evidence of the impact of exchange rate regimes on fdi flows abbott, g. & de vita, g. scottish...
TRANSCRIPT
Evidence of the Impact of Exchange Rate Regimes on FDI Flows
Abbott, G. & De Vita, G. Scottish Economic Society Conference, Perth, 21st-23rd of April 2008
AIM: To investigate the impact of different ER regimes upon FDI flows using panel data from 27 OECD & non-OECD high-income countries for the period 1980-2003
We gratefully acknowledge financial support by the ESRC (grant RES-000-22-2350).
Context
• From 1980 to 2003, 450% increase in real, world FDI flows
• Much research on determinants of FDI and its growth enhancing effects but no attention paid to ER regimes and FDI flows
• Striking given the voluminous literature on ER regimes and trade (Rose, 2000 onwards)
• Schiavo’s (2007, OEP) gravity model investigates the impact of EMU on FDI flows. OLS & Tobit results show that EMU has increased FDI flows by 160 to 320% (caution: EMU data 1999-2001)
Contribution
• CUs are only one regime among feasible policy set. First set of estimates of effect of a wide menu of ER regimes on bilateral FDI flows between country-pairs (CU-CU; CU-FLT; CU-FIX; FIX-FIX; FIX-FLT; and FLT-FLT)
• Consider which ER regime the effect is benchmarked against. We compare the specific effect of each regime combination vis-à-vis the more plausible alternative of ‘double-float’
Contribution
• In terms of the categorisation of ER regimes, comparative use of three different classifications: Reinhart and Rogoff (2004); Shambaugh (2004) and IMF (ERAR reports, various issues)
• We explicitly control for simultaneity bias and reverse causality – instrumental variable estimation within SYS-GMM framework exploits time series variation while accounting for country specific effects
Our Sample (FDI flows and shares of World FDI flows and GDP)
Country
Mean total FDI
flows
Proportion of world
FDI flows
Proportion of world
GDP United States 6,304.03 22.15% 29.80% United Kingdom 4,598.98 8.01% 4.66% Belgium-Luxembourg 2,528.65 7.28% 0.81% Germany 2,519.92 5.36% 6.31% Netherlands 2,474.61 4.09% 1.16% France 2,212.81 5.41% 4.39% Japan 1,483.85 0.68% 16.05% Switzerland 1,083.64 1.38% 0.86% Canada 955.78 3.14% 3.32% Sweden 744.18 2.19% 0.79% Spain 738.59 3.37% 1.83% Italy 612.30 1.39% 3.68% Australia 553.49 1.76% 1.19% Ireland 454.41 1.54% 0.23% Denmark 359.71 1.12% 0.53% Finland 297.87 0.60% 0.39% Singapore 228.67 1.76% 0.22% Norway 213.35 0.47% 0.52% Hong Kong 162.86 1.75% 0.48% Austria 148.97 0.56% 0.63% Portugal 136.60 0.51% 0.33% Korea 117.71 0.57% 1.32% New Zealand 114.01 0.45% 0.17% Greece 40.94 0.21% 0.38% Israel 32.70 0.32% 0.32% Iceland 5.86 0.02% 0.03% United Arab Emirates 3.36 n/a 0.21%
Model
• Our unbalanced panel (27 OECD and non-OECD high-income countries over 1980-2003), yields over 7,000 country-year observations across almost 350 country-pairs
• Drawing from standard variables typically entering the gravity equation, our baseline model is expressed (in long-run form) as:
fdiijt = β0 + β1tbtijt + β2yit +β3yjt + β4RXRVOLijt + α5disij + α6LANGij
+ α7COLij + α8COMLANij + α9FTAijt + α10CU-CUijt + εijt
where fdi is the log of total bi-lateral real FDI flows between countries i
and j at period t. Sum of inward and outward FDI flows, calculated from
the OECD’s International Direct Investment Statistics database
Data
• Gravity type variables (dis*; LANG; COL; COMLAN) - see Centre d'Etudes Prospectives et d'Informations Internationales, http://www.cepii.fr/
* Based on bilateral distances between the biggest cities of the two countries, with intercity distances being weighted by the share of the city in the overall country’s population (Mayer & Zignago, 2006).
> proximity : >trade but as dis increases : > incentive for FDI
Melitz (2001;2005) showed that distance can reflect CA.
> distance might raise (diminish) not diminish (raise), trade (FDI)
• ER regime dummies (CU-CU; CU-FIX; CU-FLT; FIX-FIX; FIX-
FLT) calculated from classifications produced by Reinhart & Rogoff (2004), Shambaugh (2004), and IMF’s ARERAR
Table 1: Correspondence between original categories and those we derived from them
Reinhart and Rogoff’s classification
IMF’s classification
Shambaugh’s classification
Our classification
1 No separate legal tender Currency union Currency union Currency union
2 Pre announced peg or currency board 3 Pre announced horizontal band ±2% 4 De facto peg
Currency board/ Currency peg
within horizontal band of ±1%
5 Pre announced crawling peg 6 Pre announced crawling band ±2% 7 De facto crawling peg
Currency peg within crawling
band of ±1% 8 De facto crawling band ±2% 9 Pre announced crawling band that is wider
than or equal to ±2% 10 De facto crawling band that is narrower
than or equal to ±5% 11 De facto moving band that is narrower than
or equal to ±2%
Currency peg within crawling
band of at least±1%
De facto currency peg
Fixed ER
12 Managed floating 13 Freely floating
Managed floating/
Independently floating
Currency float Currency
float
14 Freely falling 15 Dual market / parallel market data missing
N/A N/A N/A
Why SYS-GMM estimation?
Appropriate both conceptually and for its statistical virtues:
• Lagged values of FDI can be included to account for speed of adjustment (important since FDI might adjust slowly to changes in the regressors)
• FDI or one or more of the regressors may be simultaneously determined (with SYS-GMM every regressor is instrumented so issues of endogeneity bias are overcome)
• Including both level and first-difference equations in a stacked system allows us to investigate whether time-invariant variables
(dis & COMLAN) play a role in the determination of FDI
Table 3: Full menu of ER regime dummies vs. ‘double float’
Variable R+R
IMF
Shambaugh
tbtijt 0.061a
0.063a
0.175a
yit 0.196a
0.169a
0.183a
yjt 0.156a
0.259a
0.230a
FTAijt -0.060
-0.016
0.010
disij -0.039
0.016
0.013
COMLANij 0.122
0.114
0.137
COLij 0.103
0.127
0.142
LANGij 0.153a
0.138a
0.133a
RXRVOLijt -0.120
0.589
0.666
CU-CUijt 0.285a
0.471a
0.452a
FIX-FIXijt -0.079
0.221a
0.136a
CU-FLTijt 0.194a
0.260a
0.234a
CU-FIXijt -0.052
0.051
0.049
FIX-FLTijt -0.041
0.019
0.053
a denotes significance at 5%
Extensions and perturbations • Re-estimated excluding USA & UK. Regime dummies proved
robust to new sample. Reliability corroborated by economically sensible changes to some control variables (e.g. the investing country’s output measure and LANG coefficient now insignificant; COMLAN now +ve and significant)
• Then: (i) replaced real per capita GDP with real GDP (used in Schiavo, 2007); (ii) added three regressors: 1. a long-run measure of exchange rate volatility; 2. a proxy for informational flows; 3. the real ER
• Presence of many zeros in bilateral FDI matrix constitutes an econometric issue since log-linear structure of gravity model precludes estimation of obs for which the natural log does not exist. In our SYS-GMM we converted the series to logs by adding a +ve constant. We now also re-estimate using Tobit
Table 8: Selected coefficients from extended model (using GDP) SYS-GMM Tobit
Variable R+R IMF Shambaugh
R+R IMF Shambaugh
yit 0.083a
0.077a
0.074a
0.044a
0.040a
0.043a
yjt 0.100a
0.096a
0.081a
0.056a
0.052a
0.053a
LANGij 0.125a
0.100a
0.095a
0.083a
0.077a
0.076a
RXRijt 0.190a
0.241a
0.037
-0.018
-0.011
-0.011
RXRVOLijt 0.137
0.783
0.988
-0.127
-0.114
-0.359
LRRXRVOLijt 1.297
2.285
0.791
0.266
0.109
0.290
(infi×infj)t 0.090a
0.117a
0.113a
0.040a
0.036a
0.032a
CU-CUijt 0.037
0.219a
0.250a
0.072a
0.116a
0.124a
FIX-FIXijt -0.156a
-0.021
0.053
-0.049a
-0.086a
-0.076a
CU-FLTijt 0.022
0.090
0.125a
0.052a
0.096a
0.076a
CU-FIXijt -0.126a
-0.021
0.023
-0.077a
-0.080a
-0.073a
FIX-FLTijt -0.107a
-0.087a
-0.0007
-0.044a
-0.016a
-0.023a
a denotes significance at 5%
Conclusions
• By exerting greater or lesser stability of ERs, ER regimes do affect FDI flows
• CU membership constitutes regime type most conducive to cross-border investment. EMU membership also appears to increase FDI flows with extra-EMU countries floating their currency (vis-à-vis double float country-pairs)
• FIX-FIX combination has most –ve impact on FDI flows
• Effects of other ER regime combinations are found to be either statistically indistinguishable from that of floating currency country-pairs or significantly –ve
• In comparing ER regime classifications, R+R emerges as the least robust and least able to capture year-to-year instability of FDI data
Variable
Description Source
fdi Log of real bi-lateral FDI flows. Sum of inward & outward flows (expressed in US$, converted to 2000 prices using US GDP deflator).
OECD International Direct Investment Statistics*
tbt Log of total real bi-lateral trade. Sum of exports plus imports expressed in US$, constant prices.
IMF Direction of Trade Statistics*
yi/yj
Log of real per capita GDP or real GDP, expressed in US$.
UN common database*
RXR Real exchange rate calculated using the nominal bilateral exchange rate and consumer price indices for both countries.
IMF International Financial Statistics/ OECD Main Economic Indicators*
RXRVOL
Real ER volatility. Annual standard deviation of the monthly percentage changes in the real bi lateral ER.
LRRXRVOL Measure of long-run ER volatility. Standard deviation of the log monthly changes in the real ERs using observations from preceding five years (see Clark et al., 2004).
IMF International Financial Statistics/ OECD Main Economic Indicators*
FTA Dummy that takes the value of 1 if both countries share a free trade agreement.
World Trade Organisation
(resi×resj)t Log of the product of indices of FDI regulatory restrictiveness. OECD (2007)
infi×infj Log of the sum of informational flows. Number of fixed telephone lines per 1000 people.
World Bank, World Development Indicators*
Notes: * indicates that the series was taken from http://www.esds.ac.uk.
Selected references
Mélitz, J., 2001. Geography, trade and currency unions. CEPR Discussion Paper 2987, October.
Mélitz, J., 2005. North, south and distance in the gravity model. Mimeo, University of Strathclyde.
Reinhart, C., Rogoff, K., 2004. The modern history of exchange rate arrangements: a reinterpretation. Quarterly Journal of Economics 119, 1-48.
Rose, A., 2000. One money, one market: the effect of common currencies on trade. Economic Policy 15, 7-33.
Schiavo, S., 2007. Common currencies and FDI flows. Oxford Economic Papers (Advance Access), March 3, 1-25.
Shambaugh, J.C., 2004. The effect of fixed exchange rates on monetary policy. Quarterly Journal of Economics 119, 1, 300-351.