Welfare Aspects of FDI to Source Country: An Example of Increasng FDI n Chna Shh-Hsun Hsu, Yungho Weng Kung-Chung Hsu, Chng-Cheng Chang, Fang-Chu Tu * Abstract Ths paper takes FDI n Chna as an example to evaluate the mpacts of FDI on source countres. The smulaton results show that f captal rewards repatrated mechansm s not mplemented, although FDI could beneft the socal welfare of some source countres by term of trade effect, the loss of captal stll domnate the total welfare effect. Source countres such as Tawan, Hong Kong, Korea and Sngapore may not suffer loss f captal rewards repatrated mechansm s mplemented. Therefore, t should be consdered to source countres to mplement a captal rewards repatrated mechansm for reducng the loss of socal welfare by captal dran. Key words: foregn drect nvestment n Chna, welfare decomposton, captal rewards repatrated mechansm, computable general equlbrum model JEL classfcaton number: F21, D58 * Correspondence: Shh-Hsun Hsu, Department of Agrcultural Economcs, Natonal Tawan Unversty, Tape, Tawan. Tel: 886-2-33662677; Fax: 886-2-23628496; emal: m577@ntu.edu.tw. Shh-Hsun Hsu s a professor n the Department of Agrcultural Economcs, Natonal Tawan Unversty, Tape, Tawan. Yungho Weng s a professor n the Department of Economcs, Natonal Chengch Unversty. Kung-Chung Hsu s a graduate student of Department of Economcs, Texas A&M Unversty. Chng-Cheng Chang s a research fellow n the Insttute of Economcs, Academa Snca, Tape, Tawan and a professor n the Department of Agrcultural Economcs, Natonal Tawan Unversty, Tape, Tawan.. Fang-Chu Tu s an assocate professor n the Department of Internatonal Trade, Chhlee Insttute of Commerce, Tape, Tawan.
1. Introducton Along wth rsng nternatonal trade, foregn drect nvestment (FDI) s also becomng an mportant economc actvty worldwde. Many host countres are vyng for nvestment from multnatonal enterprses. Source countres, however, are conservatve. For the host country, FDI s not only a form of captal nflow but can also be consdered as a form of technology transfer. Therefore, many countres have been mplementng favorable nvestment polces to attract foregn nvestment. For the source country, FDI may mean some advantages for ts labor force (Skaksen and Sorensen, 2001) and can also boost the country s R&D ntensty (Lu, Lee and Weng, 2000). Nevertheless, whether the frms n a source country drectly nvest abroad or they frst plan to nvest n ther country and then transfer to host countres, for the source country, ths s a form of captal dran or ndustry eroson. Ths phenomenon may result n dendustralzaton. 1 On the other hand, f consderaton s gven to the countres blateral trade, FDI may be seen as beneftng the source country. The reason for ths s whle the host country s ncome ncreases due to FDI, t wll also stmulate the host country s purchasng power for ntermedate and fnal goods from the source country. In addton, source country government may ask outward FDI frms to remt ncome from the host country back to the source country a captal rewards repatrated mechansm. The source country s ncome thus grows and welfare ncreases. Therefore, t s debatable whether or not FDI causes a negatve mpact on the source country. The queston probably can only be answered emprcally. Ths paper s amed at explorng ths ssue. Manland Chna adopted an open door polcy n 1979. From that tme, t started an FDI boom domestcally whch was ranked second n the world snce 1993. By the end of 1999, around 307.6 bllon U.S. dollars flowed to Chna through FDI. Gven the sgnfcant FDI actvty, we wll take Chna as an example to evaluate the 1 By usng U.S. ndustral data, Staudohor and Holly (1987) conclude that captal outflow may result n dendustralzaton. 1
mpacts of FDI on the source countres. The lterature on FDI s mpact on the source country s few. But f FDI s seen as a form of nternatonal captal moblty, the lterature abounds. For example, Ruffn (1984) and Myagwa (1990) employed theoretcal models to explore captal moblty s effects on welfare of both host and source countres. They have shown that captal moblty benefts both. Ths paper follows Myagwa (1990) by usng a dual approach as a theoretcal bass. Through ths method, we can decompose the effects of FDI on welfare for both host and source countres. We then apply computable general equlbrum model to smulate the effects that are decomposed. The rest of ths paper s organzed as follows. Secton 2 modfes the general equlbrum model of Myagwa (1990). Secton 3 dscusses database and emprcal model settngs. Ths s followed by secton 4 that uses tme seres data from 1979 to 1998 to calculate smulaton shocks. Secton 5 analyzes the results whle secton 6 concludes the paper. 2. Welfare decomposton We extend the model of Myagwa (1990) to embody Armngton s (1969) assumpton that goods are dfferentated on the bass of country of orgn. In addton, we utlze a welfare decomposton method to evaluate the effects of an ncrease n FDI. Snce captal outflow decreases domestc nvestment n the short run, and hence employment, source countres may mpose some crtera to control captal flow. In the followng, we wll dscuss two cases whether or not the so-called captal rewards repatrated mechansm s mplemented. 2.1 Captal rewards repatrated mechansm s not mplemented Consder a smple general-equlbrum model where captal moves between countres wthout restrctons. In equlbrum, a country s consumpton expendture should be the value of producton. Thus, 2
j j ( p pm, u ) R ( p, px, L K ) e,, =, (1) j where e ( p p, u ), s the mnmum expendture necessary for the representatve m country to acheve utlty level u when the vectors of domestc prces p, and prces of mports from exportng country j ( j ( p px, L K ) R, j p m, j ) are obtaned. 2 The scalar, s the maxmum attanable revenue for country at the same domestc prces j p, and the prces of exports to the other countres j ( p x, j ), gven the country s mmoble labor supply L and moble captal supply K. We assume that captal s perfectly moble nternatonally. In equlbrum, the rate of return on captal s R for country wll be equal to that n the world, K * r. That * R = r, (2) K where RK = R / K. Takng the total dfferental of equaton (1) and usng equaton (2), the welfare changes ( e udu ) from captal movement through FDI can be derved as: 3 e u du j j j j * = x dpx m dpm + r dk, (3) where j m and x j are country ' s mport and export quanttes from and to country j. Under Armngton s (1969) assumpton, m j s not equal to x j. 2 Throughout ths paper, the frst and second notatons n superscrpt represent the exportng and the mportng country, respectvely. 3 We mpose market equlbrum condtons that the export equals the mport ( j j x = m, j ) for each commodty and the domestc demand ( e p = e / p ) equals the domestc supply ( R p = R / p ). 3
Hence, the mpacts of FDI on welfare, as shown n equaton (3), can be decomposed nto three effects. The frst and second are the export-prce effect (EPE) and the mport-prce effect (IPE), respectvely. The sum of the two effects s referred to as the terms-of-trade effect (TOTE). An ncrease n the export prces or a decrease n the mport prces mproves the terms of trade, and s benefcal to country. The thrd effect s the captal-moblty-effect (CME), whch s welfare mprovng for country f captal moves n. On the contrary, country ' s welfare wll decrease f captal moves out. The total effect (TE) for country from captal movement through FDI s the sum of the three effects dscussed above. Although CME has negatve effects on the source country ' s welfare, from equaton (3), welfare can be mproved from outward FDI f TOTE s postve and s greater than CME. The same reasonng can be appled to the fact that FDI s not always benefcal to the host country. If TOTE s negatve and domnates CME, the host country s welfare wll decrease. 2.2 Captal rewards repatrated mechansm s mplemented If we assume that the source country eventually wll repatrate the captal rewards, equaton (1) can be modfed as: e e a b a ba a a a ab a a f * f ( p, p, u ) R ( p, p, L, K + K ) r K m =, (4) b ab b b b ba b b f * f ( p, p, u ) R ( p, p, L, K K ) + r K m x =, (5) x where country a and b represent the host and the source country, respectvely. The p, j p m and p (, j = a, b, and j ) are the vectors of domestc prces, the j x mport and export prces, respectvely. The captal endowment, respectvely. The L and K are country ' s labor and f K s the volume of captal that moves to the host country a through FDI from the source country b. The r * ( = ) represents the rate of return n the world when the captal market s n equlbrum. R K f 4
Takng total dfferental of equatons (4) and (5), the welfare changes from captal movement through FDI on the host and source countres can be derved as: e a u du a x dp m dp K dr, ab ab ba ba f * = x m (6) e du ba ba ab ab f * = x dp m dp + K dr (7) b b u x m. Both equatons (6) and (7) show that the welfare mpacts of FDI can also be decomposed nto three effects. The frst and second effects are the export-prce effect (EPE) and the mport-prce effect (IPE) whch are dscussed n the prevous case that repatraton mechansm s not mplemented. The sum of both effects s the terms-of-trade effect (TOTE). The thrd effect that s dfferent from the prevous case s the captal-rewards-effect (CRE). When captal market s perfectly compettve, the rate of return of FDI wll be equal to the world nterest rate ( r * ). Compared to the case of captal rewards that are not repatrated and shown n equaton (3), the captal movement affects welfare ndrectly ( ( r * dk * K f dr ) nstead of drectly ). The total effect (TE) for the host and source countres from FDI s the sum of the three effects dscussed above and documented n equatons (6) and (7), respectvely. 3. Database and emprcal model settng The smulaton model and database that we use n ths paper are Global Trade Analyss Project (GTAP) whch s constructed by the Center for Global Trade Analyss at Purdue Unversty. Gven the statstcal data book publshed by the government n Chna n 1997, we select the top 15 source countres nvestng n Chna. We then aggregate the GTAP database s (Verson 5) 66 regons nto the 15 regonal groups. They are Tawan (TWN), Chna (CHN), Hong Kong (HKG), Australa (AUS), Japan (JPN), Korea (KOR), Canada (CAN), Unted States (US), Malaysa (MYS), Sngapore (SGP), Unted Kngdom (UK), Germany (GER), France (FRA), Netherlands (NLD) and the rest of world (ROW). Furthermore, we aggregate GTAP s 57 detaled sectors nto 3 sectoral groups. These are food, manufacture and servce. 5
Snce the base year of the GTAP database (Verson 5) s 1997, the welfare mpacts that we smulate wll be the mpacts from that year. However, we are more concerned wth current mpacts from the current year of 2002. Therefore, we frst evaluate the effects of FDI ncrease on the world economy from 1997 to 2001. Gven the results, we further smulate the effects of rsng FDI n Chna n 2002 on the world economy, and examne the welfare mpacts on the source countres. In order to nterpret why captal moves nto Chna, we modfy the model of Malcolm (1998). Malcolm ntroduces the concept of country rsk nto the GTAP model and explores how captal moves nto South Afrca due to country-rsk reducton durng the 1990s. In the model settng, Malcolm (1998) assumes that captal s perfectly moble. In equlbrum, the expected rate of return under country rsk consderaton wll be equal to the global net rate of return on captal stock. That s, RORE () r / RISK( r) = RORG, (8) where RORE () r s the expected rate of return wthout consderng rsk n country r, RISK () r s a rsk coeffcent n country r, and RORG s the global net rate of return on captal stock. The value of RISK ( r) s greater than 1 f country r s rsker than the rest of the world. Otherwse, t s less than 1. Followng the model of Malcolm (1998), we ntroduce the concept of nvestment attractveness to explore how captal moves nto Chna due to an ncrease n ts attractveness. When a country s more attractve n terms of potental market sze, labor productvty and so on, t wll absorb more FDI. Therefore, equaton (8) can be modfed as follows: RORE () r INVATT ( r) = RORG, (9) where INVATT () r s a coeffcent of nvestment attractveness n country r. If country r s more attractve than the rest of the world, the value of INVATT () r s 6
greater than 1. Otherwse, t s less than 1. Equaton (9) demonstrates that, gven RORG () r, the more attractve the country s, the lower s the RORE () r requred for frms to nvest n the country. We rewrte equaton (9) as a percentage-change form rore( r) = rorg nvatt( r), (10) where rore () r, () r nvatt and rorg represent the percent changes of RORE () r, INVATT () r and RORG, respectvely. Equaton (10) shows that n equlbrum the percent change of the expected rate of return wthout consderng nvestment attractveness n country r s equal to the percent change of the global net rate of return on captal stock subtractng the percent change of the nvestment attractveness n country r. 4. Calculaton of smulaton shock In ths secton, we wll estmate the level of INVATT ( r) and ts percentage change as the smulaton shock. We then mplement the shock to smulate welfare mpacts on the source and host countres. Snce nvestment attractveness cannot be quantfed, we assume that foregn nvestors evaluate the level of attractveness of a certan country accordng to the determnants of FDI n that country. In the followng analyss, we frst dentfy the determnants of FDI by usng econometrc estmaton and then calculate the elastctes and percentage changes of the determnants. We can therefore estmate the percentage change of nvestment attractveness. Prevous lterature dealng wth determnants of FDI n Chna ncludes Lu et al. (1997), Dees (1998) and Wang and Swan (1997). Usng 22 countres and 12 years of panel data, Lu et al. (1997) shows that the mport and export values, the relatve values of GDP, wage, exchange rates and cultural dfferences play mportant roles. Dees (1998) employs 11 countres and 12 years of panel data to dentfy the determnants of FDI n Chna. The results demonstrate that the market sze n Chna, cost advantages of producton, and openness to the rest of the world seem to be 7
relevant. Compared to Lu et al. (1997) and Dees (1998) usng panel data, Wang and Swan (1997) studes ths ssue by usng tme seres data. They argue that when the market sze n Chna measured by real GDP (GDP) ncreases, FDI n Chna ncreases too. Both varables are postvely related. Tarffs are one of trade barrers. Hgher tarff rates dscourage exports and encourage FDI. Hence, both tarff levels (TARIF) and FDI n Chna are postvely related. Wage rate s one of the producton costs. If wage rate n the host country ncreases compared wth the source country, frms wll have more ncentve to penetrate overseas markets through export rather than FDI. Both relatve wage rate (WAGE) and FDI n Chna are negatvely related. Long-term bond rates n the U.S. (BOND) and Chna s central bank dscount rates (DISCOUNT) represent the nvestor s opportunty costs of captal. As these two rates ncrease, FDI are dscouraged. In addton, from the theoretcal pont of vew, a devaluaton of the host country s currency (EXRT) makes the nvestors pay less to purchase land or equpment n the host country, resultng n an ncrease n FDI. The relatonshp between mports (IMP) and FDI s ndetermnate. If both are mutually substtutable, t then exhbts a negatve correlaton. If, however, both are complementary, then the relatonshp s postve. Chna s labor productvty (PRDTVY) s also another factor that attracts foregn nvestment. Hgh labor productvty attracts more FDI and thus the relatonshp s postve. In addton, Wang and Swan (1997) choose the growth rate of OECD as one of the determnants. They argue that hgher growth rate of OECD (GROECD) shows that nvestors from the member countres have more potental to nvest n Chna a postve relaton. Nevertheless, an ncrease n OECD average growth rate wll also mean a hgher domestc demand and hgher local nvestment, eatng up on possble nvestment to Chna. Gven ths possblty, the relatonshp s ndetermnate. Fnally, dummy varable (D) s used to represent the poltcal stuaton n Chna. The value of D s 1 n the years of 1986, 1987 and 1989. Otherwse, D s set to 0. In the followng, we accept Wang and Swan s (1997) arguments, expand the tme seres data to nclude 1977 to 1998, and re-estmate the determnants of FDI n Chna. 8
A detaled descrpton of the data and the sources are provded n the appendx. In addton, the relatonshp between dependent and ndependent varables s assumed to follow a Cobb-Douglas functonal form for smplfcaton. To avod the problem of spurous regresson, we frst proceed a unt root test on each of the varables. Table 1 shows the results of the unt root test on the ten varables. It can be seen that each varable s statonary after the frst dfferental s taken. Table1: The results of unt root test Lag ADF TEST Order of ntegraton ln(fdi) 1-4.10** wth drft and trend I(1) ln(gdp) 0-3.83 ** wth drft and trend I(1) ln(tarif) 1-3.95** wth drft and trend I(1) ln(discount) 8 7.828*** wth drft and trend I(1) ln(bond) 8-4.71** wth drft and trend I(1) ln(wage) 8-9.31*** wth drft and trend I(1) ln(exrt) 3-4.10*** wth drft and trend I(1) ln(imp) 8-7.49*** wth drft and trend I(1) ln(groecd) 9-9.23 *** wthout drft and trend I(1) ln(prdtvy) 1-4.48** wth drft and trend I(1) Note: The method of unt root tests s based on Doldado, Jenknson and Sosvlla-Rvero (1990). The *** and ** represent 1 and 5 percent sgnfcant n a two-taled test, respectvely. Each varable s drft and trend are determned through 5 percent sgnfcant n a two-taled test. Lag s based on AIC s smallest value. Therefore, takng the frst dfferental of each varable results n the followng: d ln ( FDI ) = α0 + β1d ln( GDP) + β2d ln( TARIEF ) + β3d ln( WAGE) + β 4d ln( BOND) + β d ln( DISCOUNT ) + β d ln( EXRT ) + β d ln( IMP) + β d ln( GROECD) 5 + β d ln( PRDTVY ) + β D + ε (11) 9 10 6 t 7 8 After runnng regresson for equaton (11) by usng step-wse regresson method, we obtan the best combnaton of predctors that are IMP, GROECD and PRDTVY. Results are dsplayed n Table 2. 4 Table 2 shows that IMP and PRDTVY are postvely correlated whle GROECD s negatvely correlated wth FDI n Chna. Ths means that there s a complementary relatonshp between Chna s mports and FDI. An ncrease n labor productvty n 4 In ths paper, we carry out the contegraton test by usng Johansen method. A detal descrpton of the testng process s provded n the appendx. 9
Chna helps n attractng FDI. Nevertheless, hgher economc growth n OECD countres wll make ther frms to nvest more n ther own countres than n Chna. Furthermore, gven the log-lnear functonal form taken n equaton (11), the coeffcent value for each varable n Table 2 represents the elastcty for the varable. Table 2: Independent varable selecton IMP GROECD PRDTVY Coeffcent 3.73-0.14 9.94 Std. Error (0.60)*** (0.08)* (3.25)*** R 0.674068 D-W 1.977670 Note: The *** and * represent 1 and 10 percent sgnfcant n a two-taled test, respectvely. Fnally, the predcted changes of IMP, GROECD and PRDTVY can be derved. They are 15.25%, -19.38% and 1.14% from 1997 to 2001, respectvely, and are 2.43%, -3.96% and 0.13% from 2002 to 2003, respectvely. The contrbuton of each varable to FDI can be obtaned through the product of the elastcty and the predcted percentage change. The sum of the contrbuton from each varable s the total contrbuton whch s the predcted change n FDI. Therefore, as shown n Table 3, the predcted change of FDI (or nvestment attractveness) s 70.97% from 1997 to 2001, and s 10.92% from 2002 to 2003. Table 3: Predcted change n FDI n Chna Unt: % IMP GROECD PRDTVY Total Elastcty: (A) 3.73-0.14 9.94 predcted changes of the varables from 1997 to 15.25-19.38 1.14 2001: (B) Contrbuton: (A) * (B) 56.83 2.77 11.34 70.97 percentage changes of the varables from 2002 to 2.43-3.96 0.13 2003: (C) Contrbuton: (A) * (C) 9.06 5.68 1.29 10.92 In addton, from equaton (10), the effect of an ncrease n nvatt (r) s equvalent to a decrease n rore (r) f we follow Malcolm (1998) to assume that a change n rore (r) has lttle mpact on rorg. Therefore, the percentage shocks on 10
rore (r) are -70.97% and -10.92% for the two perods of 1997 to 2001 and 2002 to 2003, respectvely. 5. Smulaton Results Table 4 lsts the smulaton results. The results n column (6) show that Chna, the host country, s the only benefcary f captal rewards repatrated mechansm s not mplemented. All the source countres nvestng n Chna ncur negatve welfare mpacts. In partcular, those countres, such as the U.S., Japan and Germany, havng hgher FDI ntensty n Chna receve larger reducton n welfare. As shown n column (4), welfare reducton manly comes from the reducton n CME. Although some source countres, such as Tawan, Hong Kong, Korea and Sngapore, gan n TOTE as shown n column (3), the effects are too small to cover the loss n CME. Table 4: Welfare changes under smulaton Unt: mllon U.S. dollar, 1997 EPE IPE TOTE CME CRE Total Effect (1) (2) (3)=(1)-(2) (4) (5) (6)*=(3)+(4) (7)**=(3)+(5) TWN -1290-1694 404-1790 28-1386 432 CHN 2280-5952 8232 181205-387 189437 7845 HKG -587-1239 652-2667 176-2015 828 AUS -994-1004 10-11473 3-11463 13 JPN -8291-5326 -2964-172102 37-175066 -2927 KOR -1861-2302 441-7011 18-6569 460 CAN -3640-3343 -297-18730 3-19026 -294 USA -14419-13824 -594-221758 28-222352 -567 MYS -1275-1248 -27-1699 3-1726 -24 SGP -1647-1964 318-898 22-580 340 UK -5247-5177 -70-45654 16-45723 -54 GER -9134-7660 -1474-110643 8-112117 -1466 FRA -5611-4732 -879-61868 4-62748 -875 NLD -3147-2996 -151-14483 4-14633 -147 Note: * represents that captal rewards repatrated mechansm s not mplemented. ** represents that captal rewards repatrated mechansm s mplemented. It s noted that CME s the effect of net captal flow n equlbrum. Countres wth hgher ntal FDI nvestng n Chna may not suffer hgher loss n CME. For example, Hong Kong, the bggest nvestor n Chna, suffers less loss than the U.S., Japan and Germany. The reason s that mprovement on terms of trade n the source country may attract captal to flow n. Ths wll reduce the negatve effect of CME 11
n Hong Kong. However, source countres may not suffer loss f captal rewards repatrated mechansm s mplemented. The results n column (7) show that there are stll negatve welfare mpacts on the U.S., Japan and Germany. However, compared to the results n the case of not havng captal rewards repatrated mechansm, the mpacts are relatvely small. Some source countres such as Tawan, Hong Kong, Korea, and Sngapore turn out to be the benefcares. The reason s that the TOTE plays a more mportant role than CRE on welfare changes. 6. Concluson In general, FDI s a form of captal dran or ndustry eroson for the source country. However, f blateral trade between source and host countres s consdered, FDI may be seen as beneftng the source countres. In addton, the government n source country may ask outward FDI frms to remt ncome from the host country back to the source country a captal rewards repatrated mechansm. The source country s ncome thus grows and welfare ncreases. Therefore, t s debatable whether or not FDI causes a negatve mpact on the source country. By takng FDI n Chna as an example, ths paper follows Myagwa (1990) to decompose the effects of FDI on welfare for both host and source countres. We then apply computable general equlbrum model to smulate the effects that are decomposed. The smulaton results show that f captal rewards repatrated mechansm s not mplemented, all the source countres nvestng n Chna experence negatve welfare mpacts. However, source countres such as Tawan, Hong Kong, Korea and Sngapore may not suffer loss f captal rewards repatrated mechansm s mplemented. 12
7. Appendx 7.1 Data descrptons Symbol Mean (Std. Dev.) FDI 312.31 (382.05) GDP 33.02 (16.92) Unt USD 100 mllon RMB mllon Data source Statstcal Yearbook of Chna Internatonal Fnancal Statstcs (IFS) Yearbook TARIF 9.02 (5.89) WAGE 2.46 (0.61) BOND 8.74 (2.40) DISCOUNT 7.64 (2.31) EXRT 9.76 (3.23) % Statstcal Yearbook of Chna % Chna Labour Statstcal Yearbook & Internatonal Fnancal Statstcs (IFS) Yearbook % Internatonal Fnancal Statstcs (IFS) Yearbook % Internatonal Fnancal Statstcs (IFS) Yearbook % Internatonal Fnancal Statstcs (IFS) Yearbook IMP 9.71 (2.45) USD 100 mllon Internatonal Fnancal Statstcs (IFS) Yearbook GROECD 2.75 (1.19) % OECD Economc Outlook PRDTVY 106 (3.29) None Chna Labour Statstcal Yearbook; Index of Socal Labour Productvty 13
7.2 Testng for Contegraton The ntegraton order of all the ndependent varables and dependent varable are 1. Therefore, contegraton can be used to verfy whether there s a long-run equlbrum relatonshp between the varables. We frst fnd a sutable lag length whch makes resduals to be whte nose. By Ljung-Box test, the results show that number of lags s 2 where 2 χ test s 85.204 and p-value of 0.04. After determnng the lag length, we need to choose a sutable contegraton model. Followng Pantula s (1989) prncple, the table below shows that the sutable model has no determnstc trends, contegratng equatons have ntercepts, and contegratng rank s 3. Therefore, there s a long-run equlbrum relatonshp between the varables. Johansen s contegraton test H 0 H 1 Trace 95% crtcal value Egenv. r 0 r = 1 77.68 53.42 0.8278 r 1 r = 2 40.74 34.86 0.6382 r 2 r = 3 19.39 19.99 0.4750 r 3 r = 4 5.86 9.13 0.2434 14
References Armngton, P. S., 1969, A theory of demand for products dstngushed by place of producton, Internatonal Monetary Fund Staff Papers 16, 159-78. Dees, S., 1998, Foregn drect nvestment n Chna: determnants and effects, Economcs of Plannng 31, 175-94. Dolado, J. J., T. Jenknson, and S. Sosvlla-Rvero, 1990, Contegraton and unt roots, Journal of Economc Surveys 4, 249-73. Johansen, S., 1998, Statstcal analyss of contegraton vectors, Journal of Economc Dynamcs and Control 12, 231-54. Lu X., H. Song, Y. We and P. Romlly, 1997, Country characterstcs and foregn drect nvestment n Chna: a panel data analyss, Weltwrtschaftlches Archv 133, 313-29. Lu, B. J., H. F. Lee, and Y. Weng, 2000, The mpact of FDI on technology and socal phlosophy: a dscusson of R&D and polluton preventon, Socal Phlosophy and Technology of Tawan and Chna. L. E. Chen (eds), Tape: Chung-Hua Insttuton for Economc Research Press. Lu, X., C. Wang, and Y. We, 2001, Causal lnks between foregn drect nvestment and trade n Chna, Chna Economc Revew 12, 190-202. Malcolm, G., 1998, Model country rsk and captal flows n GTAP, GTAP Techncal Paper No. 13, 1998. Myagwa, K., 1990, Internatonal captal moblty and natonal welfare, Garland, New York and London. Pantula, S. G., 1989, Testng for unt roots n tme seres data, Econometrc Theory 5, 256-71. Ruffn, R. J., 1984, Internatonal factor movements, Handbooks n Economcs seres no. 3, Jones, R. W. & P. Kenen (eds), New York: Elsever Scence. Skaksen, M. Y., and J. R. Sorensen, 2001, Should trade unons apprecate foregn drect nvestment? Journal of Internatonal Economcs 55, 379-90. Staudohor, P. D., and E. B. Holly, 1987, Dendustralzaton and plant closure, Lexngton, Mass and Toronto: Heath, Lexngton Books. Wang, Z. Q., and N. J. Swan, 1995. The determnants of foregn drect nvestment n transformng economes: emprcal evdence from Hungary and Chna, 15
Weltwrtschaftlches Archv 131, 359-82. 16