunities Standard FDI Statistics: U.S. $100 FDI in Lux from US COMPANY $100 FDI in Germany from Lux  $200 FDI globally  No US-Germany FDI link $100 investment Our New FDI Database: LUX $100 ”Real FDI” in Germany from US COMPANY $100 ”Phantom FDI” in Lux  $100 ”Real FDI” globally $100  $100 ”Phantom FDI” globally investment Our tax avoidance indicators GERMAN $100 US exposure from outward FDI COMPANY $100 German exposure from inward FDI

”Phantom” FDI is around $15 trillion (around 40% of Total FDI) 50000 0.50 Real FDI in $ billion (left axis) Phantom FDI in $ billion (left axis) Share of Phantom FDI in Total (right axis) 40000 0.40 30000 0.30 20000 0.20 10000 0.10 0 0.00 2009 2010 2011 2012 2013 2014 2015 2016 2017

Method

STEP #A: complete network of standard FDI data Use inward FDI reported by 122 countries to Coordinated Direct Investment Survey (CDIS) Use mirror outward FDI when inward FDI is not available in CDIS - Cayman Islands does not report inward FDI from Canada - Canada reports outward FDI to Cayman Islands for $34 billion - We set inward FDI in Cayman Islands from Canada at $34 billion

STEP #B: split Total FDI into ”Phantom” FDI and ”Real” FDI How can we estimate the split between Phantom and Real FDI for economies that do not report this? Intuitive idea: When FDI is large relative to GDP, only a small share of it is is Real FDI (1) Estimate structural relation between - ”FDI intensity” (FDI/GDP) - ”Real FDI share” (Real FDI/Total FDI) for reporting economies (2) Extrapolate to non-reporting economies

Very strong structural relationship between FDI/GDP and Real/Total FDI 1 Real FDI / Total FDI (in logs) 0 slope: -.504 robust s.e.: .036 r-squared: .845 period: 2013-2017 Netherlands -1 observations: 75 Hungary -2 -3 Luxembourg -2 0 2 4 Total FDI / GDP (in logs)

We extrapolate this structural relationship to all other economies 1 Predicted Reported Sweden Belgium Switzerland Portugal .8 Real FDI / Total FDI (Predicted) Denmark Iceland Austria .6 Hungary .4 Netherlands Cyprus Bahamas .2 Jersey Barbados Malta Mauritius Falkland Islands Luxembourg Gibraltar Bermuda Cayman Islands 0 0 50 100 150 200 Total FDI / GDP (Actual)

STEP #C: allocate ”Real” FDI to ultimate owner economies Use new OECD FDI data when available: some countries themselves allocate FDI to ultimate investor economies Estimate from ORBIS firm-level data when not reported: - standard stat : FDI in Spain from immediate owners in US: $23 billion - Orbis: FDI in Spain from immediate owners in US: $13 billion - Orbis: FDI in Spain from ultimate owners in US: $29 billion - our estimate : FDI in Spain from ultimate owners in US: $50 billion (23 billion × 29/13 = 50 billion)

Findings

Some uncertainty but the order of magnitude is right 20000 Baseline estimate Alternative prediction model Only reported Phantom FDI Offshore centers only 15000 Phantom FDI ($ billions) 10000 5000 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year

Major recipients of ”Phantom” FDI are offshore financial centers 4000 Self-reported Estimated 3000 Phantom FDI ($ billions) 2000 1000 0 Luxembourg Netherlands Hong Kong SAR British Virgin Islands Bermuda Singapore Cayman Islands Switzerland Ireland United Kingdom Mauritius Cyprus Malta Hungary Curacao Barbados Bahamas Jersey Gibraltar Puerto Rico

Real factors have 25% more power explaining Real FDI links Estimated Real FDI link 600 Standard FDI link Real FDI ($ billions) 400 200 0 USA-GBR USA-JPN GBR-USA USA-DEU CHN-JPN DEU-USA CHN-USA DEU-GBR CHN-DEU GBR-JPN JPN-USA USA-CHN GBR-DEU DEU-JPN CHN-GBR GBR-CHN JPN-DEU JPN-GBR DEU-CHN JPN-CHN

”Round-tripping” modest in Europe but large in China and Russia 0.30 Self-reported Estimated Round-tripping factor 0.20 0.10 0.00 a a a a y y n e m y s s a d a l a o a a i n i d i n l i c e e d i n e z n c i i s s a a d l b o a a i a a n k t n a r i i h s e t p d a n t x a m I a r a o r n r u n n l B e C S g u t I r K t r a r S l e e s A R o r T r M F n C e e z g u d i h d A i G K h t r d n e i t A t w u I u d t e i S o a N e n S S t U i n U

Exposure to tax avoidance opportunities through outward FDI into SPEs 0.80 Share of outward FDI to SPEs 0.60 0.40 0.20 0.00 Brazil Indonesia United States China Russia India Turkey United Kingdom Canada Taiwan Province of China Saudi Arabia Germany Italy Australia Mexico Japan France South Korea Spain Argentina

Exposure to tax avoidance opportunities through inward FDI from SPEs 0.80 Share of inward FDI from SPEs 0.60 0.40 0.20 0.00 Russia China Indonesia India Germany Italy United Kingdom France Spain Brazil Japan South Korea United States Turkey Canada Taiwan Province of China Argentina Australia Mexico Saudi Arabia

Conclusion

Conclusion ”Phantom” FDI and ”Real” FDI are confounded in standard FDI statistics: - cannot assess the strength of real economic ties - cannot gauge tax-motivated FDI into empty shells We estimate that: - ”Phantom” FDI accounts for 40% of Total FDI globally - vast majority of ”Phantom” FDI is into OFCs - suggestive of tax avoidance