Decentralization has been a major concern of developing countries, the international development community and researchers for two decades. The debate has centered on two sets of questions. The first examines the main driving forces and reasons for decentralization and how its overall benefits can be maximized (see, for example, Oates, 1972; Manor, 1999; Fukasaku and Mello, 1999; Dethier, 2000; Shah and Thompson, 2004). The second assesses the impact of decentralization on variables of interest such as corruption (Fisman and Gatti, 2002), government
responsiveness to local needs (Faguet, 2002), public-service delivery (Litvack and Seddon, 1999; Lieberman; 2002) and political stability (World Bank, 2000). Although decentralization has been a popular topic for a considerable time, its relationship to poverty has only recently received attention. The economic literature on poverty has ignored its potential importance in achieving poverty-reduction objectives such as the promotion of opportunities, empowerment, participation, security and rights for people who are poor and excluded at the local level (Von Braun and Grote, 2002; Romeo, 2002).
At present, the study by Von Braun and Grote (2002) seems to be the most advanced and in-depth treatment of the impact of decentralization on poverty. Based on a rigorous review of the literature and cross-country comparisons, the authors come to the conclusion that decentralization serves the poor, but only under specific conditions. The authors recommend that these conditions should be analysed within a framework that tackles political, fiscal and administrative decentralization simultaneously, while also taking into account different country specific conditions and different types of decentralization policies. Decentralization has traditionally been motivated by the following two arguments:
Decentralization can lead to an increase in efficiency. Central state authorities usually lack the “time and place knowledge” (Hayek, cited in Ostrom et al., 1993) to implement policies and programmes that reflect people’s ‘real’ needs and preferences. If properly managed, decentralization is seen as a way to improve allocative efficiency (Musgrave, 1983; Oates 1972). Decentralization can lead to improved governance. Decentralization enhances accountability and monitoring of government officials and decision makers. Unchecked authority and inadequate incentives encourage “rent-seeking behaviour” by government officials. Decentralization undermines these opportunities by creating institutional arrangements that formalise the relationship between citizens and public servants. Political decentralization, especially the election of local officials by citizens, when accompanied by a strong legal framework, can create local accountability and thereby foster officials’ legitimacy, bolstering citizen involvement and interest in politics, and deepening the democratic nature of institutions (Blair, 2000; Crook and Manor, 1998; Manor; 1999).
Both arguments are highly relevant for poverty reduction. Increased possibilities for participation, improved access to services and a more efficient way of providing public goods at the local level are major components of most anti-poverty programmes. However, the linkages might not be so straightforward. Decentralization is a multifaceted concept — its effects on poverty depend to a large extent on the form and type of decentralization in question. Regarding the latter point, it is important to differentiate between Deconcentration and devolution.
Deconcentration aims at transferring responsibilities to field and subordinate units of government, while field units basically remain under the hierarchical authority of central state authorities and have no distinct legal existence from the central state. In contrast to this, devolution refers to a transfer of competencies from the central state to distinct legal entities, e.g. area-wide regional or functional authorities, non-governmental and private organizations/private voluntary organizations. They do not belong to the central state which has no more hierarchical authority on them.
II. DECENTRALISATION AND POVERTY: A FRAMEWORK
Figure 1 highlights the basic channels of influence of decentralization on poverty. Poverty is understood in its multidimensional sense, going beyond the notion of “income poverty”. Three major dimensions of poverty, they might be influenced by decentralization policies: voicel essness, vulnerability, and limited access to social services.
In order to disentangle the various effects decentralisation might have, a distinction should be made between political and economic channels.
Political or democratic decentralisation is expected to offer citizens the possibility of increased participation in local decision-making processes, from which they have generally been excluded through lack of sufficient representation or organisation. Improved representation of formerly excluded people in local municipalities could, in turn, give the poor better access to local public services and social security schemes, reducing vulnerability and insecurity. In ethnically divided countries, decentralisation could also offer a way to share the power between local ethnic groups, thereby establishing grounds for political consensus and stability. A stabilised political system offers a foundation for the poor to build up their life and to begin investing. More generally, it can also contribute to a reduction in their vulnerability to shocks.
With respect to the economic channel, decentralisation is expected to have a strong and positive impact on poverty through increased efficiency and better targeting of services. Enhanced efficiency in service provision could directly improve poor people’s access to education, health, water, sewage and electricity, highly important poverty-related concerns. Delegating power and resources to the local level may also lead to better targeting of the poor. A more decentralised framework to identify and monitor programmes and projects could not only help to reduce costs but also to reach those most in need. In addition, it would enable greater responsiveness to local needs.
Essentially, two sets of conditions determine the impact of decentralisation on poverty: the background conditions inherited by the country and the process conditions of decentralisation.
Four variables relating to background conditions are analysed in terms of their poverty impact:
— Country setting. This includes population density, the state of infrastructure, the level of income and the level of inequalities across regions. In countries with low population density, decentralisation is likely to lead to scale-effect losses, reducing local authorities’ resources for poverty reduction and making the supply of services more expensive. In low-income countries, at least in the short run, decentralisation is likely to drain available resources and capacities for the establishment of local bodies, leaving less for poverty reduction strategies.
— The capacity of local actors and the culture of accountability and legal enforcement. In countries with low education levels, combined with a history of weak government accountability, participation of the poor is unlikely, making it difficult to initiate a pro poor decentralisation process.
— Social institutions. Inherited social institutions might contribute to or, conversely, conflict with the participation of excluded groups.
— Political power structure. The institutional framework of checks and balances in terms of the division of political power is a major factor.
With respect to the process of decentralisation, four elements appear instrumental in explaining the impact of decentralisation on poverty:
The ability and willingness to carry out reforms. This depends on factors such as political commitment at the national level, available financial resources at the local level, local human capacity and donor involvement in designing policies.
— Transparency and participation. Outcomes for the poor greatly depend on the culture of transparency and information flows.
— Elite capture and corruption. The transfer of responsibilities to the local level may lead elites to capture the decision-making process, with limited or even negative impacts on poverty. Similarly, if priorities and resources are diverted from poverty reduction policies, corruption may rise.
— Policy coherence. Decentralisation might be effective only if other policy changes are implemented simultaneously (e.g. land reform) and the process does not contradict other programmes undertaken by the country or the donor community.
Besides these two sets of conditions, the outcome of decentralisation processes depends on their overarching objectives. They can be undertaken by default or by design. The former occurs when governments are forced to decentralize in order to counter diminishing budgetary resources or to respond to other factors (e.g. ethnic diversity). Governmental ability to design the decentralisation process is limited. The policy is often imposed by donors or pursued by central government to divest itself of tasks for which it has neither sufficient resources nor power. When decentralisation is undertaken by design, governments have greater ability to shape the process. Authorities believe in the benefits of decentralisation and strongly back the process, promoting empowerment at the local level. The role of local governments shifts from the mere provision of services to promoting socio-economic development.
III. DECENTRALISATION AND POVERTY: LESSONS FROM COUNTRY EXPERIENCES
The following evaluation is based on evidence reported in the reviewed studies on how decentralisation affected poverty outcomes in particular countries (see Annex for a detailed description). Based on the results, four performance categories of decentralizations’ impact on poverty were defined: i) positive, ii) somewhat positive, iii) somewhat negative, and iv) negative.
Table 1. Country Classification by PerformanceDefinition of categories:
— Positive: successful decentralisation programmes with a significant positive impact on poverty reported.
— Somewhat positive: relatively successful programmes with some positive impact on poverty reported.
— Somewhat negative: relative failure of decentralisation programmes with very little impact on poverty reported.
— Negative: failed decentralisation programmes and no overall positive impact on poverty reported.
Source: OEDC 2009 report.
This table reveals two important conclusions with respect to the poverty impact of decentralisation: first, in more than two thirds of the cases, the impact of decentralisation was reported to be either “somewhat negative” or “negative”, indicating that one has to be very cautious in promoting decentralisation for poverty reduction. Secondly, although it appears that
the chances of pro-poor decentralisation seem to increase with the level of a country’s overall
development — all negative performers are least developed countries (LLDC’s) while most of the positive performers are middle-income countries — important exceptions, such as the Indian
state of West Bengal, remain.
On the basis of the above framework, the case studies were reviewed according to: i) the objectives of the decentralisation process; ii) information on the background of the country; and
iii) conditions related to the process of decentralisation (Annex). Comparing these conditions among the different groups of performers highlighted the following characteristics of each of the
i) Characteristics of “Positive” Performers
In the papers reviewed, decentralisation is found to have had a positive impact on poverty in the cases of Bolivia, Philippines and India (West Bengal). Without exception, these are lower middle-income countries, or less indebted low-income countries. In addition, they have a literacy rate of over 80 per cent, in sharp contrast with the bad performers, where the rate lies below 50 per cent. All these countries are qualified as free by Freedom House (FHR <= 2.5).
The process of decentralisation has generally been supported by sufficient government ability and willingness to carry out reforms, as well as by transparency, participation and policy coherence. All countries in this category adopted their decentralisation programmes by design.The authorities visibly believed in the process and had the ability to shape it. Moreover, the reforms seem to have been inspired by a desire to improve social, economic and political conditions, in the context of measures such as democratisation, improved community participation and poverty reduction. All successful countries adopted a comprehensive approach, concurrently undertaking political, fiscal and administrative decentralisation. The process went beyond deconcentration to a real delegation of power to lower tiers of government, with support from central government.
ii) Characteristics of “Somewhat Positive” Performers
China, South Africa, Mexico and Ghana are characterised by relatively successfuldecentralisation programmes, with an identifiable impact on poverty. The process fulfils only some of the criteria for an efficient, sustainable, transparent, participatory, equitable and coherent process. The official manifesto and/or implicit objective do not specifically involve poverty reduction.
In these countries, the rationale for decentralisation has been mostly economic (China, South Africa and, to a certain extent, Mexico). In some countries, central government functions have been transferred only partially in order to tackle the problem of diminishing budgetary resources. These countries all have a very high literacy rate (above 70 per cent). On the Freedom House index, their score is very good (“free”, FHR 5.0), where decentralisation has been driven mainly by the process of economic opening.
They have a higher income than the worse performers, but also higher Gini indexes, with the exception of Ghana, which qualifies as a highly indebted poor country and is considerably poorer than the others in its category. This exception reflects Ghana’s more equal income distribution when compared to worse performing countries, as well as its small surface and population size.
iii) Characteristics of “Somewhat Negative” Performers
The decentralisation programmes in Paraguay, Brazil, Nepal, Vietnam, Egypt, Sri Lanka, Ethiopia, Burkina Faso and Uganda are characterised mostly by failure in terms of pro-poor outcomes, although in some instances individual regional programmes have resulted in some poverty reduction. These countries display either low income combined with a low Gini index (e.g. Uganda, Vietnam) or higher income and a higher Gini index (e.g. Brazil and Paraguay). Thus, this category appears to group examples in which some positive elements cohabit with negative ones.
These countries are generally unstable, emerging from civil wars or ethnic conflicts or, in some cases, are still affected by political instability. The overriding objective of the decentralisation programme is political stability and the maintenance of central control through deconcentration rather than effective devolution. In many cases, decentralisation policies are aimed at preserving and re-establishing national unity. Being implemented by default, decentralisation in these countries is not designed for its benefits in terms of democratisation, greater responsiveness to local needs and community participation, the three recognized dimensions of poverty reduction. As a result of the shortcomings of the decentralisation process, the countries in this category have not pursued a comprehensive approach to decentralisation, choosing deconcentration rather than a devolution of power.
iv) Characteristics of “Negative” Performers
Guinea, Mozambique, Malawi, India (Andrah-Pradesh) and India (Madyha-Pradesh) share many characteristics with the previous category, but decentralisation has not shown any pro-poor impacts. The reform process has been flawed. All countries have pursued decentralisation reforms by default. Like countries in the previous category, they are post conflict economies and thus share similar reform objectives, but they have registered no demonstrable pro-poor outcomes from specific projects that are linked to decentralisation. The negative performers are all low income and HIPC countries. Their literacy rate is under 50 percent. None qualify as “free countries”. Their infrastructure is poor; their score on the Corruption Perception Index is rather bad (below 2.9). Their Gini index varies, meaning that no real trend can be made out.
The analysis confirms the crucial importance of the country background and the design of the process in shaping the success or failure of pro-poor decentralization. In the next section, a detailed analysis is undertaken to single out individual factors of influence within these broad categories.