Exposure to External Shocks and the Geographical Diversification of Exports Marc Bacchetta (WTO), Marion Jansen (WTO/ILO), Carolina Lennon (Universit( Université de Paris 1) and Roberta Piermartini (WTO) World Trade Organization, Geneva 20 May 2009
The issue: what explains volatility Output volatility 0.02.04.06.08 Output volatility by group 1960 1970 1980 1990 2000 2010 Year Developed LDCs Developing
OUR QUESTION Does geographical export diversification reduce income volatility? shocks may be product specific but also country specific fixed costs of entry into new markets make it difficult (costly and takes time) to redirect exports The ability to face country specific shocks will depend on geographical diversification
What do we know about levels What do we know about levels of export diversification? 200 High income Middle income Low income 150 100 50 1962 1967 1972 1977 1982 1987 1992 1997 2002 1962 1967 1972 1977 1982 1987 1992 1997 2002 1962 1967 1972 1977 1982 1987 1992 1997 2002 year (mean) Nproducts (mean) Npartners Collapsed by simple mean
What do we know about levels of export diversification? Variable income 1963-1972 1973-1982 1983-1992 1993-2002 Number of products Number of partners ToT barter High 152 143 150 153 Middle 108 109 116 139 Low 91 91 97 108 High 109 109 121 144 Middle 64 67 77 100 Low 48 55 68 73 High 4.21 3.21 Middle 15.42 9.52 Low 16.10 14.20
Measuring exposure to external shocks ECSS i = x i, j 2 ( ) var GDPgrowth j + j Xi j z x i, j X i x i, z X i cov( GDPgrowth j, GDPgrowth z ) Exposure to Country Specific Shocks depends on Volatility of GDP in export markets (var( var) Degree to which cycles in partner countries are correlated (cov) Weight of individual markets in total exports (x( i,j, x I ). Therefore related to: Geographical concentration of exports (Herfindahl( index) Number of partners
Portfolio (Growth, GDP_const) 5 4 3 5 years East Asian Financial Crisis 1997 2 1 0 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 year 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 Oil crisis P ort. Tot al Va r-c ov (Gro wth ) GDP _co nst, 5 yrs 1974 Energy Port. Cov. (Growth)GDP_const, 5 yrs Crisis 1981 Collapsed by simple mean Port. Var. (Growth) GDP_const, 5 yrs Volatility (sd) (G rowth) GDP_cons t, 5 yrs
Singapore: exposure to external shocks Port. Var-Cov 6 5 4 3 2 Portfolio (G rowth, SGP) SGP East Asian Financial Crisis 1997 1 0 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 year 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Port. Total Var-Cov (Growth) GDP_const, 5 yrs P ort. Cov. (G rowth) G DP _const, 5 yrs Port. Var. (Growth) GDP_const, 5 yrs Argentina Crisis 2001 Vol. of the Growth of GDP_const(sd), over 5 yrs
Brazil: exposure to external shocks 6 5 Portfolio (Growth, BRA) BRA East Asian Financial Crisis 1997 Port. Var-Cov 4 3 2 1 0 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 year 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 P ort. T ota l Va r-c o v (G ro wth ) G D P_ co nst, 5 yrs P ort. Cov. (G rowth) G DP _const, 5 yrs P ort. Va r. (G row th ) G D P_c on st, 5 yrs Argentina Crisis 2001 V ol. of the Gro wth of GD P_ co ns t(sd), over 5 yrs
Exposure to external shocks depends on geographical diversification (U-shape) Results from panel regressions Number of partners * world volatility - 0.039*** -0.041*** Number of partners squared * world volatility Herfindahl index * world volatility Herfindahl index * world volatility 0.000*** 0.000*** 11.842*** 11.810*** -6.141* -6.114* openness -1.028* -0.553
ToT volatility GDP volatility depends on external shocks Full sample Overlapping periods Non-overlapping overlapping periods Restricted sample 0.028*** 0.028*** 0.027*** 0.036 0.040** ECSS 0.039* 0.220** 0.199** covariance 0.570*** variance 0.037* openness 1.859*** 1.885*** 1.928*** 1.901 1.782
Stylized facts GDP volatility decreases with level of development Diversification increases with level of development, both in terms of product and in terms of geographical diversification The difference between high and middle income countries is highest in terms of geographical diversification The difference between middle and low income countries is highest in terms of product diversification
Findings of empirical exercise Exposure to external country specific shocks contributes to GDP volatility Geographical diversification decreases exposure to external shocks An optimal mix of trading partners takes into account the correlation between partners business cycles