Existing in all countries where there is more than one level of government，intergovernmental fiscal transfers have attracted a great deal of attention fromeconomists.Much of their discussion has focused on how to justify intergovernmentaltransfers.Three broad reasons have been identified.First，regions with low percapita income and high per capita needs may entail transfers from those with oppositecharacteristics to equalize their fiscal capacity.If collective action problemprevents voluntary inter-jurisdictional transfers from materializing，there isa strong ethical ground to expect the central government to do the job，for a countryshould treat its citizens equally ，no matter which jurisdiction they happen toreside.Second，due to the fact that central government normally have broader taxbases at its disposal ，central transfers may be needed to address the imbalancebetween limited resources available and indispensable expenditure responsibilitiesof sub-national governments.Third，intergovernmental transfers may be needed tocompensate for inter-jurisdictional spillover effects of such regional fiscal operationsas environment protection and promoting education.While such normative deliberationis useful for designing an equitable and effective transfer system，the real worldrarely functions according to abstract ethical principles.In practice，intergovernmentaltransfers are often used to achieve goals other than equalizing fiscal resourcesand outcomes among regions.In other words，political factors may be as importantas，and probably even more important than，ethical and economic considerationsfor policy-makers to allocate transfers.Thus ，simply asking "what ought to be"would not take us very far.Instead ，we should ask，"What really happens".Morespecifically，the real challenge is to explore positively the underlying politicallogic of intergovernmental transfers.Of course ，the political logic of fiscaltransfers is not the same in all systems all the time.In normal situations ，centralpoliticians may use transfers to reward their own constituencies or to "buy"supportfrom those who can be lured to support them ，while regional actors try to lobbyfor as much as central subsidies as possible.Wherever and whenever national unityis at stake ，however，intergovernmental fiscal transfers seem often to assumea different function，one of appeasing centrifugal forces.The most obvious examplein this regard is Germany.East and West Germany were formally unified on October3，1990.However，unification posed a new challenge，the crux of which was aregional disparity.While the "old"Federal Republic had enjoyed a high degree ofinterregional balance ，now it became politically imperative to extend this interregionalequality to the east part and to establish equivalent living conditions throughoutthe "new"Federal Republic.For this reason ，each year after 1990，the federalgovernment has transferred at least DM 150billion，or 5percent of the West GermanyGNP to East Germany.Without such infusion of resources to narrow the overall incomedifferential，the process of German unification would have been much bumpier.Asenormous as it seems，the fiscal sacrifice was a necessary political price to payfor consolidating the newly unified country.The case of Germany is by no meansexceptional.In post-Soviet Russia，for instance ，despite its inability to collecttaxes and to make ends meet ，Moscow still tried to placate separatist demandsby central largess.As a result ，it was those regions that had posed serious threatto the country's political stability and territorial integrity that were rewardedwith larger net per capital central transfers.As Treisman points out ，"the practiceof appeasing mobilized anti-center regions was one reason why ，despite separatistpressures ，economic crisis，and weakened central institutions，Russia did notdisintegrate in the early 1990s as all three other post-communist federations haddone."National unity may come at a high price，but central policy makers everywhereseem to view such price as something worth bearing.This study focuses on the caseof China，as of now，a united country that faces no imminent danger of breakingup.But the country has a tragic past.Since 221B.C.when Qinshi Huangdi firstunited China，the country has been disintegrated as often as it has been united.As recently as the 1910s and 1920s，for instance ，China was divided into feudingwarlord-run kingdoms.The purpose of this paper is to explore how important a partthe concern for national unity plays when Chinese policy makers come to allocatefiscal transfers.We do this with a set of cross-section data that covers China'sall 31provincial units .The technique used is least square multiple regression.Measures of per capita receipts of transfers are regressed upon factors that mayaffect central allocators'decision-making，which in our view is unlikely to bea process free of political influence.Our key hypothesis is that central politicianstend to allocate fiscal transfers according to their ranking order of preferencesand that the top priority for them is to maintain national unity，without whichit is unlikely for them to maintain their personal political preeminence in thecountry.Of course we cannot rule out a priori the possibility that they give firstpriority to assisting regions with poor tax bases and greater socio-economic needs.To ascertain to what extent policy-makers are motivated by a concern for equity ，we include indicators of social needs as part of independent variables in our models.Also included are measures of provinces'bargaining power ，because it is reasonableto assume that all provincial governments want to extract as much central transfersas possible.This study finds strong evidence that national unity is a far moreimportant concern than interregional equity when political elites come to make decisionsabout the allocation of fiscal transfers.The paper is organized as follows.SectionI provides estimates of fiscal disparities in China.Section II offers an overviewof China's transfer system and defines the dependent variable.Section III listspossible determinants of intergovernmental transfers，including our test variablesand some control variables.Section IV presents the results of our regression analyses，which is followed by Section V that discusses those results.The final Section discussesthe implications of our empirical findings.

　　I.Fiscal Disparities in China

　　Intergovernmental transfers can be justified only if there are substantial gapsin capacity and costs to produce a standard package of public services between sub-nationalgovernments.Presumably ，the larger the gaps are，the more indispensable thefiscal transfers become.Covering 9.6million square kilometers ，China is thethird largest country in the world.Due to its gigantic size，there have alwaysbeen significant spatial variations in geographical conditions，resource endowments，the sectoral distribution of economic activity，and the level of socio-economicdevelopment.Given the regional economic inequality ，fiscal disparities are inevitable：lower income regions may not be able to provide their residents with a standardpackage of public services that are taken granted by the residents of high incomeregions.In such a circumstance ，poor regions may seek support from rich regionsto reduce inequalities in public expenditure，be it directly or via transfers fromhigher levels of government.

　　[Table 1about here]

　　How large are fiscal disparities in today's China ？Table 1may give us somehints.For 1998，variations in per capita GDP appeared to be fairly large acrossChina's 31provinces.While Guizhou ，China's poorest province ，registered aper capita GDP（2，342Yuan）of 31.8percent of the national average（7，373Yuan），Shanghai ，China's leading industrial and commercial center ，enjoyeda per capita GDP（28，253Yuan）nearly 4times higher than the national averageand more than 12times that of Guizhou.Relative differences in per capita budgetaryrevenue were even more striking.As Table 1reveals ，the coefficient of variation-ameasure of dispersion equal to the standard deviation divided by the mean-of percapita revenue was much higher than that of per capita GDP.It is worth noting，however ，that ，compared to the inequality of per capita revenue ，inequalityof per capita expenditure was significantly smaller ，about in the same magnitudeas inequality of per capita GDP.Clearly，there were some central transfers thathelped to reduce provincial fiscal inequality.Thanks to central transfers，gapsin three key areas of public spending （education ，administration ，law and order）were not as large as income gap.But in two other key areas of public spending（health and capital construction ），gaps are larger.In any event，inequalityof per capita budgetary expenditure in all areas was pretty large by internationalstandards.Table 1seems to suggest that，while central transfers in China wereprogressive ，they were either not progressive enough or they were too small insize to alleviate inter-provincial disparities.Thus，it is natural for us to askwhich regions tend to receive more net transfers from the central government andwhy.But before trying to answer this question，a brief introduction about thechannels of intergovernmental transfers is in order.

　　II.Intergovernmental Transfers in China

　　During Mao's era，the central government enjoyed considerable control overthe distribution of resources.The fiscal system was so arranged that rich provinceshad to remit large proportions of their revenues to the central government and poorprovinces were allowed to "retain all their revenues and receive additional directsubsidies from the central government."Acting as a redistributor in between，thecentral government could use fiscal transfers to influence behavior of sub-nationalgovernments.In the first 15years of economic reforms，however，the Chinese centralgovernment's ability to allocate transfers was critically enfeebled.Under the fiscalcontract system prevailing from 1980to 1993，all the provinces，rich and poor，were compelled to become financially more independent.Most taxes were collectedand most expenditure undertaken on a jurisdiction-by-jurisdiction basis.As fiscalsurpluses from rich provinces drained off ，what was available for the centralgovernment to redistribute became increasingly limited.The result was growing fiscalgaps between provinces.With larger tax bases ，rich provinces could afford eitherdirectly invest more with their budgetary capital or offer more generous tax concessionsto potential investors.The same fiscal autonomy，however，worked to the detrimentof poor provinces ，because，with less central subsidies，they were hard up evenfor resources to support daily government operation and to provide such basic servicesas health and education ，not to speak of conducting productive investment.Bythe early 1990s ，there was a growing concern that a widening income gap betweenthe prosperous coast and the laggard interior might eventually cause the break upof the Chinese national state.To arrest this dangerous trend ，China overhauledits fiscal system at the beginning of 1994.The new system was called the tax-assignmentsystem（fenshuizhi）。One of the manifested goals of the 1994reform was to restorethe central government's redistributive ability so that it could again transfersurpluses extracted from more developed provinces to less-developed provinces.Thepost-1994Chinese fiscal flows between the central and provincial governments canbe divided into four broad categories.

　　Returned revenue：Because strictly following the new rules would lower therevenue income of every province，at the time when the 1994system was introduced，the central government made a pledge of compensating each and every province forwhat it would have to sacrifice for accepting the new system.For this purpose，each province's net loss in accepting the new system was calculated.Thereafter ，every year a province was supposed to receive a central compensation（or "returnedrevenue"）amounting to ：Rt =R0*（1+0.3*Gt）t where Rt is the central compensationin year t ；R0is the compensation baseline or the calculated net loss of the provincefor the first year；Gt is the average growth rate of VAT and consumption tax inthe province in year t；t represents the first ，second ，third……year afterthe introduction of the new system.Since the mechanism of "returned revenue"wasprimarily designed as a kind of side payment for the provinces not to resist thenew system，its distribution was not expected to conform to the equity principle.

　　Old system subsidies（or remittances）：Even after 1994，part of the revenue-sharingcontracts negotiated under the old system remained effective.The provinces continuedto remit a certain amount of their locally collected revenues to，or receive acertain amount of subsidies from，the central government as they had under thepre-1994regime.The only difference was that now the amounts of remittance or subsidieswere fixed once for all.Normally ，a province that received subsidies did nothave to remit to the central coffers.Shandong was the only exception.Among theother 30provinces，sixteen were on the recipient side and fourteen on the remittingside.The former group included all the eight provinces where minority nationalitieswere concentrated （Tibet ，Xinjiang ，Inner Magalia，Ningxia，Guangxi，Qinghai，Yunnan，and Guizhou）and other poor provinces such as Sichuan and Jiangxi.Richprovinces such as Shanghai，Beijing，Guangdong，and Liaoning all belonged tothe latter group.Thus，this mechanism contained elements of fiscal equalization.But ，the significance of this mechanism will diminish ，because，as time goesby，the relative size of such transfers will become smaller and smaller in theever growing public finance.

　　New system subsidies（or remittances）：They were "new"because they were introducedafter 1994.All the transfers in this category were for specific purposes ，suchas disaster relief，subsidies to certain regions （e.g.，Beijing，Chongqing，Xinjiang，and Yan'an ），subsidies to certain projects（e.g.，education，environmentalprotection，and industrial restructuring ），and the like.

　　"Transfer of the transition period"：First introduced in 1995，this transferwas specifically designated to address horizontal fiscal imbalance.Unlike other"new system subsidies"，this was the only formula-based type of transfer ，theallocation of which is determined by objective measurements of fiscal capacitiesand fiscal needs of the provinces.

　　Table 2presents data on central-provincial fiscal flows in 1998.The net centraltransfers to the provinces can be calculated by subtracting provincial remittancesof taxes to the central coffers from the total of all central transfers to the provinces.In 1998，this figure amounted to 272.4billion Yuan ，which accounted for abouta half of the central revenue ，or 3.4percent of China's GDP.However ，the bulkof the central transfers took the form of "returned revenues，"kind of de factoprovincial entitlements about which the central government could exercise littlediscretional power.If the "returned revenue"were to be excluded ，the net centraltransfers were barely 65billion Yuan.As for the most redistributive "transferof the transition period，"its size（6billion Yuan）was too small to be significant.

　　[Table 2about here]

　　Given the heterogeneous natures of different forms of central subsidies ，itmay be useful to distinguish two concepts of net central transfers，with one including"returned revenue"and the other excluding "returned revenue".

　　Net Transfer I =Returned revenue +Old subsidies +New subsidies +Transferof the transition period -Old remittances -New remittances

　　Net Transfer II =Old subsidies +New Subsidies +Transfer of the transitionperiod -Old remittances -New remittances

　　The dependent variables in this study are per capita net transfer I and percapita transfer II.III.Determinants of Intergovernmental Transfers

　　In theory ，central transfers should be designed primarily to equalize fiscaloutcomes or resources among provincial units.But in practice ，political factorsmay matter in the process of transfer allocation in China every bit as they do underelectoral democracies.Even though Chinese national policy makers do not have tostand for election and re-election，they ，just like their counterparts elsewhere，care a great deal about their political legitimacy and survival.So do provincialleaders.Preoccupied with unusual concern for political stability ，Chinese centralpoliticians tend to use fiscal transfers to award prospective supporters and/orto neutralize potential threats.In the same vein ，provincial governments areinclined to extract as much central transfers as possible so as to please theirown constituents.The allocation of fiscal transfers thus often becomes a focusof contention in intergovernmental relations.For this reason ，to figure out thekey determinants of central transfers in China，one has to consider not only ethicaland economic factors but also political factors.Accordingly，three categoriesof independent variables are used in this study.

　　Central decision makers'concern for equity "Per capita GDP.This is a proxyfor fiscal capacity ，that is，the capacity to raise revenue.A major source offiscal disparity arises from asymmetries in fiscal capacity ，which in turn arisesfrom asymmetries in income distribution.The lower the per capita GDP ，the lowerthe fiscal capacity.If central decision makers'objective in allocating transferswere to mitigate fiscal disparities and to level the fiscal playing ground，a negativeregression coefficient would be expected."Per capita cost of natural disaster.China is a country that is prone to natural disaster.In each particular year ，however ，some areas may incur greater losses than others.Central decision makersare supposed to take this factor into consideration when they allocate fiscal transfers.If they do，we expect a positive coefficient."The share of agriculture in theregional economy.This is an indicator of underdevelopment.If central transferswere targeted to subsidize underdeveloped regions ，one would expect a positiveregression coefficient."Dependency ratio.Such ratio refers to the share of thepopulation that is below and above working age.It serves as an indicator of fiscalneed.Regions with a larger dependent population are expected to receive largerper capita transfers."Population density.This is a proxy for the unit costs ofsocial services.Fiscal disparities can be the result of differences in revenuebases as well as in the unit cost of provision.There may be many environmentalfactors affecting costs of provision.Population density is just one of them.Presumably，the lower the population density in a region，the higher the unit cost of deliveringany particular level of social services to the population.If central policy wereto compensate cost differences，one would expect a negative coefficient.

　　Central decision makers'political concerns "The proportion of minorities inthe population or whether an area is an autonomous area.Regions with a high concentrationof non-Han inhabitants might be thought more likely to cultivate separatist aspirations.If a significant regression coefficient is found，we may conclude that the centralgovernment uses transfer as an instrument to pacify the most potentially troublesomeareas."The instances of labor disputes.This is an indicator of social stability.If social stability were a major concern for central leaders，one would expecta positive regression coefficient.

　　Regions'bargaining power "The representation in the Politburo.This is a dummyvariable，which take the value 1for all provinces that have a representative inthe Politburo ，the most powerful decision-making organ in China ，and 0for allothers.A significant positive regression coefficient on this variable would suggestthe existence of pork-barrel allocation in China."The size of the population.Beijing might be more likely to yield to pressure from a province with 50millionor more population than from a province with less than 10million."The size ofthe economy measured by provincial GDP.This is another proxy for bargaining power.Provinces with greater economic power are expected to receive higher per capitatransfers due to their political weights.

　　IV.Results

　　In this section ，we conduct regression analyses on central-provincial transfers.Dependent variables are per capita transfer 1and per capita transfer 2as definedin the section III.Independent variables are those listed in the previous section.Since there is no explicit model，regressions reported below should be seen asexploratory rather than an attempt to test a theory.They are used to explore towhat extent fiscal transfers in China are governed by economic and ethical considerationsand to what extent political considerations override economic and ethical considerations.Table 3presents the results of linear regressions for each of the two dependentvariables.In both cases，columns marked （a ）offer estimated regression coefficientswhen a broad range of theoretically relevant predictors are included，while columnsmarked（b ）give results for short regressions which exclude all independent variablesthat do not significantly improve the fit of the regression ，as judged by an F-testat the 0.10level.

　　[Table 3about here]

　　Regression（1a）in Table 3presents the results for per capita transfer 1asthe dependent variable.Three observations can be made.First ，the equity considerationdoes not appear to have played any clear role in allocating transfer 1.Four ofthe five variables in this regard have the "wrong"signs.Transfer 1is positivelyrelated to "per capita GDP"and "population density ，"and negatively related to"the share of agriculture in the economy"and "dependency ratio."In other words，the richer the province and the higher the population density ，the more the percapita transfer.And a province with less developed economy and a larger dependentpopulation tends to receive smaller rather than larger per capita transfer.In anyevent ，none of the five variables are statistically significant.Second ，pressurepolitics does not play a major role in central allocation ，either.Two of thethree variables measuring bargaining power carry the "right"signs.Membership onthe Politburo and GDP （which measures the economic importance of a province）arepositively related to per capita transfer.But neither is even close to statisticallysignificantly different from zero at conventional level.As for another measureof provincial bargaining power，population turns out to be negatively related toreceipts of transfer.Less populous provinces tend to do better in per capita terms.But this variable is not statistically significant.Third ，the most dramatic resultof Regression （1a）is the high significance of two variables measuring centralpolicy makers'political concerns.The ethnical make-up variable has a positivecoefficient that is statistically significant （significant at p =0.000）。Thehigher the proportion of the non-Han population a province is ，the more per capitatransfer it receives.Interestingly ，per capita transfer is negatively correlatedwith the variable that measures social stability（"the instances of labor disputes/thepopulation"）。Rather than to calming down less stable provinces by allocatingmore per capita transfers ，such provinces are actually penalized.Central policymakers seem to bias allocation in favor of those provinces that are able to maintainsocial stability.Indeed，it may be politically unwise to reward trouble provinceswith financial concessions.After excluding variables that fail F test，Regression（1b）retains only two variables.The provincial ethnic composition remains highlysignificant.The other variable is per capita GDP ，which is again positively relatedto per capita transfer.This is not entirely surprising given the fact that "returnedrevenue"makes up more than three-fourths of transfer 1.As pointed out above ，the "returned revenue"is the remnant of the pre-1994fiscal system ，the distributionof which by definition is not governed by the equity principle.Thus，rich provincesare able to grab more rather than less per capita "returned revenues"from Beijing.Now ，let's turn to transfer 2.Clearly Regression （2a）more or less resemblesRegression（1a）in all aspects except one，namely ，the sign of the coefficientfor per capita GDP.While per capita GDP is positively related to per capita transfer1，it is negatively related to per capita transfer 2.This implies that concernof equity plays a greater role when policy makers come to the allocation of transfersother than "returned revenues."But the negative relationship between per capitaGDP and per capita transfer 2is not statistically significant.Remarkably，thetwo measures of political concerns remain highly significant in Regression（2a）。So are they in Regression （2b）。This is impressive evidence that policy makersare motivated by political concerns more than anything else in allocating transfer2.As will be shown in the Appendix ，provincial governments in China follow thesame political logic when they allocate fiscal transfers to county governments.In sum，from the regressions reported in Table 3，we may draw three broad conclusions.First ，equity consideration seems to have played little role in central-provincialtransfers.Second ，contrary to the conventional wisdom，there is little scopefor bargaining as far as fiscal transfers are concerned.Third，the most importantdeterminant of fiscal transfers turn out to be policy makers'political concerns，in particular ，their concern over national unity.All the four regressions inTable 3reinforce each other and point to this conclusion.

　　V.Discussion

　　Fiscal transfers exist in all countries where there is more than one level ofgovernment.However ，the key determinants of fiscal transfers differ from countryto country.Some countries（e.g.the United States）favor conditional（or specificpurpose ）grants ，while others （e.g.Canada ）prefer unconditional （or generalpurpose ）grants.Where unconditional grants dominate and transfers are calculatedaccording to certain elaborate formulas ，the possibility of political interventionis reduced to the minimum （e.g.Germany）。When rules for allocating general-purposegrants are vague，there is much room open for bargaining （e.g.India and Russia）。Systems dominated by conditional grants provide a perfect institutional environmentfor pork-barrel politics（e.g.the United States）。

　　In China，general-purpose transfers account for the major part of the totaltransfers.However，the bulk of general-purpose transfers in China consist of "returnedrevenues，"the distribution of which is generally made according to certain formulas.Even though those formulas do not embody the equalization principle ，neverthelessthey limit the scope of bargaining.So are old subsidies and remittances.No wonderthat all of our variables measuring bargaining power add little to the explanation.As pointed out above，the "returned revenues"are designed to recognize the vestedinterests of the localities ，rather than to address the issue of growing regionaldisparity.That explains why it is hard to find evidence that the central distributionof transfers favors low-income areas.In a sense，it seems fair to say that，asfar as the allocation of fiscal transfers is concerned，there is not much roomfor central policy makers to act at will.Of course ，we are not suggesting thatthey are powerless.The allocation of some fiscal transfers is surely still withintheir discretion.The question is what would be their priorities in allocating fundsunder their control.The analysis in the previous section suggests that nationalunity is probably their number one concern.This is by no means a new finding.Ina study of fiscal transfers during the period of 1978-92，Martin Raiser alreadyfound that the poor provinces had all received some subsidies but the poorest provinceshad not necessarily received the highest levels of subsidies.Rather it was provinceswith predominantly non-Han population that had been given the highest levels ofsubsidies ，even though their income levels exceeded those of the poorest provinces.This finding led him to suspect that fiscal transfers in China might have been motivatedlargely by political concern rather than by concern over equity.This study providessystematic evidence to support his conjecture.Why does national unity figure soprominently in the allocation of fiscal transfers in China？

　　China is a multi-national country with 56nationalities.The majority nationality，the Han ，comprise 91percent of the population.The rest of the population isdivided into 55minority nationalities.However ，the minorities hold an importancefor China's long-term development and national security that is disproportionateto their population.First，although small in proportion ，China's minoritiesare large in number.Accounting for about 9percent of the population ，the absolutesize of China's minority nationalities actually exceed those of such large Europeancountries as Germany，France ，and the Great Britain.Few people realize thatChina has one of the world's largest Muslim populations ——nearly 20million ，more than the United Arab Emirates，Iraq ，Libya，or Malaysia.Second，the minoritynationalities have grown at much higher rates than that of the Han.While the Hanpopulation grew a total of 10percent between 1982and 1990，the minority populationgrew 35percent overall ——from 67million to 91million.A sample survey conductedamong one percent of the total population in 1995showed that 108.46million peoplebelonged to minority ethnic groups，accounting for 8.98percent of the country'stotal population of more than 1.2billion ，a 0.94percentage point increase overthe figure in 1990.If the minority populations'growth rate were to continue ，it is estimated that they could total 864million by 2080.Third，the nationalminorities are scattered over vast areas.There are minorities in every province，autonomous region and municipality directly under the Central Government，and inmost county-level units.But minority nationalities are mostly concentrated in WestChina ，which spans nearly two-thirds of the country's landmass.Fourth，the areasmost heavily populated by minorities happen to be resource-rich areas.NorthwestChina has plenty of coal，China's chief fuel ，and Southwest China has unlimitedhydroelectric potential.Qinghai's Qaidam basin has quantities of natural gas ，and Xinjiang's Tarim basin holds both oil and gas.Gold and other metals are plentifulin Tibet，Guizhou and Yunnan.Energy is a bottleneck for China's future development.West China seems to hold the key.Finally ，China's ethnic groups live mostly alongthe borders ，such as the Mongolians in the north，Uygurs and Tibetans in thewest，and the Zhuang ，Yi ，and Bai in southwestern China.Or put differently，China's border areas are mostly inhabited by minority peoples.In counties and villagesalong many border areas of Xinjiang ，Tibet，Inner Mongolia ，Yunnan and Guangxi，over 90percent of the local population belong to minority groups.From a securitypoint of view ，those areas have enormous strategic value as China's outposts fornational defense.In recognition of the minorities'official status as well as theirstrategic importance，China has created four levels of autonomous administration，including five regions，30prefectures ，120counties （or，in Inner Mongoliaand Manchuria ，banners），and more than 1，200villages.Although such autonomousareas may not enjoy true political control over local affairs ，it is neverthelessdesirable for minorities to obtain the autonomous status，for groups identifiedas minorities can receive real benefits from the implementation of various affirmativeaction programs.The most significant privileges included permission to speak andlearn their native languages，worship and practice their religions ，express theircultural differences through the arts and popular cultures，have more children ，obtain better opportunities for their children's education，have greater accessto public office，and above all pay fewer taxes and receive more subsidies.Priorto 1980，the dependence of autonomous areas on fiscal transfers was very high.So was the dependence of poor areas that were inhabited by the Han Chinese.Sincethe early 1980s ，however，the magnitude of fiscal transfers has substantiallydiminished.As a result ，poor areas no longer received as much transfers as before.But there were exceptions.As Martin Raiser noted in his 1998study ，even beforethe 1994fiscal reform，all of the large transfer recipients were provinces withbig minority populations such as Tibet，Xinjiang ，and Qinghai.This study showsthe continuity of such distributional pattern.What motivates China's policy makersto favor minorities is above all a strong desire to maintain national unity.Deepin the Chinese psyche there is a great fear of the country breaking up.Since 221B.C.when Qinshi Huangdi first united China ，the country has been disintegratedas often as it has been united.As recently as the 1910s and 1920s，for instance，China was divided into feuding warlord-run kingdoms.And the Mainland China didnot become fully united until the early 1950s.Central political leaders thus arewilling to maintain the unity of the nation at all costs.China ，as of now，isa unified country and there is no imminent danger of any sort for China to breakup again.But China's top leaders do not want to leave it to chance.What worriesChinese political leaders is twofold.First ，there have been persistent economicgaps between coastal regions where the Han Chinese dominate and western regionsthat are most heavily populated by minority nationalities.China's market-orientedreform only widens the existing gulf.With little sign that eastern prosperity hastrickled west ，central decision makers have reasons to worry that such disparitiesmay fuel age-old resentments along ethnic ，linguistic ，and cultural lines.Second，separatist activities and ethnic unrest in some of China's border areas ，especiallyXinjiang and Tibet，constitute another evil omen for those holding the reins inBeijing.In particular，the creation of several new nations on China's CentralAsian frontier has risen Chinese concern over the influence of separatist sentimentspilling over from the other side of the border into China's Muslim areas.Beijing'schallenge is to convince China's minorities that they will benefit more from cooperatingwith their national government than from breaking away.To do so，central policymakers have to take into account the interests of minority nationalities.Fiscaltransfer is a political instrument to appease ethnic or religious independence sentiments，real or potential.This is something that the Chinese government is willing andeven happy to admit.In fact，whenever the issue of fiscal transfers comes up，the central government never tires of emphasizing that ethnic regions deserve preferentialtreatments.So do provincial governments.

　　[Table 4about here]

　　Political consideration underlying fiscal transfers is evident in Table 4.Percapita locally collected revenue is generally lower in eight minority-concentratedprovinces than in the other provinces.Thanks to central transfers，however，percapita expenditure is generally much higher in those eight provinces than in therest of the country with exceptions of Beijing，Tianjin，Shanghai and Guangdong.More interestingly，when allocating fiscal transfers ，the central governmentseems to have given first priority to those provinces most susceptible to ethnicseparatism，namely ，the provinces mainly inhabited by Tibetans and Moslems（Tibet ，Xinjiang ，Qinghai，and Ningxia）。On the other hand，provinces whereminorities make little trouble（e.g.Guangxi and Guizhou）do not seem to have foundmuch favor with Beijing as far as fiscal transfers are concerned.

　　Conclusion

　　Fiscal transfers are supposed to be allocated according to the equity principle.But politicians often use transfers for pursuing other political objectives.Thisstudy reveals that fiscal transfers in China follow considerations other than regionalequity.It is central policy-makers'concern over national unity that seems to havedominated intergovernmental transfers.As a result，an area with large non-Hanpopulation tends to receive much more per capita transfer than one with predominantlyHan population，even if the former has higher per capita GDP than the latter.Forinstance，there were ten provinces whose per capita GDP were lower than Xinjiang's，but Xinjiang received a level of per capita transfers that was several times higherthan those ten provinces（Table 4）。The only explanation for such a large discrepancyis Xinjiang's predominantly non-Han population.More striking is a comparison betweenShaanxi ，Gansu，and Ningxia（In China，they are often lumped together and calledthe Shaan-Gan-Ning region ），three neighboring provinces that resembles one anotherin almost every aspect except one ：Ningxia is inhabited by a large number of ChineseMoslems whereas Shaanxi and Gansu are the legendary birthplace of the Han Chinese.That difference seems to be crucial in explaining why Ningxia received far moreper capita transfer than its two neighbors，despite the fact that its per capitaGDP is higher （Table 4）。

　　Does this finding imply that Beijing care little about the issue of regionalinequality as such？Probably not.After all，growing regional disparities evenamong Han-dominated provinces may undermine the country's political instability.The problem is that the Chinese central government simply does not have much fiscalresource at its disposal to address the issue.In 1998，its revenue（includingprovincial remittances）accounted for barely 7percent of GDP，far lower thanthe ratio prevailing in most countries（20percent and above）。Even such smallrevenue was not all within central discretion.The 1994fiscal regime obligatedthe central government to "return"a substantial share of its revenue （38%in 1998）to the provinces.What left fell far short to cover the central government's ownbasic expenses.In order to make transfers to the provinces ，the central governmenthad to run deep into debt.The weak extractive capacity forced the Chinese centralgovernment to carefully order its priorities.What Beijing deems the most fearsomeprospect is when regional disparities overlap ethnic and cultural distinctions.Therefore the top priority in allocating limited fiscal transfers was given to appeasingethnic or religious groups，at the risk of upsetting poor provinces with predominatelyHan population.In this sense ，fiscal transfers indeed represent a price of nationalunity.National unity requires a price，but，in the view of Chinese policy makers，the price is one worth bearing.

本文版权为文章原作者所有，转发请注明转自当代文化研究网：http://www.cul-studies.com