VOL &8 NO 5 OIAN AND ROLAND: FEDERALISM AND THE SOFT BUDGET CONSTRAINT 1147 Horioka(1980), extensive evidence has ac- The Go cumulated indicating substantial capital im- We consider the government budget at date mobility across countries in general (e. g 1. In each region, on the revenue side, Ti-Ti Gordon and A Lans Bovenberg, 1996). This is the net tax revenue exclusively from state includes a high correlation between domestic enterprises, where T, is total taxes and Ti is vings and investment, real interest differen- transfers to state employees. On the expendi tials across countries, and a lack of interna- ture side, the budget is used for three purposes: tional portfolio diversification. In contrast, subsidies (or bailout)S: for type 2 projects, xample, Tamim A Bayoumi and Andrew K. public goods provision 4, ent Ii, and local capital mobility within countries is high. For public infrastructure investm Rose(1993)found strong evidence of capital mobility across regions in the United Kingdom T-T;=S+1;+z using the Feldstein-Horioka methodology But international capital mobility to China Actual tax revenues and expenditures will is particularly imperfect for idiosyncratic rea- depend on whether the budget constraints of sons. Unlike most Organization for Economic enterprises are hard(action en and no bailout ooperation and Development(OECD )and of type 2 projects)or soft(action e; and bailout newly industrializing countries, China has of type 2 projects). We can rewrite the gov- maintained formal control over its capital ac- ernment budget as count. Officially, capital is not free or out of China: Capital inflow to China is con- Ti+l+z=T-s=e, trolled as well as repatriation of profits. In fact, it was not until December 1996 that the Chi- where E= EH= nR, or E=ES= anR nese government announced the current ac-(1-a)n, depending on whether budget con count convertibility of its currency, but it has straints are hard or soft no plans to make capital accounts convertible We assume no tax on returns to nonstate in the near future. These imperfections inin- capital This can be justified by the difficulty ternational capital mobility need to be con- of such taxation in practice. It is difficult to tax tal domestic nonst apital in transition econe mobility within China. Once a foreign firm mies due to the lack of good tax institutions manages to invest in China, it can freely Indeed, the great majority of taxes in China hoose its location. The only consideration is were collected from the state sector. It is also profitability. Moreover, international investors well known that foreign firms are able to evade must incur some costs before investing in income taxes in host countries by transfer pric China to overcome language barriers and ac- ing schemes. This is true in developed coun- quire knowledge of Chinese laws, etc. Once tries and more so in developing countries like these costs are sunk, it is much less costly to China. However, in Appendix a we show that move capital from, say, Guangdong to our analysis goes through and our results con tinue to hold even with taxation of returns to nonstate capital In this regard, it is useful to make a comparison be The Government's Objective Function We assume a welfare-maximizing tries such as Mexico, Thailand, Malaysia, Indonesia, and ment: The local government in region imizes w and the central gover ently. Unlike those countries, foreign capital is not al- maximizes w=2 w. We assume that the wed to go into China freel short-term and portfolio investment opportunities in China re not available to foreign investors. Such restrictions on eign capital are responsi the latter exerts an extemality on other regions to avoid financial crises like those in Mexico and Asian former does not. whether z or consumption is irrelevant for our res
148 THE AMERICAN ECONOMIC REVIEW DECEMBER /998 population in region i consists of two groups: er and the government chooses bailout, and employees in state enterprises and in nonstate hard budget constraint equilibria as equilibria firms. Let the total private benefits of the em- in which enterprises with type 2 projects ployees in nonstate firms be choose e and the government chooses x;=x(K, l)=f(Ki, l)-Ki(Ki, Ii), The Interaction Among Local Govern where x, is assumed to be an increasing and ments -Except for the case of complete cen- concave function of I,. The total private ben- tralization, there is a competition between efits of the employees in state enterprises are local governments at date 1 to attract foreign yu,where y =nBa Ti under hard budget capital and grants from the central govern- constraints and y; =anBa +(1-a)nB, + Ti ment. Local governments move simulta under soft budget constraints. The total utility neously and make choices of budget of the two groups derived from local public allocation. The equilibrium is a Nash equilib goods is u(z), which is assumed to be con- rium in which each local government takes the cave in budget allocation of other local governments Under the interpretation of K as domestic as given nonstate capital, the objective function of local government in region i is expressed as The interaction between the central and Local Governments.-The interaction between W;=f(K,)+y;+u(z;) the central and local governments at date I con- cerns allocation of grants and monetary creation assuming the government puts an equal weight (except for the cases of complete centralization between rents to nonstate employees and returns and complete decentralization). This is a se. to nonstate capital Under the interpretation of K quential interaction between the central govern as foreign capital, the objective function of local ment and local governments. We will consider government in region i becomes the case in which the central government makes decisions(such as allocating grants and creatin W=x(K,l)+y2+(x) money ) after the local governments'decisions which reflects the fact that the central govern- because the government is not concerned with ment often accommodates the local govern returns to foreign capital. The two specifica- ments'needs. We will also consider the case of tions will give qualitatively similar results; for no accommodation in which the central govern- concreteness we will use the interpretation of ment moves before local governments foreign capital We define an equilibrium of the above ga as strategy profiles of state enterprises, local equilibrium overnments, and the central government We define the game played under federal such that none of the parties has incentives to institutions as follows deviate unilaterally. Our analysis is consider ably simplified because whether an enterprise The Interaction between governments and has a soft or a hard budget constraint is deter- State Enterprises.In this sequential interac- mined by a simple comparison between the tion an equilibrium is defined by: (i)a choice benefits of bailout (Bs)and the opportunit of action e of state enterprises at date O to max- cost of doing so. Soft budget constraints are imize private benefits given the expected obtained when the benefits exceed the cost of choice of Ti, Si, I,, and zi by the government; doing the best alternative. Hard budget con- and (ii) the governments choice of Ti, Si, I straints are obtained otherwise nd z, at date 1 in order to maximize social en the choice of action by state Il. Fiscal Competition as a Commitment Device enterprises. We define soft budget constraint game-perfect equilibria in This section demonstrates a main theme of which enterprises with type 2 projects choose our paper: Fiscal competition among local