tas, plant capacity, tax rates, and material supplies. The fundamental flaws of the system are not overcome (Kornai, 1991).

There are two assumptions behind this analysis, neither of which receives the same kind of attention as the relationship between the state and the enterprise, which is the centerpiece of the argument. The first is that the budget constraints on the state itself are relatively weak: the state can manipulate the money supply or the price structure, run a foreign trade deficit, or borrow internationally to cover any shortfalls (this of course cannot work in the long run, as Eastern European economies found in the 1980s). The second assumption is in fact the reason why the state bothers to bargain with firms at all—why can it not just say no? The reason is structural. While some enterprises may provide crucial inputs for other national firms that would otherwise have to purchase their supplies abroad, all enterprises meet central nonfinancial needs for the state: they maintain full employment and fund housing and social services for large numbers of employees. As a result, very few firms are ever closed, and the large industrial complexes, often among the biggest money losers, are virtually never closed. To do otherwise would be to create severe social problems. The nonfinancial interests of the state in public firms prevent the imposition of hard budget constraints and therefore lead partial reforms into an impasse of constant bargaining, with concealment of financial resources as the central strategy—a reproduction of the central flaws of the system.

If we conceive of the state not as a single redistributive system (a not unreasonable assumption for Hungary), but as a hierarchy of thousands of such redistributive systems, and if we consider the effects of fiscal decentralization on revenues and bureaucratic incentives across levels of this hierarchy, we can see the possibility of different results. Instead of the system being shifted into a new yet unsatisfactory equilibrium, we see the processes that can lead to the dynamic yet gradual transformation of the behavior of bureaucrats and thereby the premises of the entire system. While systematic econometric evidence pertaining to some aspects of this transformation has only recently appeared, interview studies and field observations suggest strongly that this is precisely what has been happening in China.

The first step in the analysis of this dynamic transformation is to examine the impact of fiscal reforms across levels of a system in which only the small number of higher jurisdictions had the features posited by Kornai. At the pinnacle of this hierarchy in 1985 was the central government, a redistributive economy of 3,800 large industrial enterprises, each of which employed an average of 2,270 people. The revenues of the central government came directly from these large firms, which produced 20 percent of public-sector