NBER WORKING PAPER SERIES MODEL STRUCTURE AND THE COMBINED WELFARE AND TRADE EFFECTS OF CHINA'S TRADE RELATED POLICIES. Yan Dong John Whalley

Transcription

1 NBER WORKING PAPER SERIES MODEL STRUCTURE AND THE COMBINED WELFARE AND TRADE EFFECTS OF CHINA'S TRADE RELATED POLICIES Yan Dong John Whalley Workng Paper NATIONAL BUREAU OF ECONOMIC RESEARCH 050 Massachusetts Avenue Cambrdge, MA 038 September 009 We are grateful to the Ontaro Research Fund (ORF-3), CIGI, and the Socal Scences and Humantes Research Councl for fnancal support. The vews expressed heren are those of the author(s) and do not necessarly reflect the vews of the Natonal Bureau of Economc Research. NBER workng papers are crculated for dscusson and comment purposes. They have not been peerrevewed or been subject to the revew by the NBER Board of Drectors that accompanes offcal NBER publcatons. 009 by Yan Dong and John Whalley. All rghts reserved. Short sectons of text, not to exceed two paragraphs, may be quoted wthout explct permsson provded that full credt, ncludng notce, s gven to the source.

2 Model Structure and the Combned Welfare and Trade Effects of Chna s Trade Related Polces Yan Dong and John Whalley NBER Workng Paper No September 009 JEL No. F ABSTRACT Because Chna s economc structure s dfferent from that n OECD countres, usng conventonal neo-classcal compettve trade models to analyze the welfare and trade mpacts of trade related polcy change can be msleadng. In partcular, both the exchange rate regme and output and prcng polces of state owned enterprses (SOE s) wll have effects on trade and welfare whch dffer from a classcal compettve model. Ths paper present a numercal model that captures the combned and nteractve effects of three polcy elements n prototype form of tarffs, polcy towards SOEs n the ndustral sector, and an exchange rate regme supportng large trade surpluses and addtons to foregn reserves. The model has non neutral monetary features, endogenous trade mbalances and average product prcng of labor n goods. We do not clam t to be fully representatve of modern Chna, but t does go some way beyond smple compettve models used elsewhere and ponts to dfferent conclusons of polcy mpact. We calbrate our model to 006 data, and then evaluate the mpacts both sngly and n combnaton of: tarff lberalzaton, a move to more freely floatng exchange rates, and SOE enterprse reform. Results show that large dfferences n polcy mpacts relatve to a classcal compettve model. SOE reform and a freely floatng Chnese exchange rate have more mpact on Chna s welfare than tarff lberalzaton. Polces of RMB apprecaton and ncreasng Chna s money stock reduce Chna s trade surplus. In the tradtonal compettve model, trade lberalzaton mpacts both mports and exports, whle n our central case model, wth endogenously determned trade surplus, trade lberalzaton has lttle effect on exports. Most of the polcy mpact s on mports and the trade surplus. SOE reform of Chna s manufacturng sector sgnfcantly decreases producton of Chna s manufacturng sector and ncreases producton n Chna s other sectors. Yan Dong Insttute of World Economcs and Poltcs Chnese Academy of Socal Scences 5th Floor of CASS Buldng No.5 Janguomen Ne Avenue Bejng, Chna, 0073 dongyan@cass.org.cn John Whalley Department of Economcs Socal Scence Centre Unversty of Western Ontaro London, Ontaro N6A 5C CANADA and NBER jwhalley@uwo.ca

3 . Introducton The center pece of Western neo-classcal economc analyss s the Arrow-Debreu compettve equlbrum model. It has been used n numercal form to descrbe both Western and Chnese economes despte Chna s economy beng qute dfferent n structure. Notably Chna s n transton from a central-planned economy to a market economy, and other elements complcate analyss, such as SOEs, and the exchange rate regme. Here, we present a calbrated numercal model of Chna wth SOE producton, non neutral monetary and exchange rate structure, and analyze the mpacts of dfferent lberalzaton polces, not only tarff lberalzaton, but also RMB apprecaton and SOE reform, ether sngly or n combnaton. Our concluson s that usng smple compettve trade models wth ad-valorem tarff equvalent representaton of all polces can be msleadng. In the model, for the manufacturng SOE sector, we use a verson of the manageral control model of SOEs n Whalley & Zhang(006) and Fosse & Ramondos-Møller(009), n whch SOEs are assumed to be under manageral control wth managers poltcally apponted. Captal s assumed allocated freely to the SOE by credt ratonng wth a state bank recaptalzaton mechansm coverng any losses. We assume that enterprse managers maxmze personal gan from networkng and poltcal connectons, and enterprse managers maxmze enterprse sze rather than profts. The mplcaton s that f managers hre labour n a compettve labour market (a strong but greatly smplfyng assumpton for the Chnese case) and pay labour ts gong wage, they wll hre labour up to the pont that the product prce equals the average value product of labour, rather than the margnal value product as n the compettve case. We also add monetary structure nto a trade model, as n Wang & Whalley(007), to reflect a managed Chnese exchange rate and monetary regme wth a peg and RMB nconvertblty. Monetary polcy s thus non accommodatve to the chosen fxed exchange rate, and ths mples ether excess supply of foregn exchange whch the Central Bank accommodates as addtons to reserves. We calbrate ths model to 006 base year data for two countres, Chna and the Rest of the World. Our base case ncludes 006 trade, producton, and consumpton and constructed 3

4 money supply data. We then use the calbrated model to smulate the mpacts of the 3 types of polcy change on welfare, trade flows, the trade surplus and producton. These cover tarff lberalzaton, exchange rate and monetary polcy (Renmnb apprecaton and /or ncreases n the money stock), and SOE reform. Results show that tarff lberalzaton ncreases mports but has no mpact on exports and s welfare mprovng as the surplus falls. The results for RMB apprecaton and monetary polcy have effects on exports but are small and reduce trade mbalances and are also welfare mprovng.soe reform only has sgnfcant trade and welfare mpacts n the absence of monetary non neutraltes. Most of the avalable lterature on Chnese SOEs uses econometrc models wth no explct trade and monetary structure, assessng the performance, proftablty and labor hrng (examples are Bgsten, Lu, and Zhang (00), Cull and Xu (003), and Lu (00)). And whle there s lterature analyzng nteractons between trade structure and monetary polcy, they use tarff equvalents and analyze trade polcy changes ether gnorng monetary structure(such as IMF(005), Wlenbockel(006)), or trade mpacts of exchange rate changes n structures where trade pattern changes do not follow from a trade model (such as Park(005),Kamada &Takagawa(005),Marquez & Schndler(006)). 4

5 . A Prototype Model Capturng Some Key Chnese Features We analyze the potental mpacts of dfferent lberalzaton polces, tarff lberalzaton, RMB apprecaton, and SOE reform, usng a calbrated numercal general equlbrum model of Chna wth SOE producton and monetary structure. For the SOE manufacturng sector, we use the manageral control model of SOEs used n Whalley & Zhang(006) and Fosse & Ramondos-Møller(009). In ths SOEs are assumed to be under manageral control wth managers poltcally apponted. Captal s assumed allocated to the SOE by credt ratonng, wth an assumed state bank recaptalzaton mechansm whch covers any losses. In effect, captal s freely allocated. Ths more accurately reflects Chna n the 980 s and early 990 s more so than today, but s analytcally tractable and sgnfcant elements of ths structure reman. We assume that enterprse managers maxmze personal gan from networkng and poltcal connectons and hence that enterprse managers maxmze enterprse sze rather than profts. The mplcaton s that f managers hre labour n a compettve labour market (a strong smplfyng assumpton) and pay labour ts gong wage, they wll hre labour up to the pont that the product prce equals the average value product of labour, rather than the margnal value product as n the compettve case. On the monetary sde we follow Wang & Whalley (007), and specfy a smple monetary regme n a model wth monetary non-neutraltes. In ths, monetary polcy s non accommodatve to the chosen fxed exchange rate, and gven the current Renmnb peg ths mples excess supply of foregn exchange whch the Central Bank smply accumulates as reserves. More formally, the model ncorporates countres (Chna and ROW) ndexed by j or k ( j =,, k =,) to denote country,and traded goods (manufactures and other ndexed ( =, assumpton). ) to denotes goods. Goods across countres are heterogeneous (the Armngton 5

6 For each good produced n country k, we defne the seller s prce (net of tarff) as P, and allow each country j to mpose tarffs at ratet k j ( country j s tarff on good mported from country k ) on each mported good. Tarffs are set to zero for exports. Internal (gross of tarff ) prces for good produced n country k are thus P = [ + t ] P () k j k j k On the producton sde n country (Chna), we consder a SOE manufacturng sector and a compettve other sector. Both sectors have decreasng returns producton functons, wth labor as the varable nput. Captal used by the SOEs s fxed to reflect captal allocated to ths sector by ratoned credt and s unprced wth average product prcng of labor. Captal used n the other sector receves a rental return snce t s mmoble ntersectorally. In the SOE sector, the producton functon s M M M Y = ϕ ( L ) αm () k where Y M s manufacturng output n country, LM s labour used n manufacturng, ϕm s a unts term(scalar parameter), and α M < s the producton exponent. The use of captal by the SOE s captured by a fxed factor. Captal s assumed to be allocated va centralzed credt allocaton through the state bankng system, but recpents of loans (and hence captal) expect that any servcng costs of the loan wll be covered by state bank recaptalzaton of losses. We assume managers of the enterprse are poltcally apponted and are concerned wth the sze of ther personal network rather than profts, and networkng benefts are collnear wth the sze of the enterprse they manage. In ths smple model wth a sngle SOE, labour s pad ts average value product snce managers maxmze enterprse sze measured by output Y M subject to the enterprse budget constrant. Snce captal s effectvely unprced, ths mples that PY = w L + R (3) M M M M M that s w M PY R = (4) M M M LM 6

7 where R M s any surplus requred to be transferred to the state by the SOE. Ths, n turn, mples that labour receve ts average value product plus ts share of requred surplus rather than ts margnal product, and that mples that condtons for Pareto effcency are volated. In the other sector, the producton functon s Where O O O Y = ϕ ( L ) αo (5) Y O s output n the other sector n country, ϕo s a unts term(scalar parameter), and ths sector labour s pad ts margnal product,.e. the wage rate s O LO s labour used n the other sector, α < s the producton exponent. We assume that n Y α w P P ( L ) O O O = O = ϕo OαO O LO (6) and the resdual return PY w L accrues as rent to already nvested captal. ( O O O O) In country (rest of the world), the producton functons are: Y The ϕ ( L) α = ( = M, O) (7) Y are manufacturng output and other sector output n country, L are labour used n manufacturng sector and other sector, ϕ are unts terms (scalar parameters), and α < s the producton exponent. We assume that n both sectors labour s pad ts margnal product,.e. the wage rate s Y α w P P ( L ) = = ϕ α L ( = M, O) (8) On the demand sde of the model, fnal demands for commodtes n country and country as modeled are derved from maxmzaton of CES nested utlty functons defned over compostes of smlar domestcally produced and mported goods subject to country budget constrants. Ths Armngton structure allows us to drectly ncorporate substtuton elastcty parameters nto the model as the substtuton elastcty between smlar domestc and mported goods whch can then be drectly related to mport demand elastctes. Unlke n a conventonal trade model, we ncorporate endogenously determned trade surpluses and defcts drectly nto the two country budget constrants. To do ths we use a 7

8 smple monetzed extenson to a pure barter trade model wth a transactons demand for money and untary velocty of crculaton. We assume that the surplus country s predetermned as country (Chna) and the defct country as country (the rest of the world), and Country fxes ts exchange rate and has non-accommodatve monetary polcy. Country s assumed to fx ts money stock to accommodate ts trade defct. Ths mples that jontly countres and set relatve money stocks consstent wth the fxed country exchange rate and trade surplus, but monetary non neutraltes result. A change n the fxed exchange rate changes both trade flows and country s surplus. Maxmzng country s utlty yelds demands from the soluton to a stage budgetng problem, e: maxu { C ( D, D )} (9) St D + ep D = P I = P Y es + TR + R (0) I where C s the CES composte of domestc demand and mports n country, D ( ) s D the demand n country for good produced n country (), and e s the fxed exchange rate, S s the trade surplus of country, TR s the tarff revenue n country, and R are transfers to the state by the SOE n country. These are assumed recycled to consumers as lump sum transfers. Country s demands are smlarly derved by maxmzng a nested CES utlty functon defned over CES compostes of mported goods from country and domestcally produced goods,.e. maxu St { C ( D, D )} D + ep D = P I () = P Y + TR + S () I 8

9 We assume each country s money supply s fxed at m and m, and for smplcty, each country has a constant unt velocty transactons demand for money. In equlbrum, ths mples: and D + P D = P m D + P D + S = P m snce the surpluses of country mples money holdng n the currency of ($) Addng demand supply equaltes n goods gves (3) (4) D + D = Y ( = M, O) (5) D + D = Y ( = M, O) (6) and factor market clearng condtons L + L = L (7) M O L + L = L M O In equlbrum, factor prces are equalzed across the two sectors n country, and wage rates are equalzed across the two sectors n each country,.e. w w = w (8) A M = w A M and sellers commodty prces clear goods markets. These prces are consstent wth zero proft condtons n country and SOE prcng the determnaton of rent n the other secton n country. We can buld a model admssble data set for 006 for ths structure whch we use to determne model parameters through calbraton and then perform counterfactual experments for changes n dfferent polcy elements. Alternatve equlbra can easly be computed for ths system. 9

10 3. Data and Model Calbraton We buld a model compatble benchmark general equlbrum data set whch we use n calbraton. We use a base year of 006. The two countres n our smulatons are Chna and the Rest of the World. Our base case data ncludes 006 trade, producton, consumpton and country money supply data constructed to equal the value of transactons assumng untary velocty. In the model, for smplcty, we treat the entre manufacturng sector as a sngle SOE. Ths s a strong assumpton whch gnores Chna s growng prvate manufacturng sector and the role of nward FDI flows. It s adopted as a smplfcaton for purposes of tractablty. The whole Chnese manufacturng sector s also treated as producng a sngle product. We defne physcal unts for both manufacturng and the other sector product to be related to value observatons from natonal accounts followng the Harberger(96) and Shoven & Whalley(97) unts conventon that n the ntal benchmark equlbrum data world prces are unty,.e. P = P = P = P =, and factor prces are unty,.e. M M O O w = w = w = w =. M M O O To convert Chnese data nto unts consstent wth the ROW data n $, we construct an artfcal Chnese currency unt, RMB *, whch we set as RMB* = 7.97RMB, whch s the exchange rate of the US dollar to Renmnb n 006. Ths mples that unt of RMB * equals US dollar. Ths conventon s adopted so that n calbraton all equlbrum prces wll be unty for both Chnese and ROW goods and factors. In Table,GDP data s from the World Bank s WDI database and trade data s taken from the UNCOMTRADE database. F.o.b. exports values as reported by exportng countres are used, Chna s exports and mports are taken as mports and exports by the rest of the world from and to Chna. These trade data, n turn, mply Chna s trade surplus for 006. Labor nput data s from Chna s NBS data and from the U.S. Bureau of Labor of Statstcs. We assume the labor nput of ROW s 4 tmes that of U.S. on the bass that U.S. GDP s roughly ¼ of world GDP Chna. Tarff data s from the WTO Statstcal Database, and for 006, the average tarff on Chna s manufactured mport was 9%, and on other goods was 5.8%. 0

11 Elastctes n the central case model specfcaton are set as follows. There are no avalable estmates of elastctes for Chna ether on the demand or producton sdes. We set the substtuton elastctes n demands between domestc and mported commodtes at n both countres. We later use sensttvty analyss to change ths value. The substtuton elastctes between two domestc goods n each country we set at 0.5 and the substtuton elastctes between the two mported goods n each country are also set at 0.5. For Chna s SOE manufacturng sector, frst order condtons mply no value drectly for the exponent n the producton functon. We assume a value of α = 0.4,whch we loosely justfy as the share of labor n total manufacturng output from the NBS (Chna s Natonal Bureau of Statstcs) data, whch n the compettve (rather than the SOE ) case. Ths yelds a value of ϕ M from equaton (). We also perform senstvty analyss around ths settng. M Table Base Case Data n 006 for Chna and ROW used n Calbratng the Model Chna (n bllon RMB* ) Value of Producton Value of labor nput Value of Consumpton Exports to ROW Import from ROW Net trade Manufacture Other Sector Surplus Exchange rate : m ROW (n bllon US $) Value of Producton Value of labor nput Value of Consumpton Exports to Chna Import from Chna Net trade Manufacture Other Sector Surplus Exchange rate : m We construct an artfcal Chnese currency unt, RMB*, we set t as RBM*=7.97RMB, whch s the exchange * rate of the US dollar to Renmnb n 006, ths mples that unt RMB equals US dollar. Ths conventon s adapted so that n calbraton all equlbrum prces wll be unty.

12 Table reports the parameter values n producton and preferences generated by calbraton. When used n model soluton these regenerate the benchmark dataset n Table as an equlbrum. Table Model Parameterzatons Generated by Calbraton to the 006 Benchmark Data A. Parameters n Producton Functons Technology coeffcentϕ Chna exponent on labor nputα Technology coeffcentϕ Rest of the World exponent on labor nputα Manufacturng Other Sector B. Parameters n Nested CES Utlty Functons Chna Rest of the World Shares of Domestc and Import composte goods Domestc Produced Imports Domestc Produced Imports Shares of manufacturng and other goods Domestc Produced Imports Domestc Produced Imports Manufacturng Other Sector

13 4. Numercal Analyss of Dfferent Lberalzaton Polces n Chna We have used our calbrated model to smulate the mpacts of varous forms of economc lberalzaton polces on welfare, trade flows, and the trade surplus and on producton. The lberalzaton polces nclude tarff elmnaton, a freely floatng exchange rate, and SOE reform. Results are presented n Table 3 to Table 6. We reports mpacts on welfare usng Hcksan money metrc measures of welfare change. We calculate these as % of GDP. These results suggest that SOE reform and a freely floatng exchange rate have more mpact on Chna s welfare than tarff lberalzaton usng 006 data. RMB apprecaton and an ncrease n Chna s money stock reduces Chna s trade surplus as would be expected. But mpacts on exports are small. Most of the mpact s on mports. For trade flows, n the tradtonal compettve model, trade lberalzaton nfluences both mports and exports, whle n our model, wth an endogenously determned trade surplus, tarff lberalzaton has no effect on exports. Impacts nstead are on mports and the trade surplus. SOE reform of Chna s manufacturng sector sgnfcantly decreases the producton of Chna s manufacturng sector and ncreases the producton of Chna s other sectors. In Table 3, we use our central model specfcaton to analyze the mpacts of three dfferent polcy elements n lberalzaton. For tarff elmnaton, we elmnate the tarff on both Chna s manufacturng and other sectors. In exchange rate polcy, we freely float Chna s exchange rate and the trade surplus becomes zero. For SOE reform, we assume Chna s manufacturng sector changes from SOEs to compettve enterprses. The frst column of Table 3 reports the mpacts of Chna s dfferent economc lberalzaton polces on welfare usng Hcksan CV measures. Welfare mpacts usng Hcksan EVs are smlar and so n the tables, we only report the results for CVs. The most effectve polcy to mprove Chna s welfare s SOE reform, the welfare mpact on Chna s.% of GDP, freely floatng exchange s also effectve wth a welfare mpact of.093% of GDP, but for tarff elmnaton the welfare mpact on Chna s only 3.4% of GDP. 3

14 In the second column of Table 3, we report the mpacts of dfferent Chnese lberalzaton polces on trade flows. Wth a freely floatng exchange rate, Chna s mports ncrease n value terms by around %. The ncrease of the mports of manufacturng and other sectors are smlar at nearly %. For tarff elmnaton, Chna s mports ncrease n value terms by over 0%. Manufacturng mports ncrease by 9.885% and the other sector mports ncrease by.937%. For SOE reform, the mpact on Chna s mports s very small, only -0.05%. In ths model specfcaton, the mpact of all lberalzatons on exports s very small. The endogenously determned trade surplus acts as a buffer adjustng to the polcy change so that mports change sharply wth lttle change n exports. In the thrd column of Table 3, we report the mpacts of alternatve Chnese lberalzaton polces on producton. SOE reform decreases producton n Chna s manufacturng sector by over 6% n quantty terms, and ncreases the producton of Chna s other sectors by 9% n quantty terms. Under a freely floatng exchange rate, the producton of Chna s manufacturng sector decreases by 3% n quantty terms, and the producton of Chna s other sectors ncreases by 6%. Tarff elmnaton has lttle effect on Chna s producton snce exports do not respond. 4

15 Table 3 Welfare, Trade and Producton Impacts of Model Based Evaluatons of Alternatve Lberalzatons (Model wth SOE average Product Labor Prcng and Monetary non-neutraltes / Trade Surpluses) (Substtuton elastcty between domestc and mported goods =, 0.4 α = ) M Welfare Impact as % GDP (Hcksan Measure) % Change n Chna s mports % Change n % Change n producton Chna ROW Total Manufacture Other Sector Chna s exports Chna-Man Chna- ROW- ROW- Other Man Other.Tarff Elmnaton 3.4% -0.64% 0.607% 9.885%.937% 0.000% 0.000% 0.000% 0.085% -0.09%.Freely Floatng Exchange rate.093% -0.57%.37%.43%.60% -0.00% -3.78% 6.48% 0.0% % 3. SOE reform.% % -0.05% -0.05% % 0.09% -6.76% 9.007% 0.000% 0.000% 5

16 In Table 4 we present results from the three dfferent forms of lberalzaton usng three dfferent models: the model wth both SOE prcng and non-neutral money as used for Table 3, a model wth only SOE prcng (neutral money) and a classcal compettve model wth nether of these features. Our purpose s to compare the welfare, trade and producton mpacts of alternatve lberalzatons across models to gauge how mportant elements of Chnese economc structure are for polcy evaluaton. We see large dfference among the results form these three models. For tarff elmnaton, n the compettve model, Chna s welfare decreases by -.00% of GDP due to an adverse term of trade effect. Addng SOE prcng and keepng money neutral ncreases the negatve mpact on Chna s welfare to -.58% as the terms of trade effects ntensfy. But n the model wth SOE prcng and non-neutral money, Chna s welfare ncreases by 3.4% of GDP. In ths model, tarff elmnaton has lttle effect on exports, snce most of the mpact s drected to the surplus, whle n compettve model and model wth only SOE prcng, the effects of tarff elmnaton on mports and exports are smlar. For SOE reform, the non-neutral money model shows a welfare gan of.% of GDP and the neutral money model shows an even hgher welfare gan of 5.85% of GDP. The latter model results show that SOE reform has a sgnfcant effect on Chna s mports (4.580%) and exports (3.56%). 6

17 Table 4 Comparng Welfare, Trade and Producton Impacts of Alternatve Lberalzatons Across Models (Substtuton elastcty between domestc and mported goods=, α = 0.4 ) M Welfare Impact as % GDP (Hcksan Measure) Model wth SOE prcng and non-neutral money % Change n Chna s mports % Change n Chna s exports Welfare Impact as % GDP (Hcksan Measure) Model wth only SOE prcng (neutral money) % Change n Chna s mports % Change n Chna s exports Welfare Impact as % GDP (Hcksan Measure) Compettve Model % Change n Chna s mports % Change n Chna s exports Chna ROW Chna ROW Chna ROW.Tarff Elmnaton 3.4% -0.64% 0.607% 0.000% -.58% 0.04% 7.37% 5.554% -.0% 0.06% 8.709% 7.4%.Freely Floatng Exchange rate.093% -0.57%.37% -0.00% SOE reform.% % -0.05% 0.09% 5.85% -0.64% 4.580% 3.56%

18 Table 5 then reports more detaled analyses of alternatve lberalzaton mpacts usng our central case model. In the trade polcy component, we frst elmnate only tarffs on manufactures, and then elmnate only other sector tarffs. In monetary polcy, we frst consder RMB apprecaton of 5%,0% respectvely wth a fxed money stock n both Chna and the Rest of the World, then we fx the RMB exchange rate and ncrease the money stock n Chna by 5% and 0% respectvely. Results n Table 5 show that monetary polcy agan seemngly has more mpact than tarff lberalzaton on Chna s welfare, trade and producton, whle Renmnb apprecaton and ncreasng monetary stock have smlar mpacts. Wth RMB apprecaton of 5% and 0%, Chna s welfare gans are 4.5% and 8.039% of GDP respectvely, whle wth an ncrease n Chna s money stock of 5%,0%, Chna s welfare gans are 4.4%, 8.0% of GDP respectvely, and elmnaton only of Chna s manufacturng tarffs leads to Chna s welfare gans of.986% of GDP. Elmnaton of only other sector tarffs leads to Chnese welfare gans by.9% of GDP. In the base model; tarff lberalzaton and monetary polcy have lttle mpact on exports. The largest mpact s on mports and the surplus. Wth RMB apprecaton of 5% and 0%, Chna s mports ncrease by 8.83%, 7.487% respectvely, wth ncrease n Chna s monetary stock by 5%,0%, Chna s mports ncrease by 7.869%, 5.738% respectvely, and elmnaton only of Chna s manufacturng tarffs leads to a Chnese mport ncrease of 6.756%. Elmnaton only of other sector tarffs leads to a Chnese mport ncrease of 3.63%. Trade polcy has lttle effect on producton, whle under Renmnb apprecaton and an ncrease n Chna s monetary stock Chna s manufacturng producton decreases and Chna s other sector producton ncreases. 8

19 Table 5 More Detaled Analyss of Alternatve Lberalzaton Impacts (Model wth SOE average Product Labor Prcng and Monetary non-neutraltes/surpluses) (Substtuton elastcty between domestc and mported goods=, α = 0.4 ) M.Elmnaton only Manufacturng tarffs.elmnaton only other sector tarffs 3. 5% Renmnb apprecaton 4. 0% Renmnb apprecaton 5. 5% ncrease n Chna s monetary stock 6. 0% ncrease n Chna s monetary stock Welfare Impact as % GDP (Hcksan Measure) % Change n Chna s mports % Change n % Change n producton Chna ROW Total Manufacture Other Sector Chna s exports Chna-Man.986% -0.03% 6.756% 7.894% 3.08% 0.000% 0.000% 0.000% 0.076% -0.06%.9% % 3.63%.788% 9.505% 0.000% 0.000% 0.000% 0.008% % 4.5% -0.94% 8.83% 8.347% 8.074% % -.56%.54% 0.075% -0.06% 8.039% % 7.487% 7.634% 7.0% % -.548% 5.040% 0.58% % 4.4% -0.85% 7.869% 7.930% 7.67% % -.95%.394% 0.07% -0.04% 8.0% % 5.738% 5.868% 5.36% % -.30% 4.580% 0.43% % Chna- Other ROW- Man ROW- Other 9

20 Three parameters n the model are crucal for these results. One s the top level trade elastctes; a second s the bottom level trade elastctes, the thrd s α M n the producton functon of Chna s SOE manufacturng sector. We thus perform senstvty analyses on the results n Table 3 by varyng these three parameters. Table 6 provdes the results. The welfare effect of tarff elmnaton s very senstve to the top level elastctes used, whle changng the bottom level trade elastctes has lttle mpact on welfare effects of tarff elmnaton. Wth varyng α M the welfare mpacts of trade elmnaton and freely floatng exchange rates change greatly. When the αm changes from 0.4 to 0.75, the welfare mpacts of tarff elmnaton changes from 3.4% to.979%. The welfare mpact of freely floatng exchange rate changes from.093% to 7.386%,the welfare effect of SOE reform changes from.% to 4.630%.But whle results are elastcty dependent, the theme of results that departures from the classcal compettve model matters greatly for results remans. 0

21 Table 6 Senstvty Analyses of the Results on the Impacts of Alternatve Lberalzatons n Table 3 (Model wth SOE average Product Labor Prcng and Monetary non-neutraltes/surpluses) Welfare Impact as % GDP (Hcksan Measure) Chna ROW.Tarff Elmnaton Central Case 3.4% -0.64% ½ top level trade elastctes 0.498% -0.08% Increase top level trade elastctes by 50% 5.0% -0.44% ½ bottom level trade elastctes 3.065% -0.6% Double bottom level trade elastctes 3.73% -0.66% Set producton exponent n SOE sector to % 8.979%.Freely Floatng Exchange rate Central Case.093% -0.57% ½ top level trade elastctes 6.074% % Increase top level trade elastctes by 50%.47% -0.79% ½ bottom level trade elastctes 9.556% % Double bottom level trade elastctes 9.88% % Set producton exponent n SOE sector to % 8.643% 3.SOE Reform Central Case.% % ½ top level trade elastctes 0.944% % Increase top level trade elastctes by 50%.98% -.370% ½ bottom level trade elastctes.433% -0.54% Double bottom level trade elastctes.84% % Set producton exponent n SOE sector to % 9.6%

22 5. Conclusons In ths paper, we present numercal general equlbrum models of Chna wth SOE behavor and non-neutral monetary structure that capture n a smplfed form features of Chnese economc structure mssng n conventonal compettve models. We calbrate these models to 006 data, and usng the calbrated parameters, we analyze the mpacts of three dfferent lberalzaton polces on welfare and trade flows: tarff lberalzaton, a freely floatng RMB exchange rate, and SOE reform. We note that Chna s economy s qute dfferent from a typcal OECD market economy, and so t s mportant to assess how far the compettve model now wdely used n numercal smulaton work on Chna can be msleadng. In our model, for the manufacturng SOE sector, we use the manageral control model of SOEs n Whalley & Zhang(006) and Fosse & Ramondos-Møller(009).In ths SOEs are assumed to be under manageral control wth managers poltcally apponted. Captal s assumed allocated freely to the SOE by state bank credt ratonng wth a recaptalzaton mechansm coverng losses. We assume that enterprse managers seek to maxmze personal gan from networkng and poltcal connectons. We thus assume that enterprse managers seek to maxmze enterprse sze rather than profts. We add nonneutral monetary structure nto our trade mode as n Wang & Whalley(007), to reflect the actual Chnese exchange rate and monetary regme. In ths model monetary polcy s non accommodatve to the chosen fxed exchange rate, and at 006 exchange rates ths mples excess supply of foregn exchange and reserve accumulaton. Counterfactual model results show that large dfferences n polcy mpacts relatve to a classcal compettve model, ncludng dfferences of sgn. SOE reform and a freely floatng Chnese exchange rate are more effectve polces for mprovng Chna s welfare than tarff lberalzaton. Polces of RMB apprecaton and ncreasng Chna s money stock reduce Chna s trade surplus. In the tradtonal compettve model, trade lberalzaton mpacts both mports and exports, whle n our central case model, wth an endogenously determned trade surplus, trade lberalzaton has lttle effect on exports. Most of the mpact s on mports and the trade surplus. SOE reform of Chna s manufacturng sector sgnfcantly decreases producton of Chna s manufacturng sector and ncreases producton n Chna s other sectors.

23 References Bgsten,A.,X.X. Lu and J.H. Zheng (00), Effcency, Techncal Progress, and Best Practce n Chnese State Enterprses, Chna Economc Quarterly : Cull R., X.L and C. Xu (000), Bureaucrats, State Banks, and the Effcency of Credt Allocaton: The Experence of Chnese State-Owned Enterprses, Journal of Comparatve Economcs. 8: - 3. Fosse,H. and P. Ramondos-Møller(009), Trade Lberalzaton wth State-Owned Enterprses: The Case of Vetnam, Paper Presented at CESfo Global Economy Area Meetng, Feb 7-8, 009. He,F. and Q.Y. Xu (009), Chna s Exchange Rate Regme: Back to the Past. Paper Presented at the th NBER-CCER Annual Conferences, Bejng, July 009. IMF (005), Asa-Pacfc Economc Outlook, Internatonal Monetary Fund, Washngton, DC. Kamada,K., and I.,Takagawa(005), Polcy Coordnaton n East Asa and across the Pacfc, Bank of Japan Workng Paper Seres No. 05-E-4 (Bank of Japan, Tokyo). Lu,D.Q. (00) Managers n State-Owned Enterprses: Ablty, Efforts, and Productve effcency A Case Study of the Iron and Steel Industry n Chna, Chna Economc Quarterly : Marquez, J., and J. Schndler( 006), Exchange Rate Effects on Chna s Trade: An Interm Report, Federal Reserve Internatonal Fnance Dscusson Paper No. 86 (Federal Reserve Board, Washngton). Natonal Bureau of Statstcs, PRC, Chnese Statstcal Yearbook, Park, C.(005), Copng wth Global Imbalances and Asan Currences, Thorbecke, W.(006), How Would an Apprecaton of the Renmnb Affect the U.S. Trade Defct wth Chna? Topcs n Macroeconomcs Volume 6, Issue Artcle 3, Wllenbockel, D. (006), Structural Effects of a Real Exchange Rate Revaluaton n Chna: A CGE Assessment, MPRA Paper No. 90, 90/ Wang,L. and J.Whalley(007), The Impacts of Renmnb Apprecaton on Trade Flows and Reserve Accumulaton n a Monetary Trade Model, NBER Workng Paper No Whalley,J. and S.M. Zhang(006), State-Owned Enterprse Behavour Responses to Trade Reforms: Some Analytcs and Numercal Smulaton Results Usng Chnese Data, NBER Workng Paper No