LIST OF TABLES AND FIGURES

Transcription

1

2

3 ACKNOWLEDGMENTS The authos ae gateful to Mchael Johnson fo hs contbutons n many dscussons that helped fomalze the stuctue of the pape. The authos also thank two anonymous efeees fo the helpful comments and suggestons and Phylls Skllman fo Englsh edtng.

4 TABLE OF CONTENTS ACKNOWLEDGMENTS...2 TABLE OF CONTENTS...4 LIST OF TABLES AND FIGURES...5 ABSTRACT...6 I. INTRODUCTION...1 II. AGRICULTURE IN THE SOUTHERN AFRICAN ECONOMY...5 The Role of Agcultue n the Economy: Low- vesus Mddle-Income Goups...5 Dsappontng 20 Yeas fo Southen Afca s Agcultual Pefomance...9 Oppotuntes to Expand Regonal Tade...11 Regonal Agcultual Gowth Oppotuntes...18 III. ANALYZING GROWTH LINKAGES IN SOUTHERN AFRICA...23 A Regonal Geneal Equlbum Model fo Southen Afca...23 The Model and Data Descpton...23 Smulaton Scenaos...27 Altenatve Gowth Scenaos fo Southen Afca s Agcultue...28 Agcultue n Low-Income Countes Benefts fom Gowth n South Afca...28 Agcultue Has Stong Gowth Lnkages to Nonagcultue...33 Gowth n Mddle-Income Countes Can Help Low-Income Countes Ovecome the Domestc Demand Constants fo Gans...35 IV. CONCLUSION...39 REFERENCES...42 APPENDIX A. Supplemetay Tables...46 APPENDIX B. Mathematc Pesentaton of the Regonal CGE Model...53 LIST OF DSGD DISCUSSION PAPERS

5 Tables LIST OF TABLES AND FIGURES 1. Income and Povety fo Southen Afcan Countes Gowth Decomposton by Secto n the Low Income Southen Afcan Countes (Aveage ) Intaegonal Agcultual Tade n Southen Afca (US$ mllon) Southen Afcan Countes Agcultual Expot Intensty n Dffeent Makets Numbe of Matches Between a County wth Compaatve Advantage and a County wth Compaatve Dsadvantage fo a Smla Agcultual Commodty aveage Land Poductvty n Low-ncome Southen Afca Compaed to Land Poductvty n South Afca (n klogams/hectae) CGE Model Smulaton Scenaos Aggegate Effect of CGE Model Smulatons Effects on Agcultual Subsectos of CGE Model Smulatons Gowth n Nontadtonal Expots n Scenao Fgues 1. Expots fom Southen Afcan Countes to Dffeent Destnatons (US$ mllon cuent pces) Shaes of Destnaton Regons n Southen Afcan Agcultual Expots Numbe of Matches Between Southen Afcan Countes wth Compaatve Advantage and Dsadvantage fo a Smla Commodty ( Aveage)

6 ABSTRACT Consdeng the heteogenety of the countes of southen Afca and the pesence of South Afca and othe mddle-ncome countes n the egon southen Afca has a unque oppotunty to explot agcultual potental and egonal tade oppotuntes though egonal dynamcs and ntegaton. We analyze the mplcatons of such oppotuntes fo the gowth of the low-ncome countes usng a egonal geneal equlbum model that captues gowth lnkages. We fnd that gowth n the mddlencome southen Afcan countes such as South Afca benefts the egon s lowncome countes though nceased demand fo the agcultual expots. Agcultual poductvty gowth howeve s necessay fo low-ncome countes to take advantage of South Afca s gowth. Poductvty gowth n the low-ncome countes gan and lvestock sectos geneates moe gowth n GDP and food consumpton than gowth n nontadtonal expot cops. Unlke othe egons whee gowth n gan poducton s lkely to be constaned by domestc demand expandng mddle-ncome economes n southen Afca povde addtonal demand fo gans and lvestock slowng the declne n gan pces n the egon. 6

7 v

8 EXPLORING GROWTH LINKAGES AND MARKET OPPORTUNITIES FOR AGRICULTURE IN SOUTHERN AFRICA Alejando Nn Patt and Xnshen Dao 1 I. INTRODUCTION Stengthenng egonal economc lnkages that offe mutual benefts acoss countes s an mpotant pat of development stateges n Sub-Sahaan Afca leadng to economc gowth and povety educton. Regonalsm n fact has eceved nceasng attenton as a esult of gowng feas n Afca and n the ntenatonal communty of Afcan magnalzaton n the global economy. As a esult seveal egonal ntatves have been developed acoss the contnent n patcula n southen Afca. The need fo the ceaton of nsttutonal famewoks and pogams that can mpove food secuty n the egon has been cental to coopeaton effots though egonal schemes such as the Common Maket fo Easten and Southen Afca (COMESA) the Southen Afca Development Communty (SADC) and the Southen Afca Custom Unon (SACU). 2 Effots by SADC and COMESA to establsh a fee tade aea (FTA) and customs unons ae all steps n movng towad an economc aea that ultmately allows the fee movement of people goods and sevces as well as factos of poducton (captal and labo). Both the SADC and COMESA schemes have ted to addess ctcal ssues such as emoval of taff and nontaff baes; development of ules of ogn; coopeaton n customs admnstaton techncal standads and santay and phytosantay standads; and pomoton of coss-bode nvestment. 1 Alejando Nn Patt s a Reseach Fellow and Xnshen Dao s Seno Reseach Fellow of IFPRI s Development Stategy and Govenance Dvson. 2 COMESA membe countes ae Angola Buund Comoos the Democatc Republc of Congo (DRC) Djbout Egypt Etea Ethopa Kenya Lbya Madagasca Malaw Mautus Rwanda Seychelles Sudan Swazland Uganda Zamba and Zmbabwe. SADC membe countes ae Angola Botswana DRC Lesotho Madagasca Malaw Mautus Mozambque Namba South Afca Swazland Tanzana Zamba and Zmbabwe. SACU membe countes ae Botswana Lesotho Namba South Afca and Swazland. 1

9 Pogess has also been made n mpovng the egon s oad netwok. Though SADC s Tanspot Communcaton and Meteoology Potocol fo example membe countes ae hamonzng oad desgn standads; adoptng standadzed oad taffc sgns dves tanng manuals and engneeng specfcatons fo oad and bdge constucton; and ehabltatng majo tanspot codos such as the Maputo Development Codo and the Tans-Kalaha Codo. 3 Investments n the egon have also been gowng as South Afcan sevce and manufactung companes ncludng supemaket chans lke Shopte and Pck n Pay have expanded to othe countes n the egon. Despte the pogess beng made the egon s stll a long way fom takng full advantage of the oppotuntes to futhe ntegate and stmulate economc development. Food defcts ae stll an ssue n the low-ncome countes n the egon wth poductvty of ceeal poducton stll below the Afcan aveage. Inceased nvestments ae needed f countes ae to eap geate benefts fom FTAs and to futhe the ntegaton pocess. Foegn dect nvestment (FDI) nflows have lagged because ndvdual countes have small makets weak nfastuctue and unfavoable nvestment clmates caused by cvl was poltcal unest and cuency nstablty. Pooly functonng makets also lead to hgh tansacton costs. Although southen Afca has well-establshed tanspotaton codos tanspot costs eman excessvely hgh: they have been estmated to be pecent of valued added of goods sold n the subegon (Ktznge-van Nekek and Moea 2002). Seveal studes consstently show that hgh tanspotaton costs act as a estctve bae to nceased tade and captal flows (Yeats and Amjad 1999; Busse 2003; Amjad and Yeats 1995). In ths context the key queston s what oppotuntes do egonal ntegaton and coopeaton offe southen Afcan countes fo gowth and povety allevaton? Thee ae at least fou aeas n whch southen Afcan countes can beneft fom egonal ntegaton and coopeaton: (1) the economc dvesty of the egon (2) egonal food secuty (3) egonal nfastuctue and (4) tade and nvestment. Othe ssues lke 3 SADC dves lcenses have been ssued n eght countes. 2

10 HIV/AIDS and amed and poltcal stfe ae also aeas whee ntegaton could play an mpotant ole. Ths study focuses on the economc lnkages between mddle- and lowncome countes gven that southen Afca s the only egon on the contnent whee thee ae a numbe of mddle- and low-ncome countes n close poxmty to each othe. Dffeences n ncome level often epesent dffeences n development stages. Thus southen Afca s economc dvesty s geneally vewed as a key eason fo pomotng geate egonal ntegaton fo stmulatng gowth and povety educton. Pe capta ncomes n the egon vay wdely and benefts fom geate egonal ntegaton ae expected to come fom the natual ole South Afca can play as an engne of gowth fo the egon both n tems of povdng a dynamc maket fo egonal expots and a souce of nvestment and technology dffuson. Many countes n southen Afca have elatvely small agcultual sectos ethe because the economes ae moe advanced and dvesfed o because they have a hgh dependency on mneal esouces: Angola Botswana Namba South Afca Swazland. Agcultue howeve emans the pmay souce of employment and ncome n the egon s low-ncome countes Malaw Mozambque Zamba and Zmbabwe. In these countes povety and hunge ae stll pedomnantly ual phenomena. Most southen Afcan countes stll have unexploted agcultual potental especally Angola nothen Zamba nothen Mozambque and Zmbabwe. Combnng ths potental wth the po-poo featue of agcultual-led gowth suggests that agcultue can play a cental ole n educng povety n the egon. Gven that some natonal nvestments wll geneate postve extenaltes and splloves to the neghbong countes nceased effcency gans could be obtaned fom egonal nvestment stateges especally fom nvestments n eseach and development. Hee we analyze the economc lnkages n southen Afca and the mplcatons of such lnkages fo the gowth of low-ncome countes usng a egonal geneal equlbum model developed fo ths study. We fnd that gowth n the mddle-ncome southen Afcan countes benefts the low-ncome countes n the egon though nceased demand fo the agcultual expots. Agcultual poductvty gowth 3

11 howeve s necessay fo low-ncome countes to take advantage of South Afca s beng a gowth pole fo the egon. Poductvty gowth n the low-ncome countes gan and lvestock sectos geneates moe gowth n goss domestc poduct (GDP) and food consumpton than gowth n the nontadtonal expot cops. Unlke othe egons whee gowth n gan poducton s lkely to be constaned by lmted domestc demand gowng mddle-ncome economes n southen Afca povde addtonal demand fo gans and lvestock slowng the declne of gan pces n the egon. The est of ths pape s oganzed as follows: secton 2 focuses on the chaactestcs of southen Afcan economes. We estmate a measue of compaatve advantage n tade and use t to detemne tade complementaty between countes n the egon n ode to assess the potental fo expandng egonal tade. Secton 3 pesents the egonal computable geneal equlbum (CGE) model and the model smulaton esults focusng on dffeent subsectos potental contbutons to food secuty economy-wde gowth and tade expanson. Secton 4 povdes ecommendatons and conclusons of the study. Appendx A compses a set of supplementay tables and Appendx B pesents the vaables and equatons that make up the CGE model. 4

12 II. AGRICULTURE IN THE SOUTHERN AFRICAN ECONOMY The Role of Agcultue n the Economy: Low- vesus Mddle-Income Goups The theoetcal and empcal lteatue suggests that the ole of agcultue n the economy s hghly elated to a county s stage of development (Johnston and Mello 1961; Block and Tmme 1994; Kydd et al. 2004; Hazell 2005). Usng pe capta ncome as a poxy fo development the 11 southen Afcan countes can be classfed nto two goups (Table 1): sx countes belong to the mddle-ncome goup and the emanng fve ae n the low-ncome goup. Accodng to the Wold Bank defnton annual GDP pe capta n the mddle-ncome goup was moe than $735 n Of the fve countes n the low-ncome goup two actually moved down fom beng mddle-ncome countes n the ealy 1980s (Zamba and Zmbabwe). As mddle-ncome countes account fo moe than 40 pecent of southen Afca s total populaton the egon as a whole had aveage annual pe capta ncome of $1510 n 2002 much hghe than that of othe Sub- Sahaan Afcan countes (many of whch have pe capta ncome below $300). The agcultue secto accounts fo only 3 pecent of total GDP fo the egon s mddlencome countes as a goup but accounts fo 20 pecent of total low-ncome countes GDP. Thee s only one county Malaw n whch agcultue s 34 pecent shae n GDP s above the aveage (31 pecent) fo all low-ncome Sub-Sahaan Afcan countes as a goup. Despte elatvely small agcultual sectos most southen Afcan countes have lage ual populatons accountng fo 48 pecent of populaton n mddle-ncome countes and 68 pecent n low-ncome countes. Moeove the povety ate s just as hgh as n othe Sub-Sahaan Afcan countes ncludng n mddle-ncome southen Afcan countes such as Botswana Namba and Swazland. In these countes a vast majoty of the poo lve n ual aeas and ae dependent on agcultual ncomes. Although Swazland has dvesfed ts manufactung secto snce the md-1980s wth suga and wood pulp now man foegn exchange eanes subsstence agcultue stll occupes moe than 80 pecent of ts populaton wth fames facng poblems of 5

13 ovegazng sol depleton and dought. In the case of Namba the economy s heavly dependent on the extacton and pocessng of mneals fo expot (damonds uanum lead znc tn slve and tungsten) but the mnng secto employs only about 3 pecent of the natonal labo foce. Seventy-fve pecent of ts people depend on low-poductvty subsstence agcultue cash tansfe pensons and wage ncome on commecal fams fo the lvelhoods (Stone and Gaomab 1994). In Botswana unemployment offcally stands at 24 pecent but unoffcal estmates place t close to 40 pecent despte the county s hgh economc gowth ates snce ndependence n 1966 (CIA 2006). Theefoe whle agcultue may not be a domnant secto n the egon t stll plays an mpotant ole n educng povety. Table 1. Income and Povety fo Southen Afcan Countes County GDP Pe Capta c Rual Populaton c Povety Head Count a AgGDP c US$ (%) (%) Yea (%) Mddle-ncome countes b Mautus Botswana South Afca Namba Swazland Angola Low-ncome countes b Lesotho Zmbabwe Zamba Mozambque Malaw Southen Afca Sub-Sahaan Afca Sub-Sahaan Afca not - ncludng Southen Afca a Povety headcount ato at $1 a day (PPP) (% of populaton). Povety head count fo Swazland s fom FAOSTAT Food Secuty Statstcs. b Weghted aveages. Low-ncome countes ae Lesotho Malaw Mozambque Zamba and Zmbabwe. Mddle-ncome countes ae Angola Botswana Mautus Namba South Afca and Swazland. c Yea 2002 Souce: Wold Bank Wold Development Indcatos

14 To bette undestand the ole of agcultue n the egon t s necessay to dstngush among countes accodng to a ange of ndcatos that eflect agcultual potental and altenatve souces of gowth. Agcultual potental daws on a classfcatoy scheme developed by Dxon Gullve and Gbbon (2001) whch ncludes measues such as ago-ecologcal condtons and populaton denstes. Accodng to these ndcatos all fve low-ncome southen Afcan countes have agcultual potental. Howeve even n countes whee condtons ae favoable agcultue competes wth othe sectos fo lmted esouces. Countes wth ch mneal o ol endowments may have altenatve souces of gowth. And coastal countes may have advantages n expotoented agcultue o geate oppotuntes n nonagcultue. Theefoe we wll dscuss the fve low-ncome countes accodng to whethe they ae coastal land-locked o mneal-ch. Mozambque s the only coastal county among the fve low-ncome southen Afcan countes. The county has elatvely favoable agcultual condtons and few natual baes to tade. Whle coastal countes may have bette potental fo expot-led agcultual gowth oppotuntes fom nonagcultual sectos may ceate altenatve gowth optons. Indeed Mozambque s GDP expanded at an annual ate of 5.7 pecent between 1985 and 2002 wth gowth dven by both agcultual and nonagcultual sectos (Table 2). Howeve the county s one of the two pooest southen Afcan countes wth annual pe capta ncome below $250. Moe than 60 pecent of Mozambque s populaton lves n ual aeas and most of the poo depend on agcultue fo the lvng. Hence Mozambque needs a gowng agcultual secto to sustan gowth. Whle a 5.3 pecent ate of annual gowth n agcultue between 1985 and 2002 s hghe than most countes n the egon o n Sub-Sahaan Afca agcultual gowth seems to have slowed down n ecent yeas. Ou analyss ncludes thee land-locked low-ncome countes: Lesotho Malaw and Zmbabwe. Whle beng land-locked can epesent a sgnfcant natual bae to tade and undemne expot oppotuntes ntegaton wth neghbong countes can actually ovecome such baes. As the pooest county n the egon Malaw has the 7

15 hghest agcultual GDP shae (34 pecent) and ual populaton shae (85 pecent). Moeove agcultual gowth s the man dve fo the oveall economc gowth. Extemely low gowth n the nonagcultual sectos dung esulted n an annual gowth ate of GDP of only 2.36 pecent and neglgble gowth n pe capta tems (Table 2). Table 2. Gowth Decomposton by Secto n the Low Income Southen Afcan Countes (Aveage ) Shae n GDP n 1985 (%) Gowth Rate (%) Contbuton to GDP Gowth (%) County Agcultue Industy Sevces Agcultue Industy Sevces GDP Agcultue Industy Sevces Low-ncome countes a Coastal Mozambque Land-locked Lesotho Malaw Zmbabwe Mneal-based Zamba a/ Weghted aveages. Low-ncome countes ae Lesotho Malaw Mozambque Zamba and Zmbabwe. Souce: Wold Bank. Wold Development Indcatos 2005 The othe two land-locked countes Lesotho and Zmbabwe togethe wth the mneal-ch county Zamba have evolved n dffeent ways. Although Lesotho s economy s stll pmaly based on subsstence agcultue especally lvestock t has developed a small manufactung secto based on fam poducts and a apdly expandng appael assembly secto. The latte has gown sgnfcantly manly because Lesotho qualfes fo tade benefts unde the Afca Gowth and Oppotunty Act (Lesotho Kngdom of 2006). Despte the eale status as mddle-ncome countes Zmbabwe and Zamba s agcultual sectos and economes as a whole have pefomed the wost fo vaous easons manly poltcal nstablty and conflcts. Snce the agcultual GDP of 8

16 these two countes accounts fo almost 20 pecent of the egon s total the poo pefomance of the agcultual sectos has advesely affected southen Afca s total agcultual gowth. The contbuton of agcultue to total GDP gowth n these countes has esulted n a low 2.7 pecent gowth of annual agcultual GDP fo the lowncome southen Afcan countes as a goup. Howeve Malaw and Mozambque had annual GDP gowth ates of 3.6 and 5.3 pecent espectvely dung ths peod. Dsappontng 20 Yeas fo Southen Afca s Agcultual Pefomance In southen Afca food staple poducton s the domnant agcultual actvty. Moe than 50 pecent of agcultual land s allocated to ceeals whle maze alone accounts fo moe than 40 pecent of the total havested aea (see Appendx Table A.1). Roots and tubes cuently account fo 8 pecent of the total cop aea fo the low-ncome countes as a goup. In total the staple cops occupy almost 66 pecent of cop land n the low-ncome goup leavng just ove 30 pecent of land fo othe cops manly tadtonal expotables such as cotton (7 pecent); tobacco tea coffee spces and suga (5 pecent); olseeds (10 pecent); and futs and vegetables and pulses (11 pecent). The composton of the anmal stock shows that moe than 70 pecent of the anmals ae beef and day cattle n both low-and mddle-ncome countes. The shae of poulty n total anmal stock gew steadly between 1985 and 2002; chcken cuently epesents 10 pecent of the total anmal stock n the egon compaed wth only 2 pecent n Revenue fom cops epesents two-thds of egonal agcultual evenue wth mddle-ncome countes poducng almost 65 pecent of ceeals 80 pecent of futs and vegetables and moe than 80 pecent of beef and poulty meat (Appendx Table A.2). The egon poduced 2.6 mllon tons of meat and 3.6 mllon tons of mlk n pecent of whch was poduced by South Afca. But low-ncome countes poduce 60 pecent of oots and tubes and 80 pecent of tadtonal expotable cops (tobacco coffee and tea). 9

17 Although the egon allocates 50 pecent of agcultual land to ceeal poducton southen Afca as a whole has become a gan-defct egon n ecent yeas. Ceeal mpots nceased fom 12 pecent of ceeal demand n to 22 pecent n wth a gap between demand and poducton of 20 pecent n low-ncome countes and 17 pecent n mddle-ncome countes (Appendx Tables A.3 and A.4). Moeove 10 of the 11 southen Afcan countes all except South Afca cuently ae maze-defct countes and the defcts n the fve low-ncome countes anged fom 42 pecent of domestc consumpton n Lesotho to 6 pecent n Zmbabwe between 1998 and Stagnant poductvty gowth n agcultue s the man facto that caused the egon to become dependent on food mpots. Compaed wth 1981 the land aea allocated to maze n 2002 nceased by moe than 30 pecent n the low-ncome county goup. Although land allocated to ceeal and maze poducton fell n the mddle-ncome goup the egon s total aea allocated to ceeals stll nceased n ths peod. Despte ths total ceeal as well as maze poducton deceased (Appendx Table A.5). Wth stong populaton gowth thoughout the egon and nceased pe capta ncome n some mddle-ncome countes food demand fo ceeals has nceased by 50 pecent n the past 20 yeas. These two factos wokng togethe have shfted the egon fom a gan suplus n the ealy 1980s to a gan defct n ecent yeas. In addton to the defcts n ceeal supply food secuty s unde pessue fom nceasng populatons. When govenment suppot polces fo maze and othe ceeals wee emoved afte the mplementaton of the stuctual adjustment pogams n the late 1980s oot and tube poducton n the low-ncome southen Afcan countes nceased. Moeove unlke the ceeal secto poductvty gowth n oots and tubes seems to be qute successful n many southen Afcan countes. Whle total poducton aea of oots and tubes nceased by 50 pecent ove the last 20 yeas the output nceased by moe than 150 pecent n the same peod whch sgnfcantly contbuted to the food secuty of many poo ual households. The lvestock secto has pefomed bette than the gan secto n the egon. Compaed wth the aveage fo the egon s total meat poducton has nceased 10

18 by 1.92 pecent pe yea on aveage ove the past 20 yeas. Howeve demand has gown moe apdly at 2.57 pecent pe yea n the same peod. Thus the egon has shfted fom a meat suplus n the ealy 1980s (wth net expots accountng fo moe than 6 pecent of total poducton n ) to a defct (Appendx Table A.6). In the lowncome goup 14 pecent of the mlk consumed and 6 pecent of the poulty meat consumed s mpoted evey yea. Fo the egon as a whole 11 pecent of the meat and 17 pecent of the mlk consumed s mpoted ( aveage). Oppotuntes to Expand Regonal Tade As mentoned eale egonal schemes to foste coopeaton among southen Afcan countes such as COMESA SADC and SACU have placed geat mpotance on ntegaton n the egon s development stategy. In ths context emoval of taffs s an mpotant ssue n the egon because taffs affect tade between mddle- and lowncome countes that do not belong to SACU (such as Malaw Mozambque Zamba and Zmbabwe). On the one hand South Afca mposes hgh taffs on mpots of day poducts ceeals and textles sectos wth potental fo low-ncome countes n the egon to ncease expots. On the othe hand the low-ncome countes mpose hgh taffs on textles futs vegetables and pocessed food poducts sectos wth potental fo ntaegonal tade (Appendx Table A.7). 4 The elmnaton of agcultual taffs among SADC countes would beneft the egon n tems of eal agcultual GDP natonal ncome and agcultual output (see fo example Dao and Robnson 2003; Kang Swadana and Ronge 2002). Howeve taffs ae not the only obstacle to nceased egonal tade. The analyss of ntegaton n southen Afca goes beyond tade lbealzaton; to explan low tade n the egon seveal studes have stessed the mpotance of tanspot and tansacton costs and the lack of dvesfcaton n compaatve advantages (see fo example Chauvn and Gaule 2002; Cassm 2000; Daves 1996; Geda and Kbet 2002; Goldsten 2004; Holden 1996; Jenkns Leape and Thomas 2000; Longo and Sekkat 2001; Radelet 4 Zamba s an excepton wth lowe taffs on these poducts than othe low-ncome countes n the egon. 11

19 1997). Ths study depats fom pevous analyss focusng on egonal economc lnkages and the mplcatons of such lnkages fo the gowth of low-ncome countes. Although we ecognze the mpact of tade and nvestment polces on poductvty and economc gowth a moe sophstcated ntetempoal dynamc model s needed to fully take nto account the endogenous lnkages between these polces and economy wde gowth at the ndvdual county level (see fo example Dao 2001). The development of ths model s beyond the scope of ths study and wll be the focus of authos futue eseach effots. Fgue 1. Expots fom Southen Afcan Countes to Dffeent Destnatons (US$ mllon cuent pces) EU SSA Othe developng Regon Souce: Authos calculatons fom COMTRADE data. Fgue 2. Shaes of Destnaton Regons n Southen Afcan Agcultual Expots Aveage southen Afca 7% Aveage southen Afca 11% Othe developng 7% SSA 1% Othe developng 14% Othe developed 20% SSA 4% EU 52% EU 65% Othe developed 19% Souce: Authos calculatons fom COMTRADE data. 12

20 In ths secton we focus on the ecent evoluton of egonal tade usng hstocal data to analyze compaatve advantage and tade complementaty and assess the potental to expand egonal tade n southen Afca. Regonal tade saw a sgnfcant expanson dung the 1990s (Fgues 1 and 2). Whle total agcultual expots fom the egon expanded at a ate of 7.5 pecent a yea nta-southen Afcan expots gew by 13 pecent annually between 1990 and 1999 esultng n nceased ntaegonal tade shaes fo agcultual commodtes (sng fom 7 pecent n 1990 to 11 pecent n 1999). Whle Oganzaton fo Economc Coopeaton and Development (OECD) countes ae stll the most mpotant tade patnes of southen Afca a new tend seems to be developng wheeby southen Afcan expots ae shftng to the makets n developng countes ncludng Asan makets and egonal makets n southen Afca and n Sub-Sahaan Afca n geneal. The shae of OECD countes n the egon s total expots fell to 70 pecent n 1999 fom 85 pecent n The expanson of egonal tade s assocated wth South Afca s nceasng nvolvement n the egon snce the county was eadmtted to the global economc communty n Snce then South Afca has been an actve nvesto n all SADC countes accountng fo 25 pecent of total foegn dect nvestment (FDI) flowng nto the SADC egon (Rumney and Pngo 2004). South Afca has also nceased ts tade wth ts neghbos snce About 75 pecent of egonal expot expanson s explaned by nceased expots fom South Afca 5 whle Mozambque Zmbabwe and Zamba togethe explan the emanng 30 pecent (Table 3). On the mpot sde only 9 pecent of mpot gowth s explaned by South Afca. Mozambque Zmbabwe Zamba and Angola explan almost 80 pecent of the ncease n mpots. Whle SACU sgnfcantly expanded net expots to the egon othe expotng countes lke Mozambque Zamba and Zmbabwe expeenced a educton n net expots to the egon. In 1990 South Afca was a net mpote n the egon (wth net mpots of US$58 mllon). By the end of the decade South Afca had become a net expote to the egon wth US$317 mllons of net expots whle all othe countes saw lage nceases n the mpots fom South Afca. In patcula Zmbabwe whch was 5 No dsaggegated data of tade of SACU countes ae avalable but SACU tade n the egon s manly explaned by South Afca 13

21 the only net expote to the egon n 1990 s stll a net expote but n a decade ts net expots wee educed to half of the 1990 value. Table 3. County Intaegonal Agcultual Tade n Southen Afca (US$ mllon) Expots Impots Incease Incease Angola Malaw Mautus Mozambque South Afca a Zamba Zmbabwe Total a/ Tade of SACU countes manly South Afca s tade Souce: Authos calculatons usng COMTRADE 2005 data To gve us a bette sense of the mpotance of the egonal maket fo southen Afcan countes we measue tade ntensty n Table 4. Tade ntensty measues show that thee ae stong tade lnkages between countes n the egon gven that the shae of tade gong to the egon s much lage than expected accodng to the shae of the egon n total wold tade. The excepton s Angola whch shows weak lnkages wth southen Afcan countes whle t appeas to be ovetadng wth othe Afcan countes gven that ts tade shae wth these countes s lage than the shae of these countes n total wold tade. In contast wth the expots to the egon all countes (except Mautus) show low expot ntensty to hgh-ncome countes and othe egons whle most countes ovetade wth the est of Sub-Sahaan Afca. The domnant ole of South Afca as an expote n the egon can be seen n the blateal tade ntensty measues pesented n Table 4 whee South Afca s expot ntensty s always lage than that of any of ts tade patnes. Fo nstance the ntensty of South Afca s expots to Angola s 17.0 whle the ntensty of Angola s expots to South Afca s only 1.2; smla esults ae obtaned by compang South Afca s expot ntensty wth that of othe countes. 14

22 Table 4. Southen Afcan Countes Agcultual Expot Intensty n Dffeent Makets 1999 County Impot Makets EU Othe Developed Countes Rest of SSA Angola Malaw Mautus Southen Afca Angola Malaw Mautus Mozambque South Afca a Zamba Zmbabwe Note: Andeson and Nohem (1998) defne the ndex of tade ntensty between a specfc county and a goup of countes (egon) as: xj Ij = = m j ( q ) j x j j whee x j s the shae of county s expots gong to county goup j; m j s the shae of goup j n wold mpots (net of county s); q j s the shae of county j n wold GDP; and j s j s mpot-to-gdp ato dvded by the wold s (net of county s) mpot-to-gdp ato. If thee s no egonal bas that s f the shae of tade fom county gong to egon j s equal to the shae of j total mpots n wold tade then the ndex wll have a value of 1. a / Tade of SACU countes manly South Afca s tade Souce: Authos calculatons fom COMTRADE 2005 data Mozambque South Afca Zamba Zmbabwe Regon In ode to analyze the possbltes fo expandng egonal tade of agcultual poducts t s mpotant to dentfy the commodtes n whch countes n the egon have compaatve advantages and dsadvantages. Geate possbltes fo egonal tade expanson exst fo those commodtes n whch some countes have compaatve advantages whle othes have compaatve dsadvantages (complementaty). We use the evealed compaatve advantage (RCA) ndcato fo such analyss. The ndex s measued by the ato of expots fo a specfc commodty n a county's total expots elatve to the shae of ths commodty's tade n wold total tade. We assume that f the value of the ndex s geate than 2 (the shae of the good n the county's expots s twce the shae of ths good n wold tade) the county has a stong evealed compaatve 15

23 advantage n expotng that commodty. If the value of the ndex s less than 2 the county s consdeed to have a stong compaatve dsadvantage n that good. 6 The RCA ndces ae used to analyze tade complementates between countes. It s expected that countes wth dffeent compaatve advantage pofles would have n geneal moe oppotuntes to tade than countes wth smla specalzaton pattens. We use the numbe of matches between commodtes wth RCAs n one county (expote) and commodtes wth evealed compaatve dsadvantages (RCD) n the othe countes (mpotes) to vefy the degee of potental tade complementaty n the egon. Complementaty between expotes and mpotes s then measued by countng the matched numbe of commodtes blateally (Table 5). Compang the total numbe of commodtes wth RCA n each county that ae matched by commodtes wth RCD n the egon t appeas that Zmbabwe and South Afca ae the countes wth the best oppotuntes to ncease expots of agcultual poducts to the egon. Convesely Mautus Angola and Mozambque n that ode ae the countes wth the lagest numbe of commodtes wth compaatve dsadvantages fo whch othe countes n the egon show a compaatve advantage. The numbe of matches between commodtes wth RCA and RCD shows that thee ae egonal tade oppotuntes fo ceeals tadtonal expots futs and vegetables lvestock olseeds and ols and cotton (Fgue 3). 6 Followng Feto and Hubbad (2003) we use a global measue of elatve tade advantage (RTA) whch accounts fo mpots as well as expots. Ths measue s calculated as the dffeence between elatve expot advantage (RXA) and an ndex of elatve mpot advantage (RMA) as follows: RTA k whee k k k k [ x x ] [ m m ] = k w w x epesents the shae of expots of commodty k fom county n total s expots; shae of mpots of commodty k n total mpots of county ; and m s equal to k w k m s the k x w and epesents the shae of wold total tade of commodty k n total value of wold tade. Postve values of ths measue eveal a compaatve advantage n tade of commodty k by county whle negatve values show compaatve dsadvantages (RCD). 16

24 Fgue 3. Numbe of Matches Between Southen Afcan Countes wth Compaatve Advantage and Dsadvantage fo a Smla Commodty a ( Aveage) 25 Low-ncome countes Mddle-ncome countes 20 # of matches Ceeals Lvestock Cotton Futs & vegetables Olseeds & poducts Tadtonal expots Othes Note: Othe ncludes beveages leathe and wood poducts fsh and aw mateals. Souce: Authos calculatons fom COMTRADE data. In sum the analyss shows that thee ae oppotuntes fo the low-ncome countes to expand and dvesfy agcultual tade wthn southen Afca. Such oppotuntes ae condtoned by the gowth of South Afca whch wll geneate moe demand fo egonal agcultual expots and oppotuntes fo FDI gong to low-ncome countes. Howeve the low-ncome countes also face challenges fom gowth n South Afca as the unbalanced expanson of ntaegonal tade s manly explaned by gowth n South Afca s expots. As dscussed by Daves (2001) egonal ntegaton could exacebate the tendency towad polazaton callng fo an appoach to ntegaton wth a developmental focus (Ramsamy 2001). Rathe than tade ntegaton alone the egon needs a pogam that combnes tade ntegaton sectoal coopeaton and polcy coodnaton to addess the majo challenges faced by the low-ncome countes (Daves 2001). 17

25 Table 5. Numbe of Matches Between a County wth Compaatve Advantage and a County wth Compaatve Dsadvantage fo a Smla Agcultual Commodty a aveage Expotes Impotes Angola Malaw Mautus Mozambque South Afca Zamba Zmbabwe Total Matches Expotes Angola Malaw Mautus Mozambque South Afca Zamba Zmbabwe Total matches mpotes Note: a/ Fve-dgt level SITC classfcaton b/ Tade of SACU countes manly South Afca s tade Souce: Authos calculaton usng COMTRADE 2005 data Regonal Agcultual Gowth Oppotuntes The analyss of the man economc chaactestcs of southen Afca and the stuctue and evoluton of agcultual poducton and tade n the egon esulted n the dentfcaton of seveal chaactestcs that offe southen Afca specal oppotuntes to foste development and agcultual gowth though egonal lnkages. Hee we hghlght thee of these chaactestcs: (1) complementates between low- and mddle-ncome economes and hence stong tade and nvestment lnkages acoss countes (2) unexploted agcultual gowth potental and (3) unexploted agcultual tade oppotuntes. Southen Afca s the only egon n the Afcan contnent wth a numbe of mddle- and low-ncome countes n close poxmty to each othe. South Afca s aleady the egon's engne of gowth wth pe capta ncome of $3002 pe yea 38 pecent of the egon s total populaton and moe than 70 pecent of ts GDP. Futhemoe two othe mddle-ncome countes Botswana and Mautus though elatvely small ae seen as the most successful examples of economc development n 18

26 Afca. Obvously economc development among the lowe-ncome southen Afcan countes and the fosteng of agcultual gowth depends ctcally on how these countes can best take advantage of a unque oppotunty to beneft fom the egonal dynamcs affoded by the moe advanced neghbos. South Afca could nfluence gowth n othe countes though dffeent channels: ntenatonal tade spllove effects FDI and fnancal lnkages. Ths county could also affect busness and consume confdence n othe Afcan countes gven the sze of ts economy and ts leadeshp ole n egonal economc and poltcal ntatves. Aoa and Vamvakdes (2005) econometcally estmate ths potental effect usng data fo the peod The esults ndcate that an ncease of 1 pecentage pont n South Afcan economc gowth s coelated wth a pecentage pont ncease n gowth n the est of southen Afca. Although South Afca has been a hgh-mddle-ncome county snce the 1970s Apathed and the sanctons that followed t effectvely solated t fom the est of the wold and pompted polces amed at ensung self-suffcency. Fo example past subsdes of lage-scale agcultue by South Afca s govenment have left the county wth an extemely captal-ntensve agcultual secto whch none of the neghbong countes' small-scale fams can compete aganst. Howeve the lftng of the sanctons n the ealy 1990s and the esultng esugence of the South Afcan economy have allowed the county to sgnfcantly ncease ts foegn tade ncludng tade wth ts SADC neghbos (Thulow 2004). Futhe lbealzaton of captal makets dung the late 1990s also caused huge captal outflows fom South Afca nto the SADC egon and the est of Afca. Many of these nvestments have been n agcultue o agcultue-elated sectos. Fo example South Afcan supemakets have ceated demand fo hgh-value locally poduced poducts and have establshed supply chans both wthn and outsde of the egon. Thee have also been South Afcan nvestments n oads pots and othe maketelated nfastuctue n neghbong countes whch also mpove maket condtons fo both agcultual and nonagcultual expots n the egon as a whole. 19

27 Angola s anothe county that has potental to geneate egonal gowth dynamcs n southen Afca. Snce ts postwa economc ecovey began some 10 yeas ago Angola has aveaged almost 7 pecent n annual GDP gowth. As a county ch n natual esouces and wth annual pe capta ncome of $803 Angola depends on mpots fo most agcultual poducts. Fo example almost 60 pecent of the county s ceeal demand has to be met by mpots: fo maze alone 30 pecent of domestc supply s mpoted. The county also mpots 30 pecent of the pulses consumed domestcally and these account fo 20 pecent of the egon's total pulse mpots. 7 If Angola s economy contnues to gow apdly n the comng yeas t could become an mpotant maket fo agcultual expots fom othe countes n the egon. Regonal gowth oppotuntes also come fom the egon s agcultual potental. Howeve the poo pefomance of the agcultual secto manly a esult of bad polces o poltcally unstable envonments has constaned the egon fom explotng ts agcultual potental. Fo example an uban bas n economc development polces that lagely emphaszes the mneal secto has sgnfcantly hut Zamba s agcultual gowth (Thulow and Wobst 2004). In Zmbabwe ecent poltcal nstablty has esulted n declnng agcultual poducton. Whle fve-yea aveage yelds fo maze poducton n Zamba and Zmbabwe wee only pecent below South Afca s level dung the ealy 1980s the yeld gap has nceased to 50 pecent n Zamba and 80 pecent n Zmbabwe n ecent yeas ( ) (Table 6). These and othe falues to explot the egon s agcultual potental have been lagely esponsble fo the tansfomaton of southen Afca nto a food-defct egon. A ecovey of maze poductvty to ts hstocal hghest level could sgnfcantly mpove low-ncome countes compettveness and esult n mpot substtuton of maze lvestock and othe commodtes povdng these countes wth moe gowth oppotuntes n agcultue. 7 In tems of Angola's demand fo cash cops all suga consumed n the county s mpoted whch agan accounts fo 30 pecent of the egon s suga mpots. Futhemoe 13 pecent of vegetables ae mpoted accountng fo 17 pecent of the egonal vegetable tade. Whle the county has a elatvely lage lvestock secto domestc poducton could not meet domestc demand even back n the late 1970s. In ecent yeas moe than 33 pecent of meat demand n the county has been met by mpots ncludng 85 pecent of the poulty consumed. 20

28 Table 6. Land Poductvty n Low-ncome Southen Afca Compaed to Land Poductvty n South Afca (n klogams/hectae) aveage aveage Malaw Mozambque Zamba Zmbabwe South Afca Malaw Mozambque Zamba Zmbabwe South Afca Maze Wheat Rce Roots & tubes Pulses Olcops Futs Vegetables Cotton Sugacane Tobacco Souce: Calculated fom FAOSTAT 2006 Evdence of the potental mpotance of the othe poducts fo the egon (such as futs vegetables olseeds and cotton as nontadtonal cops) can be deved fom the expanson of tade of these poducts n the egon and fom gowng egonal nvestments n the value chans of poducton. South Afca s FDI to the egon mostly n mneal ndustes sevces and food etalng (such as supemakets and fast food chans) has been gowng apdly. Total South Afcan FDI n Afca amounted to 3.33 bllon dollas n pecent hghe than n Most of ths nvestment went to neghbong countes n southen Afca. Between 1994 and 2003 South Afcan companes nvested 2.8 bllon dollas n Mozambque whle the DRC Namba Zamba and Zmbabwe also eceved lage amounts (Rumney and Pngo 2004). Though most of the nvestment has gone to mnng basc steel and nonfeous ndustes and utltes a sgnfcant amount has been nvested n food etal (see Weathespoon and Readon 2003). Although these nvestments ae n tun helpng to ncease expots fom South Afca ths s expected to change n the futue as the etal and agbusness fms n each county nceasngly 21

29 nvest n local dstbuton netwoks and become dependent on local supples. Moeove by ncopoatng local supples nto egonal value chans domestc agcultual sectos could become moe dvesfed and even specalzed as egonal tade flow nceases. As dscussed by Nyabu (2004) the majo baes to ntaegonal tade ae not taffs and nontaff egulatoy egmes but undedeveloped poducton stuctue and nadequate nfastuctue. The fst of these factos s eflected n low poductvty. Oppotuntes also exst n expot agcultue. Olseeds and textle fbes (cotton) appea to be the commodtes wth potental to expand expots fom the egon to SACU at pesent levels of poductvty and compettveness. Moe oppotuntes could esult fom nceasng poductvty and compettveness of othe cops. Whle the egon expoted 2.3 mllon tons of fut and vegetables fo a value of almost US$ 1 bllon n pecent of these expots ae fom South Afca. At cuent technology levels most lowncome countes n the egon can hadly compete wth South Afca fo such expot makets. Fo example aveage yelds of fut and vegetables n the egon s low-ncome countes ae only half that of South Afca and much lage gaps exst n the qualty of many commodtes. Thee also exst nceased expot oppotuntes n commodtes that ae tadtonal expotables n some countes but nontadtonal n othe countes. Cotton n Zamba s a good example: ts cotton expots have nceased sevenfold ove the last two decades. Zamba now s the thd lagest cotton expote n southen Afca (afte Zmbabwe and Mozambque). And almost all of Zamba s cotton s poduced by smallholdes. In sum we have pesented a numbe of dstnctve chaactestcs of southen Afca s economy that togethe offe a unque oppotunty to foste the egon s economc development and agcultual gowth though egonal lnkages. These chaactestcs nclude complementates between low- and mddle-ncome southen Afcan economes stong tade and nvestment lnkages and unexploted agcultual tade oppotuntes. The next secton analyzes the potental mpact of South Afca s economc gowth fo the egon usng a egonal computable geneal equlbum (CGE) model. 22

30 III. ANALYZING GROWTH LINKAGES IN SOUTHERN AFRICA A Regonal Geneal Equlbum Model fo Southen Afca The analyss of southen Afca s compaatve advantage n agcultual tade n the pevous secton showed that nontadtonal expots seem to offe the best oppotunty to ncease tade n the egon. Howeve these esults have lmtatons because they ae obtaned fom hstocal data dung a peod of low gowth n South Afca and poo agcultual poducton and poductvty gowth n the low- ncome countes. The esults ae also lmted because they do not allow us to compae the mpact of dffeent agcultual subsectos on economc gowth n low-ncome countes o to dentfy gowth lnkages n the egon. To evaluate fully the ole of agcultual subsectos n economc gowth and food secuty n the egon t s necessay to have an economywde vew. Theefoe we pesent a egonal geneal equlbum model n ths secton and apply the model to assess how economc lnkages n the egon affect stategc optons and potes fo agcultual development n southen Afca. A detaled descpton of the model can be found n Appendx B. The Model and Data Descpton The computable geneal equlbum (CGE) model as ts name suggests conssts of an economy-wde multsectoal model that solves smultaneously and endogenously fo both quanttes and pces. As the coe of the model conssts of the econclaton of potental demand and supply mbalances n commodty and facto makets afte ntoducng any shock (such as tade polcy change and poductvty gowth) the CGE model s a useful tool to captue both consumpton and poducton lnkages between agcultue and the est of the economy. In addton to these featues whch ae common to all CGE models n the egonal CGE model used fo ou study equlbum between commodty demand and supply n the wold maket s also obtaned allowng the model to captue the blateal tade elatonshps between the countes ncluded n the model. The model also solves fo wold commodty pces smultaneously wth othe endogenous vaables. 23

31 The technologcal and behavoal functons fo both poduces and consumes consst of nonlnea and substtuton possbltes among factos of poducton and among commodtes n fnal demand. Poducton technology s epesented by fxed nput-output coeffcents fo ntemedate goods and constant elastcty of substtuton (CES) functon fo the followng pmay nputs: two types of labo (sklled and unsklled) land othe natual esouces and captal. Whle supply of othe poducton factos s assumed to be fxed wthn each county the model assumes the exstence of unemployment n unsklled labo among low-ncome southen Afcan countes. Poducton technology vaes acoss sectos and countes and s calbated to the countes data. Whle poducton-demand lnkages ae manly captued by the nput-output elatonshps ncluded n the model n most low-ncome southen Afcan countes such lnkages between agcultue and nonagcultue ae weak gven that the level of ntemedate nput use s qute low n most agcultual actvtes. As value-added s the majo component of poducton evenue evaluated at poduce pces consumpton lnkages ae sgnfcantly affected by the facto ntensty whch vaes acoss sectos and countes. A captal-ntensve secto may geneate fewe consumpton lnkages among poo consumes whose ncomes ae manly fom wage eanngs. Ths s one of the majo easons why gowth n smallholde agcultue has elatvely stong coss-secto lnkages n developng countes. The empcal analyss pefomed n ths study evaluates the magntude of these lnkages. Consumpton demand lnkages ae hghly affected by ncome levels consumpton pattens and magnal popensty to consume each of whch vaes acoss countes. In a geneal equlbum model pce esponses (expessed though pce elastctes of demand) ae also mpotant as all pces n domestc makets ae endogenously solved n the model. The ncomes of consumes ae detemned n the facto makets afte subtactng taxes. The demand fo commodtes by secto s detemned fom these ncomes (gven household savngs popenstes) and fom the govenment consumpton functons. Ou egonal CGE model solves consume demand by maxmzng a Stone- Geay utlty functon whch mples lnea expendtue systems (LES) fo ndvdual 24

32 commodtes. The ncome elastctes used to deve the magnal budget shaes fo consumpton ae fom Reme and Hetel 2004; fo example ncome elastctes fo gans ange fom 0.4 to 0.5 fo the low-ncome Afcan countes. The subsstence paametes n the demand functons ae calculated by assumng a Fsch paamete (togethe wth ncome elastctes) fo each ndvdual county. Once we know the ncome elastctes and subsstence paametes pce elastctes (ncludng own and coss pce ones) can be deved by mposng the homogenety condton on the LES functons. Ths pocedue esults n pce elastctes of demand fo gans fo example of between and The model assumes pce-senstve substtuton (mpefect substtuton) among foegn goods and domestc poducton and among goods poduced by dffeent tadng patnes. Because of ths assumpton domestc goods cannot fully substtute fo mpots even f poductvty mpoves n the domestc poducton secto. Impefect substtuton mples that poductvty mpovements n the agcultual secto ae not enough and addtonal tadng facltes and mpovng maketng condtons ae necessay to mpove substtuton between domestc and foegn goods. The model ncludes sx ndvdual southen Afcan countes: Botswana Malaw Mozambque South Afca Zamba and Zmbabwe and two aggegate subegons: the est of SACU and the est of southen Afca. The model also ncludes thee countes n East Afca: Madagasca Tanzana and Uganda a est of Sub-Sahaan Afca egon two Noth Afcan countes (Moocco and Tunsa) and a est of Noth Afca egon. Outsde Afca the model ncludes two bg Asan countes (Chna and Inda) and a est of Asa egon as well as Afca s two majo tadng patnes (the Unted States and the Euopean Unon) and the othe Euopean countes as a goup. The est of the wold s ncluded as anothe sepaate egon aggegatng all othe countes not ncluded above. 25

33 The focus of the study s low-ncome countes n southen Afca 8 whch ae explctly defned n the Global Tade Analyss Poject (GTAP) database used n the study. 9 The model focuses on agcultue and ncludes 21 agcultual and agcultueelated sectos and 11 nonagcultual sectos many of whch such as tanspotaton and textles dectly lnk to the agcultual secto. Incluson of moe dsaggegated agcultual subsectos s constaned by the GTAP database. In the latte many egonally mpotant agcultual commodtes (such as tobacco fo expot o cassava and othe oot and tube cops to meet domestc demand) ae ncluded n an aggegate secto called the othe cop secto and cannot be dstngushed as ndvdual commodtes. Fo the puposes of ths study we adjusted ths secto accodng to the degee of maket oentaton. Specfcally we splt the othe cop secto ncluded n the GTAP database nto two: expot othe cops and domestcally consumed othe cops. We use expot othe cops to epesent tadtonal expot tee cops and tobacco whle the domestcally consumed othe cop secto epesents oots and tubes used as staples. Smlaly we splt the GTAP s aggegated vegetable and futs secto n two: nontadtonal expotables and futs and vegetables fo domestc makets. Two tanspot sectos n the GTAP database wate and othe tanspot povde data on nputs consumed by othe sectos n the poducton pocess and also affect pce magns fo ntenatonal tade. 10 Intenatonal tanspotaton magns ae calculated fo Afcan countes usng blateal data on c..f. and f.o.b. pces based on nfomaton fom Lmao and Venables (2002). Whle the maket value of such pce gaps s teated as expots of tanspotaton sevces fom expotng countes to mpotng countes 11 the magns wll be endogenously affected by the changes n the poduce pce fo the domestc tanspotaton secto. Impovng the tanspotaton secto s poductvty lowes the unt cost of sevces povded by the secto whch causes expots to become moe 8 They ae Malaw Mozambque Zamba and Zmbabwe; n the ognal database Lesotho was aggegated nto a egon called est of SACU. 9 The GTAP database veson 6.1 not the GTAP model tself s used n ths study. The GTAP s a poject of Pudue Unvesty. The GTAP data veson 6.1 epesents the wold n 2001(Dmaanan 2006). 10 Due to data lmtatons we dd not consde pce magns n domestc makets. 11 Even though ntenatonal tanspotaton sevces can be povded by expotng o mpotng countes n ealty they ae often povded by a thd paty. 26