Do Japanese Foreign Direct Investment and Trade Stimulate Agricultural Growth in East Asia? Panel Cointegration Analysis

Transcription

1 Do Japanese Foregn Drect Investment and Trade Stmulate Agrcultural Growth n East Asa? Panel Contegraton Analyss Weerapong Sattaphon Doctoral student: wsattaphon@hotmal.com Department of Agrcultural and Resource Economcs, The Unversy of Tokyo Yayo, Bunkyo-ku, Tokyo Contrbuted paper prepared for presentaton at Internatonal Assocaton of Agrcultural Economsts Conference, Gold Coast, Australa, August 12-18, by Weerapong Sattaphon. All rght reserved. Readers may make verbatm copes of ths document for non-commercal purpose by any means, provded that ths copyrght notce appears on all such copes. 1

2 Do Japanese Foregn Drect Investment and Trade Stmulate Agrcultural Growth n East Asa? Panel Contegraton Analyss JEL Classfcaton: O4, Q17, R0 Keywords: Agrcultural Growth; East Asa; Foregn Drect Investment; Trade: Panel Data 1. Introducton The most mportant ssue n economc studes s how to promote countres growth, snce growth can have many mpacts on ther economes. Foregn drect nvestment (FDI) and trade are often seen as mportant catalysts for economc growth. FDI s an mportant vehcle for technology transfer from developed countres to developng countres and FDI could also promote growth by provdng addonal employment n a labor surplus economy. Internatonal trade s also recognzed as an nstrument of economc growth, snce trade faclates more effcent producton of goods and servces by shftng producton to more competve. Even though past studes show that FDI and trade have a posve mpact on growth, the observed growth n such studes was the overall growth of the economy. There was, however, ltle specfc focus on growth n the agrcultural sector, whch s the most mportant and sgnfcant sector n developng countes. East Asa s sad to be the most dynamc economc development regon, however the agrculture sector has been gnored as a possble catalyst for promotng economc growth. Even though agrcultural gross domestc product (GDP) only accounts for a small proporton of total GDP, the labor force n ths sector represents a hgh proporton of the total labor populaton (Table1). In order to help the many people who rely on ths sector s mportant to understand what factors help to stmulate growth n ths sector. Japanese FDI has played a val role n economc development n East Asa for many decades. Accordng to statstcal data of the Mnstry of Fnance, the number of Japanese multnatonal enterprses (MNE) nvestng n East Asan countres from 1987 to 2003 was 952, wh a value of $3.5 bllon (Table 2). It s generally understood that Japanese FDI helps to accelerate growth by ncreasng local employment and enhancng productvy va technology 2

3 transfers and spllover effects (Urata and Kawa, 2000), but there have not yet been any studes to confrm ths argument. Therefore, n order for the host countres to make the proper polces towards Japanese FDI, s necessary to nvestgate these mpacts emprcally. The purpose of ths paper s to examne the factors that contrbute to the growth of the agrcultural sector n East Asan countres, by partcularly focusng on Japanese FDI 1 and nternatonal trade as the catalysts for growth. 2. Lerature Revew of FDI, Trade and Growth A number of studes have dscussed how nward FDI can contrbute to the growth of a host country s economy (e.g. Balasubramanyam, 1996; Borensten, 1999; Buckley et al, 2002; Carkovc et al, 2005; Choong, 2004; Kosempel, 2004; de Mello, 1999). In general, these studes argue that the mpact of FDI by multnatonal enterprses (MNEs) on growth s complex. Frst, through capal accumulaton n a recpent economy, FDI s expected to be growth-enhancng by encouragng the ncorporaton of new nputs and foregn technologes nto the producton functon of the host economy. Second, FDI mproves the effcency of locally-owned host country frms va contract and demonstraton effects, and ther exposure to ferce competon. Last and most mportantly, FDI s beleved to be a leadng source of technologcal change and human capal augmentaton n developng countres. Technologcal progress takes place through a process of capal deepenng n the form of new varetes of knowledge-based capal goods. It also proceeds va specfc productvy-ncreasng labor tranng and skll acquson promoted by MNEs. In addon, nternatonal trade has long been consdered as a catalyst for growth. Most leratures suggest that trade contrbutes to economc growth largely by openng access to ntermedate nputs, amplfyng the learnng-by-dong process, and expandng the sze of global 1 Although other sources of FDI may also have an mpact on growth, n ths paper we wll only consder Japanese FDI, for two reasons. Frst, n terms of the growng dea for regonal ntegraton n East Asa, we are nterested to see f Japanese FDI has an mpact on agrcultural growth and can enhance regonal ntegraton. Second, no data s avalable for nward FDI n the agrcultural sector of East Asan countres, or s only reported on a BOP bass, and so we need to use outward data, whch only Japan provdes. 3

4 market (Edwards, 1993: Frankel and Romer, 1996; Krshna et al., 2003). The effect of trade on growth can also be seen as n term of exports and mports. The export-led growth hypothess was emprcally nvestgated and supported by many studes whch mostly found that exports promote growth by mprovng effcency n the allocaton of resources, encouragng factors of producton to move from low- to hgh-productvy sectors and rasng total factor productvy (Esfahan, 1991; Ghrmay, 2001; Jn, 2002; Kavouss, 1984; Tremblay, 1990; Tyler, 1981). Recently, there s evdence of an ncrease n mports along wh exports. Snce there s much need of exchange of ntermedate and capal goods (Esfahan, 1991; Tremblay, 1990), thus resultng n many developng countres have experence much actvy emergng from mportng. Imports of foregn nputs are an mportant factor lnkng trade and growth, and so nternatonal trade can help to promote economc growth by provdng essental and effcent foregn nputs for the ndustral sector (Lee, 1995; Walde and Wood, 2004). 3. Econometrcs Model Specfcaton The conventonal neo-classcal producton functon was used to test the mpact of FDI on growth, but added foregn drect nvestment as an addonal varable because FDI s the prme source of human capal and new technology. Moreover, trade, exports and mports were also ntroduced as varables n the model to assess the export- and mport-led growth hypothess. Thus, the basc producton functon can be wrten as 2 Y = F( K, L, IMP, EXP, TRADE, FDI, t) (1) where Y s real agrcultural GDP, K s agrcultural land used, L s the labor force n the agrcultural sector, IMP and EXP are real agrcultural mports and exports respectvely, TRADE s the total of real agrcultural exports plus mports, whch s used to determne the level of trade lberalzaton, FDI s stock Japanese FDI n the agrcultural sector, t s a tme trend whch captures the mprovement n productvy due to techncal progress. Assumng the producton 2 Neher fertlzer use nor agrcultural machnery was ncluded n the agrcultural producton functon, due to lack of data. Domestc nvestment could also help to stmulate growth, but agan lmed avalably of data and problems wh natonal data recordng systems (BOP bass), prevented ths varable from beng ncluded n the model. 4

5 functon to be log-lnear, takng logs and dfferentatng wh respect to tme, thus the model can be llustrated as: Y & = α + β K& + β L& + β IMP & + β EXP & + β TRADE & + β FD& I + ε (2) where a dot over a varable denotes the rate of growth of ndvdual varables and ε denotes a random error term. In ths study, the panel data econometrc approach was appled. Therefore, the model needed to deal wh cross-secton (N) and tme seres (T) data. Therefore, the growth functon to be estmated can be llustrated as follows: Y& = α + β K& + β L& + β IMP & + β EXP & + β TRADE & + β FDI & + β D ε (3) = 1,2,,N t = 1,2,,T where refers to the number of ndvdual countres n the panel, t refers to the number of observatons over tme. In addon, a dummy varable representng the Asan crss of 1997 was also ntroduced nto the equaton to see whether would affect the equaton structure. It was assumed that the crss would have had an mpact n the followng year, In the context of panel data, the exstence of unobservable growth determnants that are specfc to regons or countres can be acknowledged and taken nto account n the estmaton procedures. Generally, a panel data set can be estmated n one of three ways, dependng on whether the ndvdual crosssecton effects are consdered to be constant, fxed or random. The correspondng statstcal models are the ordnary least squares (OLS) model, the random effects (RE) model, and the fxed effects (FE) model. OLS smply assumes that the unobservable ndvdual-specfc effects do not dffer.e., they are homogenous effects, whereas RE and FE consder these effects nto the model. However, the RE model was consdered napproprate for use n ths study because of s assumpton of the error term from unobservable ndvdual-specfc effects s uncorrelated wh We confrmed the mpact of 1997 Asan crss n two ways. Frstly, by plottng graph, the value of agrcultural GDP was dramatcally dropped after Secondly, we calculated the equaton (3) by comparng a 1997 tme dummy wh a 1998 tme dummy. The result clearly ndcated that there was a stronger mpact of the Asan crss n

6 the exogenous varables seems not to be real. Thus, n our study the homogeney test was nally appled to see whether the ndvdual effect stll exsted, and f no ndvdual effect exsted then the estmaton was smply determned by applyng the OLS model. If there was an ndvdual effect then the FE model would be appled. To start wh, equaton (3) s a pooled model representng a behavoral equaton wh the same parameters across countres. Equaton (4) s the same behavoral equaton but wh dfferent parameters across countres, and s assumed to be constant over tme. Thus, we can postulate a separate regresson for each ndvdual country: Y = α + β X + µ (4) = 1,2,,N t = 1,2,, T where α and β = β, β,..., β ) are 1 x 1 and 1 x K vectors of constants that vary across, ( 1 2 K X s a 1 x K vector of the exogenous varable n equaton (3), and u s the error term. The homogeney test could be conducted smply by testng the null hypothess of whether the parameters n the ndvdual models vared from country to country.e., whether they shared common ntercept and slope coeffcents: H o : α 1 = α 2 = = α N, β 1 = β 2 = = β N The test could be executed by applyng the followng F-test (Hao, 2003). F = ( ) ( N 1)( K + 1) 1 S [ ] 1 [ NT N( K + 1) ] S 2 (5) S S2, the resdual sums of squares (RSS), was obtaned from the pooled OLS and S1 was obtaned from summng the RSS from N ndvdual OLS regressons and K s the number of slope varables. If we assume u = N(0,σ 2 I NT ) then, under the null hypothess, ths test statstc s dstrbuted as F((N-1)(K+1), N(T-(K+1)). 4. The Data and Applcaton of Panel Contegraton Annual data were used coverng the perod Agrculture GDP, the labor force and agrcultural trade data were obtaned from Key Indcators of Developng Asan and Pacfc 6

7 Countres, ADB. Agrcultural land use was obtaned from the FAO, except for Tawan data, whch was obtaned from Tawan statstcal Data Book Japanese agrcultural sector FDI, accumulaton of food ndustry, fshery, agrculture and forestry sector FDI were obtaned from Fnancal Statstcs Monthly, Mnstry of Fnance, Japan. Snce the Japanese government does not report the stock value of FDI, a proxy was constructed by accumulatng the outward FDI annual flow values. Moreover, Japanese FDI data are reported accordng to ther fscal year, runnng from Aprl to March the followng year, thus FDI data were converted to calendar year data by the formula X = X + 3X ) / 4 before makng any estmates. The data on CY ( FY ( t 1) FY ( t) GDP, exports, trade and FDI were converted nto 1995 constant prces by usng the consumer prce ndex (CPI), whch was obtaned from the IMF, from each country. Mssng data were generated by the standard growth model. We nally nvestgated the tme seres propertes of the data to check that our data s consstent wh the model assumpton, to enable us to test for the long-run relatonshps between dependent varables and explanatory varables. 4.1 Panel Un Root Test We tested the data to determne for non-statonary varables aganst the alternatve of trend statonary varables, allowng dfferent ntercepts and tme trends for each ndvdual country. We appled the IPS un root test proposed by Im et al. (2003) snce ths test allows each panel member to have a dfferent autoregressve parameter and short run dynamcs under the alternatve hypothess of trend statonary. The IPS test statstc, referred to as the t-bar statstc, s based on the average value of the Augmented Dckey Fuller (ADF) test statstcs as follow: Ρ y = α + γ trend + β y + ρ y + t, 1 j, t j j = 1 ε (6) = 1, 2,,N; t = 1, 2,,T 7

8 where y s the varable under consderaton, s the frst order dfference operator, j = 1, 2,.,N, p s the lag length of y whch s chosen based on Akake Informaton Creron (AIC), ε s the error term and s ndependent and dentcally dstrbute and ρ s the estmate vector of coeffcent on the augmented lagged changes. All varables are expressed n natural logarhm. The null hypothess of un roots n the panel data s defned as β = 0 for all. (H 0 ) aganst the alternatves that all seres are statonary processes.to test the hypothess, IPS proposes a standardzed W-tbar statstc gven by W tbar ( Ρ, ρ ) = N t bar 1 N 1 NT N N Var = 1 N Ε[ ( Ρ ) = ] = t,0 1 T β 0 [ t ( Ρ,0) = 0] T β (7) where t-bar NT s the average of ndvdual ADF statstcs, whch can be expressed as t bar N NT = t = 1 T ( p, ρ ) N, and the value of E (, 0) β = 0 and ( 0) [ ] t T p [, = 0] Var t T p β are tabulated n Im et al. (2003). 4.2 Panel Contegraton Test In the emprcal applcaton of the panel contergaton test, we appled Pedron s contergaton test methodology (1999 and 2004) snce ths methodology allowed us to test for the presence of long-run equlbra n multvarate panels, whle permtng the dynamc, and even the long-run contegratng vectors to be heterogeneous across ndvdual members. In general, the Pedron s contegratng regresson takes the form of: y, t = + δ + β1 x1, t + β2 x2, t + + βm xm, t + e, t α K (8) for t = 1,, T ; = 1,, N ; m = 1,, M Pedron (1999, 2004) dscusses the constructon of seven panel contegraton statstcs for testng the null of no-contegraton, four based on poolng along the whn-dmenson, and three based on poolng along the between-dmenson. In each category, one test s parametrc ADFbased, and the other test nvolve the use of non-parametrc correctons. The test statstcs n the 8

9 whn-dmenson category are based on estmators that effectvely pool the autoregressve coeffcent across dfferent members, for the un root tests on the estmated resduals. In contrast, the test statstcs gven n the between-dmenson category are based on estmators that smply average the ndvdually estmated coeffcents for each member. The exstence of contegraton relatonshps between the varables was nvestgated through the statonary of the error term n equaton (8). The test statstcs n both categores have asymptotcally normal dstrbutons; x N, T ν µ N Ν ( 0,1) where x, s the approprately standardzed form for each of the statstcs. Also µ and υ are the N T correspondng values for each test of the mean and varance, respectvely, and gven n Pedron (1999). 5. Emprcal Result The results of the IPS un root tests ndcated that most of varables were not statonary at I(0) (Table 3). After performng the frst dfference operaton, however, all data were statonary at I(1). These results confrmed the necessary condons for performng panel contegraton, thus we contnued wh the dentfcaton long-run growth relatonshps by applyng the Pedron panel contegraton test. The results reported n Table 4 ndcated that the null of no contegraton was rejected by four of the panel tests, namely the panel pp-statstc, panel adf-statstc, group pp-statstc and group adf-statstc. These four rejectons of the null strongly ndcated the exstence of contegraton relatonshps n our growth relaton functon. Therefore, we concluded that our estmaton model could be nterpreted as a long-run growth relatonshp. In order to determne whether the estmaton of model (3) should be an OLS model or a Fxed effect (FE) model, the Homogeney F-test (5) was executed to assess whether the countres share a common ntercept and slope coeffcent. The results of the homogeney F-test 9

10 ndcated that there was no common slope and ntercept for all countres and regons mplyng that unobservable country-specfc effects exsted n all estmatons. Therefore the FE model was the most approprate to apply n our estmaton The results from the FE regresson ndcated that all external factors had a posve mpact on agrcultural growth (Table 5). The contrbuton of Japanese FDI to growth also showed a posve mpact; however, the mpact was much smaller than that of mports, exports and trade. Ths outcome may have been due to the fact that the characterstcs of the agrcultural sector, land and ntensve labor, were the man constrants that meant that the benefs from FDI va technology transfer 4 were lower n the agrcultural sector than n the manufacturng sector. Also we found that the Asan economc crss n 1997 had a negatve mpact on agrcultural growth n East Asa. In AGRIBASE countres the results ndcated that the mpact of nternal factors such as land and labor were not consstent. However, the mode that ncluded exports can help to explan the mpact of these varables on growth. Exports can help to mprove productvy of land and labor drectly as dscussed n secton 2, thus resultng n a posve mpact on growth. Snce agrcultural producton requres land and labor ntensve, a posve mpact would be expected n these countres where land and abundant labor preval. Imports and exports have a very posve mpact on agrcultural growth. The export-led growth hypothess that was confrmed agan n ths study was an expected outcome, snce these countres are mostly agrcultural exporters and have a comparatve advantage n ths sector. However, the mport-led growth hypothess was also found to be true n these countres. Ths evdence strongly suggests that f these countres only promote exports and lower mports e.g., mport substuton polcy, ths would lower growth. To obtan growth n agrcultural GDP, trade needs to be encouraged n both drectons. Moreover, the trade varable whch s a representaton of trade lberalzaton clearly ndcated that more 4 For more detal of technology transfer n agrcultural sector and the adaptaton of agrcultural technology and nvestment see Hayam and Ruttan (1985). 10

11 trade and exchange of agrcultural commodes n these countres would be benefcal n helpng accelerate economc growth. Japanese FDI also has a posve mpact on growth n these sectors, although s small. As dscussed n the theoretcal secton, unlke trade whch has drect mpact on growth, FDI seems to have a complex effect on growth, and the agrcultural sector whch seems to be more constraned n absorbng benefs drectly from FDI, would only recevng a small contrbuton from FDI. Even though there s some concern about the small amount of FDI contrbuton to the agrcultural sector, these results mply that Japanese FDI has played a sgnfcant role n promotng growth. In NON-AGRIBASE countres, agrcultural GDP growth was strongly related to the avalably of agrcultural land, wh a 1 per cent ncrease n land ncreasng agrcultural growth by up to 3 per cent. Ths strong relatonshp may due to the fact that land used for agrculture has become scare for these countres, where alternatve use for land may gve hgher benefs, thus resultng n a gradual decrease n land avalable for agrcultural producton. In contrast, labor had a negatve mpact on growth, possble due to export effects, snce these countres are mostly mporters of agrcultural products rather than exporters. Consequently labor s unable to gan any mproved productvy va exports causng the negatve mpact of labor on growth. Therefore, the export-led growth hypothess probably does not work well n these countres. However, mports had a strong posve mpact on growth, as a 1 per cent ncrease n mports lead to a 0.8 per cent ncrease n agrcultural growth. These results mght have been expected because these NON- AGRIBASE countres economes are mostly agrcultural mporters and so do not rely much on agrculture. Consequently, trade varable also ndcated that more trade lberalzaton n these countres could help stmulate agrcultural growth. The mpact of Japanese FDI on ther agrcultural growth was mldly posve, but was not sgnfcant. Therefore appears that the effect of Japanese FDI vares between countres/regons. 11

12 In AGRIBASE countes, Japanese FDI n the agrcultural sector was manly vertcal nvestment (Sattaphon and Kmnam, 2006). In these countres Japanese MNE undertook FDI to establsh producton base because of the abundant resources avalable and exported most of the products back to the Japanese home market. Any remanng goods were sold to local markets or exported to thrd countres. Therefore, FDI mproved the host economy growth va local employment, technologcal transfer and other spll-over effects. In contrast, FDI n NON- AGRIBASE countres has been mostly horzontal nvestment n order to access to ther local markets, and therefore, ths FDI has had ltle economc mpact. Local employment has been ncrease, but there was no ncrease n agrcultural productvy va technology transfer, snce ths knd of FDI competes wh local producton. 6. Concluson Remarks Ths study analyzed the mpact of nternatonal trade and Japanese foregn drect nvestment (FDI) on growth n the agrcultural sector of East Asan countres by applyng the panel data econometrc approach. The results ndcated that, n the case of Chna, Malaysa, Indonesa, Phlppnes, and Thaland, exports are the most mportant factor contrbutng to growth. Moreover, mport-led growth was also detected n these countres suggestng that mportng capal or ntermedate goods from abroad also helps to accelerate agrcultural growth. Japanese FDI also had a posve mpact on stmulatng the growth process, but the effect was not large. In contrast, n Korea and Tawan, agrcultural growth was greatly affected by ncreased agrcultural land use. Imports clearly had a more posve mpact n stmulatng agrcultural growth than exports. However, there was only a slght posve mpact of Japanese FDI on growth n these countres. The fndng of ths study wll help polcy makers to understand the lkely mplcatons of possble future polces. The agrcultural sectors of countres cannot rely solely on nternal 12

13 factors such as land and labor to promote agrcultural growth. Agrcultural growth requres more trade lberalzaton, especally when mports are an mportant factor for promotng growth. Attemptng to lower mports e.g., mport substuton polcy would lead to retarded growth. Meanwhle, approprate polces to attract Japanese FDI nto the agrcultural sector are also mportant n order to help accelerate growth n ths sector. Acknowledgement The Author would lke to thanks Prof. Kawamura and Prof. Ndara for ther comments on the frst draft of ths paper whch was presented at the Japan Secton of Regonal Scences Assocaton Internatonal (JSRSAI), 8-10 September 2005, Tottor Unversy, Japan. And also, the authors are ndebted to Dr. Kmnam for hs valuable comments and supports, and Dr. Taro for hs techncal assstance. 7. References Balasubramanyam V.N. et al., Foregn Drect Investment and Growth n EP and IS countres, Econ. J. 106, Borensten E. et al., How does foregn drect nvestment affect economc growth, J. Int. Econ. 45, Buckley P. J. et al., FDI, regonal dfferences and economc growth: panel data evdence from Chna, Transnat. Corporatons 11, 1-28 Carkovc M. et al., Does foregn Drect Investment Accelerate Economc Growth? In: Moran T. H. et al.(ed.), Does Foregn Drect Investment Promote Development?, Instute For Internatonal Economcs, pp Choong C. K. et al., Foregn Drect Investment, Economc Growth, and Fnancal Sector Development: A Comparatve Analyss, ASEAN Econ. Bull. 21, de Mello L.R., Jr., Foregn drect nvestment-led growth: evdence from tme seres and panel data, Oxford Econ. Pap. 51, Edwards S., Openess, Trade Lberalzaton, and Growth n Developng Countres, J. Econ. L. 31, Esfahan H.S., Exports, mports, and economcs growth n sem-ndustralzed countres, J. Devel. Econ. 35,

14 Frankel J.A., Trade and Growth n East Asa Countres: Cause and Effect? NBER workng paper Ghrmay T. et al., Export, Investment, Effcency and Economc growth n LCD: an emprcal nvestgaton, Appl. Econ. 33, Hayam Y., Ruttan V.W., Agrcultural Development: An Internatonal Perspectve, The John Hopkns Unversy Press. Hsao C., Analyss of Panel Data second edon, Cambrdge Unversy Press. Im K.S. et al., Testng for un root n heterogeneous panels, J. Econometrcs 115, Jn C.J., Export and growth: s the export-led growth hypothess vald for provncal economes? Appl. Econ. 34, Kavouss R. M., Export Expanson and Economc Growth: Further Emprcal Evdence, J. Devel. Econ. 14, Kosempel S., A theory of development and long run growth, J. Devel. Econ. 75, Krshna et al., Trade, nvestment and growth: nexus, analyss and prognoss, J. Devel. Econ. 70, Lee J-W., Capals goods mports and long-run growth, J. Devel. Econ. 48, Pedron P., Panel Contegraton: Asymtotc and Fne Sample Propertes of Pooled Tme Seres Test wh an Applcaton to the PPP Hypothess, Econometrc Theory, 20, Pedron P., Crcal Values for Contegraton Tests n Heterogeneous Panel wh Multple Regressors, Oxford Bull. Econ. Statst. 61, Sattaphon W., Kmnam A., The Effects of Japanese FDI on Blateral Food Trade wh East Asan Countres: An Applcaton of Panel Data Econometrcs, Studes n Regonal Scence. 35 (Forthcomng) Tremblay R., The Export-Import Effects and Economc Growth, N. Amer. Rev. Econ. Fnance. 1, Tyler W.G., Growth and Export Expanson n Developng Countres, J. Devel. Econ. 9, Urata S., Kawa H., Intrafrm Technology Transfer by Japanese Manufacturng Frm n Asa, In: Ito T. and Krueger A.O. (Ed.) The role of foregn drect nvestment n East Asan economc development, Unversy of Chcago Press, pp Walde, K., Wood, C., The emprcs of trade and growth: where are the polcy recommendatons? Int. Econ. Econ. Pol. 14

15 Table 1: Agrcultural GDP and labor force share of the total economy (percentage) Year Country GDP LABOR GDP LABOR GDP LABOR Chna Indonesa Malaysa Phlppnes Thaland Korea Tawan Source: Author s own calculaton based on Key Indcators of Developng Asan and Pacfc Countres, varous ssues, Asan Development Bank (ADB). Table 2: Dstrbuton of Japanese FDI durng (1000 US$) Sector Country Food Fshery Agrculture&Forestry Total CASE VALUE CASE VALUE CASE VALUE CASE VALUE Chna Indonesa Malaysa Phlppnes Thaland Korea Tawan Total Source: Fnancal Statstcs Monthly, varous ssues, Mnstry of Fnance, Japan. Table 3: IPS Un Root Test Varable W-tbar wh Trend W-tbar whout Trend Level I(0) Frst Dff I(1) Level I(0) Frst Dff I(1) LAGDP *** *** LLAND *** LLABOR *** *** LEXP *** *** LIMP *** *** LTRADE *** *** LFDI *** *** *** *** Note: 1) *** ndcated sgnfcant from zero at 1% 2) Crcal values for W-tbar wh trend at 1%, 5% and 10% are -2.88, and respectvely. 3) Crcal values for W-tbar whout trend at 1%, 5% and 10% are -2.24, and respectvely. Table 4: Pedron Panel Contegraton Test Panel Contegraton Test Test statstc (x) Panel v-statstc Panel rho-statstc Panel pp-statstc Panel adf-statstc Group rho-statstc Group pp-statstc Group adf-statstc Note: The Pedron (1999) statstcs are one-sded tests wh a crcal value of 1.65 (x < mples rejecton of the null), except the v-statstc that has a crcal value of 1.65 (x > 1.65 suggests rejecton of the null).the test nclude a constant and heterogeneous tme trend n the data. 15

16 Table 5: The Fxed effects (FE) regresson results of ndvdual varables on the agrcultural growth functon Explanatory ALL AGRIBASE NON-AGRIBASE varables (1) (2) (3) (4) (1) (2) (3) (4) (1) (2) (3) (4) Land c a c b b a (0.374) (0.387) (0.322) (0.584) (0.412) (0.471) (0.386) (0.706) (1.235) (0.994) (0.968) (1.149) Labor b a (0.144) (0.152) (0.127) (0.212) (0.232) (0.252) (0.212) (0.362) (0.311) (0.302) (0.265) (0.331) Exports c c (0.070) (0.078) (0.219) Imports Trade FDI Crss c (0.042) c (0.047) c (0.035) c (0.065) c (0.033) c (0.031) b (0.046) c (0.044) c (0.530) c (0.042) c (0.062) c (0.038) b (0.037) c (0.594) c (0.091) c (0.216) c (0.424) c (0.243) a (0.101) (0.064) c (0.070) Homogeney F-test c c c c c c c c c c c c R-square F-test c c c c c c c c c c c c No. obs Note: 1) All has ntercept value and standard errors are n parentheses. 2) a, b and c ndcate that the coeffcent s sgnfcantly dfferent from zero at 10%, 5% and 1% levels, respectvely. 3) AGRIBASE comprses Chna, Malaysa, Indonesa, Phlppnes, Thaland and NON-AGBASE comprse Korea and Tawan 16