This article deals primarily with the complexity of the decentralization problem, outlining what needs to be taken into account if decentralization is to be accomplished successfully in developing countries. With decentralization currently occurring by deliberate design, political necessity or default, the range of possible outcomes has neither been realized nor envisioned.

The distinction between decentralization in developing countries and their developed counterparts is taken as twofold. First, many of the assumptions made in literature on decentralization (in industrialized nations) such as the presence of voice and exit do not hold in developing countries. In developing countries, exit is often constrained by lack of mobility (poor information, weak markets for land, labor and capital and risk aversion due to the absence/inadequacy of social safety nets). Voice is constrained by problematic electoral systems. Both lead to a lack of accountability on the part of the government.

The second chapter goes to great length to point out that decentralization as a concept/ideology is neither inherently good nor bad, but is one of many tools for institutional restructuring. There is no decisive consensus on what the effects of decentralization will be. At its best, decentralization can affect equity (through allocation) and efficiency increases, as well as macroeconomic stability. At its worst it can fail to improve service delivery, promote the risk of national destabilization and induce undesirable second and third tier effects.

The debate on whether decentralization is good or bad is simply unproductive: decentralization is a political reality in many developing countries, as opposed to an option for deliberation. Thus, it is equally important to realize that the complexity of the problem does not lend itself to a simple, unified solution but calls for appraisal on a case by case basis. The search for solutions is limited by the lack of empirical evidence.

How then to proceed? Chapter 3 deals with the specific institutional design framework needed to address the problems of exit, voice and accountability. Fiscal federalism provides such a framework. Fiscal federalism is the assignment of expenditure responsibilities to each tier of government, often facilitated by intergovernmental transfer of funds. It is premised on the idea that decisionmaking should occur at the lowest level of government consistent with allocative efficiency (chpt 3, p.10). This would dictate the optimal size for the jurisdiction of each service and service component. The authors provide examples of services, such as water supply and sewerage, for which there are roles and implications for different tiers of government.

To assign the appropriate tier of government, it first must be determined what outcomes are deemed desirable for the central government and how they are to be accomplished (e.g. direct provision, delegation, etc). Control of those issues that are determined to be of less importance to the central government should be relinquished to local authorities. Although detailed control over local use of funds is not desirable, some sort of monitoring is necessary to ensure the quality of service provided meets standards.

Raising revenue is part of this equation in that lower-tier governments are often unable to finance the expenditures for which they are responsible with own-source revenues. Three methods to bridge the gap between revenue and expenditure are: intergovernmental transfers, provision of taxation power for subnational governments, and subnational borrowing.

Allowing subnational governments to levy taxes is aimed at burdening local residents for the benefits they receive and can take two forms: a retail sales tax or a personal income tax. Litvak et al rule out a retail sales tax, citing administrative difficulties. The personal income tax is given more weight and is considered effective only when local authorities retain the ability to set the rate.

Intergovernmental transfers provide a more economically acceptable if logistically more difficult option. Transfers can be broadly divided into non-matching (lump sum) and matching transfers, the former further divided into selective (conditional) and general (unconditional) categories. Matching transfers require that funds be spent on specific purposes and that the recipient to some extent matches the grant. These transfers have the potential to distort local priorities and favor more affluent jurisdictions. Non-matching conditional transfers require the recipient spend at least the full amount received on the designated function. Problems of substitution (fungibility) come up when money that would have been spent on the designated function is diverted once the grant has been received, making the impact of the grant itself unclear. Unconditional non-matching grants do not have a specific use and therefore their impact varies. Subnational borrowing is considered a last resort due to the chance of increasing the pressures of cyclical borrowing.

If key services are provided through decentralized governments, there must be methods of ensuring that (1) the price facing service providers is right, (2) the targeted groups are receiving the desired services at an acceptable level of quality and (3) non-compliance is dealt with in an equitable way. This will undoubtedly necessitate different tiers of government having common goals.

Chapter 4 addresses how institutions organizations, the rules they follow and the means of enforcement determine the success of decentralization. The authors focus on the regulatory framework and decentralized borrowing, the organization of service delivery, and the establishment of information systems and competitive governments.

The regulatory framework and decentralized borrowing are critical parts of decentralization. There are good reasons for allowing sub-national governments to finance their own investments by borrowing, but without the proper rules such borrowing can get out of hand. Hence, local governments should be allowed to raise their own revenue and be forced to pay back any loans by raising local taxes. Borrowing should be done only for capital investments, and should require national approval. They also argue that the banking system must be kept separate from direct political intervention, but that the central government must ensure transparency and monitor the level and nature of the banking systems liabilities.

The organization of service delivery is another important factor. The question of who pays for a service is independent of how it is delivered. Each type of service has a different optimal way for funding and delivering it, so each service should be examined individually.

It is very important to establish information systems and competitive governments. One of the major arguments for decentralization is that local governments have better information about local wants and needs. Decentralization and open information allow citizens to compare the level of taxes they pay and the level of services they get with other municipalities. The central government can facilitate this by requiring uniform and complete budget and financial reports and by encouraging open budget hearings.

Not all sub-national governments are equally able to manage the challenges of decentralization. It is important that decentralization be synchronized across the fiscal, administrative, and political realms.

In the final chapter, the authors re-emphasize the need to enhance local accountability, and point out that this can be done through decentralization. Decentralization can increase community participation and transparency, allow citizens to compare government performance across municipalities, and create incentives for fiscal responsibility on the local level. Some local governments do not have the capacity to manage the greater responsibility that decentralization brings. However, the authors suggest that local governments can increase their capacity or work with private corporations and NGOs to meet the greater responsibility.