Decentralisation in Zambia

Final report

Decentralisation

in Zambia

A comparative

analysis of strategies

and barriers to

implementation

Nava Ashraf

Oriana Bandiera

Florian Blum

November 2016

When citing this paper, please

use the title and the following

reference number:

S-41306-ZMB-1

Decentralisation in Zambia:

A Comparative Analysis of Strategies and Barriers to

Implementation

Florian Blum

Oriana Bandiera and Nava Ashraf

November 2016

Contents

1. Introduction ....................................................................................................................... 1

1.1. Focus Country Summary .............................................................................................. 3

1.2. Decentralisation Reform Overview ............................................................................. 4

2. Analytical Approach .......................................................................................................... 7

3. Administrative Decentralisation ....................................................................................... 7

3.1. Risks in Devolving Policy Making Power ...................................................................... 8

3.2. Risks in Devolving HR Management ............................................................................ 9

3.3. Overcoming Limited Administrative Autonomy .......................................................... 9

3.4. Achieving Local Accountability after Administrative Decentralization ..................... 12

4. Fiscal Decentralisation ..................................................................................................... 13

4.1. Risks in Assigning Meaningful Expenditure Autonomy ............................................. 14

4.2. Risks in Local Revenue Collection .............................................................................. 14

4.3. Designing Transfer Systems ....................................................................................... 15

4.3.1. Transfer Systems in Focus Countries ................................................................. 17

4.3.2. Practical Challenges with Transfer Systems ....................................................... 17

4.4. Achieving Local Accountability after Fiscal Decentralization .................................... 18

5. Conclusion ........................................................................................................................ 19

6. Bibliography ..................................................................................................................... 20

7. Annex: Focus Country Comparison Based on the 10 Areas of the Zambian

Decentralisation Implementation Plan .................................................................................... 25

List of Tables

Table 1: Focus Country Overview ............................................................................................... 3

List of Figures

Figure 1: Ranking of Decentralization Reforms .......................................................................... 4

1. Introduction

Decentralisation reforms have been at the centre of the public policy agenda for multiple

decades. International institutions such as the World Bank embrace it as one of the major

governance reforms on their agenda. In addition, policymakers and researchers alike consider

decentralisation in China and India to be a determinant of their recent economic success

stories (Bardhan, 2002). The classical case in favour of decentralisation highlights the benefits

achieved through increased flexibility under localised decision-making while cautioning that

local decision makers might fail to internalize the externalities created through their decisions.

This notion, first formalized by Oates (1972), hence highlights that decentralisation is

beneficial when tastes are heterogeneous and there are no spillovers across jurisdictions.

In more recent theoretical work, this view has been extended to also incorporate transaction

and agency costs. Seabright (1996) argues that local governments have an informational

advantage over central governments as it is less costly for them to acquire information on

preferences and production costs. In addition, local governments might have stronger

incentives to make use of their superior information as they are more accountable to local

recipients of public services. Besley and Case (1995) suggest that increased accountability can

arise due to competition between local jurisdictions, because people can vote with their feet

and through yardstick competition.

Some caveats to this analysis exist. On the one hand, Bardhan and Mookherjee (2000) point

out that increased accountability of local decision-makers can backfire, as they might be more

prone to local elite capture. On the other hand, competition between local jurisdictions is

constrained by the presence of high moving costs and local specialisations in public good

provision.

Two additional key determinants of the success of decentralisation reforms exist. On the one

hand, as pointed out by Bardhan and Mookherjee (2000), the extent of fiscal autonomy given

to local governments determines the extent to which decentralisation reforms are prone to

elite capture. When local governments don’t only have power to determine expenditure

patterns but are also given tax collection power, the risk of elite capture is high. This is because

the ability to raise tax revenue not only allows captured decision-makers to redistribute funds

from all citizens to the capturing elites, but may also allow local elites to evade taxes such as

local property taxes. A reliance on user fees can reduce the extent of cross subsidization as

only public service recipients can be asked to contribute to the cost of the service. Yet, user

fees are often not feasible, for example when asking the recipients of the service to contribute

to its cost would defeat the purpose of the program. This could be the case for anti-poverty

programs aimed at redistributing income from rich to poor households, for example. An

alternative that also reduces the risk of elite capture is a reliance on grants from the central

government, which disconnects expenditure devolution from decentralisation in revenue

collection. While central government financing requires little administrative capacity at the

local level it also generates weak incentives for cost savings. When trying to maximize

production efficiency as part of a decentralisation reform, policymakers hence face a trade-

off between cost-minimization and the risk of elite capture.

The second key determinant of decentralisation success is the ability of local governments to

handle the assigned tasks. As has been pointed out by Bardhan (2002), disparities in technical

and administrative capacity between the central and the local governments are the main

cause of unsuccessful decentralisation reforms. The challenge with decentralisation reforms

is hence that central governments don’t know what to do and local governments don’t know

how to do it.

Given those countervailing theoretical arguments, the Zambian government has requested an

empirical investigation that evaluates the risks and opportunities associated with

decentralization. This paper reviews the literature on decentralization reforms in four Sub-

Saharan African countries (Tanzania, South Africa, Sierra Leone and Ethiopia) to understand

which factors can contribute to a successful decentralisation reform and how the design of

reforms affects their impact.

The four countries that this review focusses on were chosen because their context is

comparable to Zambia and they have experienced comprehensive decentralization reforms

which produced mixed results which maximises learning opportunities for Zambia. Today, all

countries have a de-jure system in which the administrative responsibility for health,

education and infrastructure has been devolved to local government’s authorities (LGAs). In

addition, LGAs in all focus countries have (limited) revenue raising and expenditure power.

1.1. Focus Country Summary

Table 1 provides summary information about the focus countries. In order to assess the

comparability of the focus countries to the Zambian context, it also benchmarks the

information to Zambia. The table puts a particular emphasis on the political structure of the

countries, as the literature considers this to be a major determinant of successful

decentralisation reforms.

The indicators highlight the dimensions that determine successful decentralization in which

the focus countries are comparable to Zambia. In particular, we focus on three dimensions:

Aggregate economic activity, the political system, and trade. First, in terms of aggregate

economic activity, column 2 in table 1 highlights that Zambia is comparable to Tanzania,

Ethiopia and Sierra Leone, whereas South Africa has a substantially higher GDP per capita.

South Africa also employs a higher fraction of GDP as government expenditure than the other

focus countries and Zambia. Column 3 shows that including South Africa is still important to

understand the Zambian context, as its democratic system is most comparable to Zambia. In

addition, column 4 highlights that, as opposed to Tanzania and Ethiopia, Sierra Leone, Zambia

and South Africa are all export based economies. As such, depending on the dimension of

interest, all focus countries provide relevant comparisons to Zambia.

Table 1: Focus Country Overview

Country

GDP per

capita in PPP

US$ (2015)

EIU

Democracy

Index (scale

from 1 to 10,

US: 8.11)

Dominated

by single

party?

Merchandise

trade (% of

GDP-2015)

Government

Expenditure

(% of GDP-

2015)

Tanzania

2,667

5.77

Partially

Ethiopia

1,626

3.72

Yes

Sierra Leone

1,591

4.56

South Africa

13,165

7.82

Yes

Zambia

3,853

6.39

Sources: GDP per capita and trade figures-World Bank World Development Indicators; Government expenditure figures – International

Monetary Fund; Democracy Index – Economist Intelligence Unit

1.2. Decentralisation Reform Overview

In all four countries decentralization reforms shared three characteristics. First, similar to the

Zambian decentralisation plans, reforms in the focus countries involved the devolution of

basic service delivery powers to local governments. Second, in all cases the central

government retains authority over policy and standard setting in order to assure a minimum

quality level of public service delivery. Finally, while all focus countries allocate some revenue

collection power to local governments, financing of LGA activities primarily occurs through

central government grants.

Decentralisation reforms in the focus countries had various levels of effectiveness. Figure 1

presents a proposed ranking, with darker colours indicating more successful decentralization

reforms:

Figure 1: Ranking of Decentralization Reforms

Tanzania had the least effective experience. LGAs in Tanzania were established in 1982 and

have since then sequentially received increasing administrative power. The current local

government framework has been in place since 1996 and was comprehensively reformed in

1999. As part of this reform LGAs were assigned responsibility over basic service delivery

functions such as primary health and education, agricultural extension, local water supply and

road construction. Local government staffing and planning is still primarily controlled by the

central government, which is also the main provider of LGAs’ funds. Tanzania’s

decentralization efforts are widely considered to have been ineffective at best, and

destructive at worst. As the major decentralization reforms weren’t driven by a political will

to decentralize, but can be partially considered a response to external (donor) demand, its

ability to succeed was limited from the start. The main shortcoming of the Tanzanian

Tanzania

Sierra

Leone

South

Africa

Ethiopia

decentralization efforts is that the decentralization was never fully executed and instead

created breaks in the chain of command that now significantly limit service delivery. Local

governments rely on central government funding, capacity and structures to deliver public

services, yet there are no formal mechanisms that allows central government ministries and

local government agencies to cooperate (Fermet-Quinet et. al., 2008; Venugopal and Yilmaz,

2010).

In contrast to Tanzania, Sierra Leone embarked on a self-initiated decentralisation reform

after the end of the civil war in 2002 as a newly instated Truth and Reconciliation Commission

called for more locally inclusive policies. The main motivations for the decentralization reform

were to reconfigure political institutions to reduce risk of further sources of conflict and to

enhance government legitimacy and increase political support by taking power away from

local chieftains and strengthening LGAs. As part of this reform, Sierra Leone recreated the

institutions of local councils and formalised traditional chieftains, both of which had been

abolished in 1972. The government then proceeded to devolve administrative power over

primary and mid-secondary education, primary and secondary health facilities, roads,

agriculture, water and social welfare functions to the local institutions. This highlights a key

positive aspect of decentralization: In ethnically fractionalized states, such as Sierra Leone,

decentralization can work towards overcoming ethnic tensions as it provides autonomy to

communities that avoids sources of friction.

LGAs are primarily financed through a system of inter-governmental transfers.

Decentralization has, however, been slower then expected, with especially the

decentralisation of road construction still facing significant political constraints due to an

unwillingness of the central government to limit its authority. Given those constraints, the

evaluation of Sierra Leone’s decentralisation efforts is generally mixed. In particular, existing

evidence finds no effect of decentralisation on health care coverage and education indicators.

In addition, staff attendance in health and education facilities has decreased post-

decentralisation, possibly due to a lack of capacity to monitor attendance in the LGAs

(Edwards, Yilmaz & Boex, 2015; Srivastava and Larizza, 2011).

Decentralization in South Africa was initiated as a response to challenges after the end of

Apartheid and was needed to reincorporate the former homelands (‘bantustans’) in the

country. Decentralisation was finalised as part of the post-Apartheid constitution, passed in

1996, which allocated social service provision powers to LGAs while allowing the central

government to assist through coordination and policy making. A particular characteristic of

the South African decentralisation process is that it devolved decision making power to two

different sub-national layers of government. On the one hand, provincial governments are

responsible for the administration of major social services such as education and health.

Districts councils, on the other hand, organise infrastructure related service provision, such as

water, sewerage and electricity provision. While there is some evidence that the devolution

of power in South Africa has improved the targeting of HIV treatment expenditure to areas

most affected, strong influence from the ruling African National Congress limits de-facto local

autonomy and therefore LGAs’ ability to tailor public services to local needs (Wittenberg,

2003).

Finally, the literature widely considers the Ethiopian decentralisation attempt a successful

reform (e.g. Halvorsen, Smith and Shenkut, 2005). Decentralisation in Ethiopia started after

the end of socialist rule in 1991. The process was primarily driven by the Ethiopian People’s

Revolutionary Democratic Front

which aimed to reverse previous policies of homogenisation,

in order to assuage regions and ethnic groups demanding increased control and participation

(Cohen, 1995; Turton, 2006). The reform was completed in 1995 with the passing of a new

constitution that recognized the local right to self-determination and created a federal

structure. As part of the constitutional reform, LGAs were given full responsibility over basic

service delivery functions, including primary education, agriculture, health, water, sanitation,

roads. Similar to the reforms in the other focus countries the federal government retains

authority over setting policies and standards in the devolved administrative areas. Given

positive experience with the constitutional reform, decentralisation was extended in 2002

when a subset of decision powers was devolved further from regional states to local level

districts (“Woredas”). Local service delivery is primarily funded through (ear-marked) block

grants from the central government. Evaluations of Ethiopia’s decentralisation experience

The EPRDF is the ruling party in Ethiopia that was formed out of a coalition of militias that overthrew the

military junta in 1991.

point to decentralisation having improved the performance of the public sector, both through

increased efficiency in the use of resources and enhancing the outreach of public service

delivery.

As an explanation for this success the literature primarily points towards the

existence of strong party ties between local and central government officials that facilitate the

flow of information and generate strong incentives for local officials to enhance service

delivery (e.g. Khan et. al., 2014b).

2. Analytical Approach

The structure of this literature review is based on the methodology developed in Yilmaz, Beris

and Serrano-Berthet (2008). The main premise of this framework is that successful

decentralisation can only be achieved if it sustainably implements and enforces local

discretion and local accountability. The idea behind this premise is that local governments

need the autonomy, means and incentives to respond to citizens’ demands. At the same time,

citizens require the ability and opportunity to demand accountability from their LGA

representatives. We apply this framework to the focus countries in order to identify the

characteristics of decentralisation reforms that determine their success. We perform the

analysis of the literature separately for administrative and fiscal decentralization, focussing

first on the determinants of administrative autonomy and accountability before turning to

fiscal autonomy and accountability.

3. Administrative Decentralisation

The framework developed by Yilmaz, Beris and Serrano-Berthet (2008) identifies a number of

factors that determine the extent to which LGAs have de-facto autonomy over administrative

issues. This section will focus on a subset of two factors: First, local governments need to have

the de-jure and de-factor power to design and implement local policies. Second, the central

government needs to devolve HR management functions to LGAs. The following section will

discuss challenges with administrative autonomy separately for those two areas.

For example, a $1-dollar increase in decentralised expenditure is associated with a 3.7% increase in

educational enrolment rates (Khan et. al. 2014a).

3.1. Risks in Devolving Policy Making Power

Challenges with decentralisation reforms can be traced back to shortcomings in the allocation

of autonomy to LGAs to design and implement local policies. In general, the design of

decentralization reforms in focus countries either formally or informally restricts local

government autonomy, therefore limiting LGAs’ ability to use superior information to target

public expenditure and public service provision. In Sierra Leone, for example, devolution of

policy making power is undermined by a “divide and rule” strategy employed by the central

government to keep local councils weaker than they could be. This strategy allows the central

government to control LGAs by playing off traditional authorities (“chieftaincies”) against local

authorities. The resulting vacuum of power is facilitated through the design of the

decentralisation reform, which remained vague about the formal division of power between

LGAs and chieftaincies. This ambiguity is supported through a 2010 amendment of the

decentralisation reform in which the national government established a “district officer”

position at the local level that provides a direct link between national government and

chieftains, therefore bypassing LGAs and undermining the power of local councils (Srivastava

and Larizza, 2011).

A similar situation has been documented for South Africa, where competencies are more

dispersed at the local level. In particular, responsibilities are spread over three layers of local

government: provincial governments are responsible for the administration of major social

services (especially education and health), whereas infrastructure services are administered

by district level councils. Overlapping competencies create policy uncertainty that reduces

local autonomy. In addition, district councils have passed service delivery functions on to local

councils but maintain the responsibility for their financing. This division was initially born out

of the necessity to address administrative capacity gaps at the local council level but has

persisted even after the capacity had been built and now creates unnecessarily long chain of

commands (Wittenberg, 2003).

While local administrative autonomy is limited through mostly informal channels for Sierra

Leone and South Africa, a number of authors suggest that the major risk to decentralization

in Ethiopia originates in the formal channels developed by the central government (Yilmaz and

Venugopal, 2011). In particular, strong central party influence significantly restricts local

autonomy, as local government officials are incentivized by and accountable to the (central)

party administration. The performance of local level development agents, for example, is

primarily measured through the achievement of planning targets set centrally by the party

(Dom and Mussa, 2006). Taken together, this experience therefore suggests that

understanding the motivations of central government decision makers, and making sure that

central stakeholders are committed to devolving power to the local level, is crucial.

3.2. Risks in Devolving HR Management

Similar to the devolution of power to design and implement local policies, the decentralisation

of staffing decisions is also limited in the focus countries. In Tanzania, for example, all district

level operations are overseen and managed by the District Executive Director (“DED”) who is

appointed by the central government. The DED can himself appoint the executives of lower

level LGAs, leaving no space for local decision making in appointing executives. This pattern

of central influence also exists for non-executive positions. Tanzania’s LGA staffing decisions

are made by local employment boards, which have 3 members from the central and 2

members from the local governments, therefore limiting local staffing autonomy. Similarly,

doctors, secondary school teachers, accountants, nurses and agricultural extension officers

are typically recruited by the responsible central ministry and then deployed to LGAs

(Venugopal and Yilmaz, 2010).

A similar situation exists in Ethiopia, where all local hiring and firing decisions require approval

from higher levels of government. For example, regional bureaucrats need to approve hiring

decisions for teachers and zonal as well as regional executives have the ability to overrule

woreda level hiring decisions. Similarly, the EPRDF retains a decisive role in the appointment

of high-level executives at the local levels (Yilmaz and Venugopal, 2011).

3.3. Overcoming Limited Administrative Autonomy

A possible explanation for de-facto low levels of administrative autonomy is the prevalence of

capacity constraints at the local level. Experience from the focus countries shows that

especially skills necessary for budget execution, such as planning and accounting skills, as well

as financial and HR management skills, are in short supply at the local level. For example, an

assessment of LGA performance in Tanzania shows that only 8.6% of councils have high

financial management capabilities, with most rural councils receiving the lowest possible

rating (World Bank, 2001). A more recent assessment reveals that while the situation has

improved for some LGA’s, particularly those close to urban areas and the capital city,

inadequacies of key professional staff with some of the above skills, such as accountants,

continue to persist, particularly in rural councils (Parliamentary Centre, 2011).

In addition, LGAs are constrained by insufficient management skills. For the case of South

Africa, limited staffing ability of local governments especially for elected positions led to

inadequate staff establishments (Koelble & Siddle, 2012). This did not only result in a high rate

of vacancies in local governments but also led LGAs to install non-elected bureaucrats in

leading (political) LGA position, therefore undermining local accountability (Wittenberg,

2003). Similarly, Sierra Leone has experimented with decentralized recruitment of staff but

has resumed central recruitment after it was found that the skill sets of the newly recruited

staff were far below those of (local) staff assigned by the (central) Office of the Establishment

Secretary (Kanu, 2009).

Countries have experimented with multiple ways of overcoming human capacity constraints.

First, development partners in Tanzania provided computerised systems for accounting,

planning, M&E and budget reporting as part of the decentralization reform. Donors also

supported the hiring of technical advisors who are tasked with supplementing a lack of

administrative capacity at the local government level. While there exists no evaluation for this

program, anecdotal evidence suggests that while this program was effective in providing the

necessary infrastructure for decentralization, it did not allow LGAs to satisfactorily fulfil their

duties in the long run (Frumence et. al., 2013).

A second capacity building measure regularly employed by central governments involves the

provision of training to local government staff. This is either organized through capacity

building grants allocated to LGAs or through standardized countrywide training provisions.

The evidence on such programs is generally positive. Tanzania has so far invested a total of 6

billion Tanzanian Shillings (approx. 3 million USD) in capacity building grants, mostly targeted

at improving planning and financial management skills. The literature perceives this program

to have been successful in improving the level and targeting of service delivery (REPOA, 2008).

Similarly, the Ethiopian central government maintains training programs for elected LGA

officials that focus on agricultural development, basic management, financial management,

integrated rural development and ethics and are generally reviewed positively (Beyene, 2000).

The South African government undertook a special capacity building measure to train LGA

staff on effective public service provision. As part of this program, LGAs cooperated with

private companies tasked with training LGA personnel in the area of water management. This

measure has been evaluated very positively as it generated local government capacity and

also supported emerging companies in the area of water management (Elhiraika, 2007).

Finally, all focus countries have also experimented with transferring experienced central

government staff to LGAs in order to fill capacity gaps. While this is generally viewed as an

effective measure to overcome human capital constraints, central government staff is

typically unfamiliar with local conditions. This program therefore reduces LGAs’ ability to

target its services to local demand conditions.

While capacity building attempts are generally viewed favourably in the literature, two major

challenges have arisen. First, when wages for LGA staff are fixed between locations, qualified

staff tends to sort into urban or more developed areas, leading to disadvantages for poorer

regions. It is hence key to devolve de facto autonomy over staff compensation to local

councils, and assure that financing for compensation is readily available (see also section on

financial decentralization below). Second, LGAs don’t only compete with each other but also

with the private sector, given that especially budget execution skills are also crucial for the

success of businesses. As such, attrition, especially after capacity building programs have been

completed, is a major concern (Girishankar et. al., 2006). The economic literature highlights a

number of ways that retention of government workers can be assured. First, increased

compensation and performance pay can foster retention (e.g. Deserranno, 2016; Dal Bo et. al.

2013). In addition, opportunities for career development, continuing education, , resource

availability and recognition/appreciation have been shown to reduce health worker attrition

in various developing country contexts (see the literature review by Willis-Shattuck et. al.,

2008).

3.4. Achieving Local Accountability after Administrative

Decentralization

When considering the determinants of local accountability, Yilmaz, Beris and Serrano-Berthet

(2008) separate public and social accountability. In their terminology, public accountability

refers to the accountability of leading administrators at the local level to their top

administrative officers or auditors. Social accountability, on the other hand, refers to

community-based auditing of the quality of service provision.

Focus countries’ experiences with public accountability are mixed. On the one hand, high

levels of public accountability are a possible explanation for the positive performance of South

Africa’s and Ethiopia’s decentralization reform. To overcome communication challenges, the

South African central government has established sector specific intergovernmental relations

committees of ministers and members of provincial councils. While such committees are only

consultative with limited executive powers, they can still provide incentives for service

providers to increase accountability and enhance information flows (Eaton, Kaiser and Smoke,

2010). Similarly, public accountability in Ethiopia is driven by strong incentives provided by

the centralised party structure that makes career progression of civil-servants dependent on

service delivery outcomes (World Bank, 2005; Yilmaz and Venugopal, 2011).

While accountability in Ethiopia works for service providers, it works significantly less for local

councillors tasked with overseeing executive policy implementation and service delivery. As

they are partially accountable to higher level (party) bureaucracies and partially to local

electorates, the aforementioned incentive channel is weakened, therefore undermining

public accountability (Khan et. al., 2014a). A similar picture emerges for Tanzania where all of

the executive staff is centrally appointed but monitored by local level councils. As local

councils have no instruments to discipline centrally appointed staff and there are no formal

linkages that allow the central government to effectively monitor local level executives, an

agency problem arises. This set-up therefore generates weak incentives for service providers

which undermines the effectiveness of decentralisation reforms (Venugopal and Yilmaz,

2010).

Achieving social accountability has proven to be significantly more difficult than achieving

public accountability in the focus countries as there is little evidence of local citizens

effectively sanctioning poorly performing local government officials. In Tanzania, for example,

the standard method to assure social accountability is through village assembly meetings.

While this assembly has the de-facto authority to overrule village leaders’ decisions,

Venugopal and Yilmaz (2010) report that there are no examples of such behaviour. At the

same time, village assemblies are not equipped with the authority to sanction local council

members, therefore limiting opportunities for social accountability.

A success story supporting the effectiveness of social accountability mechanisms comes from

Ethiopia, where structured feedback sessions that brought together citizens and service

providers significantly strengthened citizen participation and improved service delivery

outcomes. Similarly, grievance redress mechanisms through a designated ombudsman,

formalised citizen feedback mechanisms and joint service improvement plans led to

improvements in basic service provision (Khan et. al., 2014b). However, Ethiopia’s strong party

leadership, one of the drivers of successful public accountability, also undermines more

traditional social accountability channels. Traditionally, village level assembly (“kebele”)

meetings used to be the first point of contact for citizens to provide feedback about service

provision. In the recent past, it has however been argued that kebeles have come to be

primarily controlled by party cadres, and are hence unable to act as a platform that supports

social accountability (Pausewang et. al., 2002; Human Rights Watch, 2010).

4. Fiscal Decentralisation

The framework developed by Yilmaz, Beris and Serrano-Berthet (2008) identifies three factors

that determine the extent to which LGAs have de-facto autonomy over fiscal issues. First, LGAs

require a meaningful level of expenditure responsibility. For LGAs to be able to respond to

local needs, it is crucial that they retain discretion to make their own expenditure allocation

decisions. Second, LGAs need to have the autonomy and capacity to collect their own revenue.

LGAs should have rate-setting authority over locally assigned revenues and should be allowed

to define their own tax base. This argument is associated with the notion that local

governments are more accountable when relying on their own tax bases (Faguet, 2008).

Finally, LGAs need to receive financing assistance through transfers. This is because own-tax

revenue typically isn’t sufficient to cover LGAs expenditure requirements. To bridge this gap,

local authorities rely on transfers from the central government. Maintaining autonomy and

accountability in the design of this transfer system has proven to be a major challenge in focus

countries.

4.1. Risks in Assigning Meaningful Expenditure Autonomy

Expenditure autonomy is typically limited by central regulation. In particular, experiences

from the focus countries show that the de-facto expenditure autonomy allocated to local

governments differs substantially from the de-jure status. In Tanzania, for example, while

LGAs have de-jure autonomy over expenditure allocation, 78.5% of local government

spending is recurrent, mostly due to high staff costs, leave little space for decentralised

expenditure targeting.

Expenditure autonomy is mostly undermined through the employed financing mechanisms

and, among those, primarily through the reliance of LGAs on central government transfers.

With the aim of increasing effectiveness, such transfers are typically earmarked or are

associated with prescribed sector-specific maximum spending. In Tanzania, for example, LGAs

have the de-jure autonomy to determine the wage structure for their employees

independently. Yet, spending limits on wages, together with a transfer allocation that is

dependent on the number of LGA employees, typically prescribes a specific level set by the

central government (World Bank, 2001).

A similar picture emerges for the case of South Africa, where LGAs are responsible for

financing the salary of staff at the local level through unconditional grants received from the

central government. While this set-up allocates a high de-jure autonomy to local

governments, central regulation de-facto limits local discretion by describing salary levels for

local staff and making transfer allocation conditional on the expected wage bill, leaving little

space for discretionary budget allocation (Wittenberg, 2003).

4.2. Risks in Local Revenue Collection

The framework developed by Yilmaz, Beris and Serrano-Berthet (2008) suggests that the

ability of local level governments to finance their operations through own revenue sources is

a crucial determinant of LGA autonomy. Experience from focus countries does, however,

reveal three challenges with local revenue collection.

First, levels of locally collected revenue are generally low. For example, while Sierra Leone’s

LGAs finance 25% to 30% of their total expenditure through self-collected revenue, Tanzanian

LGAs only achieve a mere 6.9% (USAID, 2010; Searle, 2009). This is driven by the fact that in

the absence of transfers only narrow tax bases can be put under the control of local

governments as the taxation of large tax bases, such as corporate profit, generates spillovers

between LGAs that aren’t internalized under a decentralized system. In addition, devolution

of income and corporate tax bases significantly exacerbate the inequality between urban and

rural LGAs. Given those constraints, only narrow tax bases, such as property value, are

typically delegated to local governments. Those bases, however, are too narrow to finance

the recurrent expenditure of especially rural LGAs (Oates, 1972).

Second, local revenue collection can exacerbate political tension and undermine social

accountability. On the one hand, even the devolution of narrow tax bases, such as property

value, is likely to generate significant inequality in tax collection rates between urban and rural

areas and hence supports differential service provision levels in the absence of transfers. On

the other hand, local revenue collection makes LGAs prone to elite capture. In Sierra Leone,

for example, anecdotal evidence suggests that the devolution of property tax collection power

significantly increased pressure from (property owning) local elites to reduce tax collection

efforts, therefore undermining local fiscal capacity (Jibao and Prichard, 2013).

Third, fiscal capacity is not only undermined by local elites but also through the incentives

created by fiscal transfer mechanisms. In Ethiopia, for example, Woredas receive targets for

tax collection but are responsible to transfer any excess tax collected back to the regional

level. As this arrangement significantly reduces Woredas’ benefits from tax collection, it

generates limited incentives to create tax collection capacity at the local level (Adal et. al.,

2005).

4.3. Designing Transfer Systems

Establishing a fiscal transfer system to finance LGA expenditure is a central part of any

decentralisation reform. On the one hand, transfers are suitable mechanisms to redistribute

income between rich (urban) and poor (rural) areas in order to equalize levels of public service

provision (e.g. Bordignon, Manasse and Tabellini, 2001). Transfers can also be used to

internalize inter-regional spillovers. For example, LGAs might be more likely to invest in

infrastructure that also benefits other regions (for example building power plants that serve

multiple regions) if inter-regional transfers assure that they are compensated for the resulting

benefits. Finally, transfers can also act as an insurance against local economic shocks, for

example allowing LGAs to raise government expenditure in response to high local

unemployment (e.g. Lockwood, 1999).

Designing intergovernmental transfer systems is, however, significantly constrained by inter-

regional heterogeneity and information asymmetries between the central and local

government (Bordignon, Manasse and Tabellini, 2001). For example, there might be spatially

diverging preferences for different public goods (Cremer et. al., 1996). Alternatively, localized

access to technologies used for the production of public goods can lead to differences in cost

structures (Cornes and Silva, 2000). When those differences aren’t directly observable by the

central government they can lead regions with lower costs for public good production to

receive higher transfers than they require, leaving high cost regions short of funding. Similarly,

central government transfers can also weaken LGAs’ incentive structure, for example by

reducing effort exerted on local revenue collection.

When designing transfer systems, policy makers therefore need to assure that they maintain

local incentives for revenue collection and efficient public good production while also basing

transfer allocations on indicators that are proportional to preferences for public goods and

cost measures. One example used in the literature for such an indicator is the income or

property tax rate chosen by the local government: Higher tax rates are costly but imply a

higher valuation for tax revenue which is used to signal a region with higher demand for public

goods or higher costs. Regions with higher tax rates should therefore be rewarded with higher

transfers (Bordignon, Manasse and Tabellini, 2001).

In reality, best practice suggests using one of two possible options to design transfer systems.

On the one hand, the amount of transfers can be based on a “normative” measure of the costs

associated with public service delivery. Examples that have been used include local population

size and estimates of the costs associated with public good production. This option is primarily

used by developed countries where reliable accounting data is available. For example, most

OECD countries use fixed distribution formulas that reflect the average or normative cost of

the basic package of public services provided by LGAs. It is, however, unclear to what extent

such measures can be obtained in a developing country context (Bergvall et. al., 2006).

On the other hand, transfer amounts can be based on measures of tax collection effort. Recent

developing country experience suggests that tying transfer formulas to revenue generation

can serve as a motivator for subnational governments to embark on wealth enhancing

policies. The challenge is to identify measures of tax collection effort that are independent of

local conditions to not further exacerbate inequality between rural and urban areas. Potential

measures include effective tax rates and tax collection relative to an independent assessment

of the tax base size (Poeschl and Weingast, 2013).

4.3.1. Transfer Systems in Focus Countries

Transfer systems in the focus countries typically take one of two possible shapes. On the one

hand, they can be indicator-based, meaning that the size of central government transfers is

determined by LGA-specific characteristics, such as the local development indices or poverty

and literacy rates. A key challenge with this approach is that fund allocation is unrelated to

costs. Similarly, indicator-based approaches are typically biased towards relatively large and

poor LGAs. Taken together, an indicator-based approach typically leaves richer LGAs with high

costs short of financing. In addition, this approach requires regularly updated information on

development indicators collected at the local government level, which is costly in itself.

An alternative to the indicator-based approach used in the focus countries is a cost-based

approach to transfer allocation, which determines transfer size as a function of LGA

administration costs. While the cost based-approach is certain to recover LGA expenditure,

experience from Sub-Saharan African countries shows that it can lead to blowing up of staff,

or breaking down of LGA units to increase transfer allocation (Smoke, 2003 and 2008).

4.3.2. Practical Challenges with Transfer Systems

Experience from the focus countries shows that the disparity between rich (urban) and poor

(rural) LGAs is a key challenge to effective transfer systems design. In South Africa, for

example, LGAs’ share of transfer income spans from 15% in urban areas to 80% in more rural

LGAs. Urban municipalities are therefore able to achieve the same level of service delivery as

rural regions while maintaining a higher independence from the central government, which

creates significant spatial heterogeneity in local autonomy and the effectiveness of

decentralisation reforms (Koelble & Siddle, 2012).

A second challenge that has arisen in focus countries relates to the timeliness and reliability

of transfer payments. In Sierra Leone, for example, the central government demands detailed

documentation and justifications of previous expenditure before approving additional

payments. As the assessment of such reports is time consuming, this system causes

substantial uncertainty, which has shown to particularly delay infrastructure investments

(Edwards, Yilmaz and Boex, 2015).

Finally, the central government in South Africa regularly engages in large discretionary

investments in areas allocated to local governments, such as road construction. As

coordination between LGAs and the central government is insufficient, such investments

regularly conflict with LGA budgeting, therefore not only reducing local government

autonomy but also leading to inefficient resource allocation (Wittenberg, 2003).

4.4. Achieving Local Accountability after Fiscal Decentralization

Similar to administrative decentralisation, Yilmaz, Beris and Serrano-Berthet (2008) separate

public and social accountability when considering the determinants of local accountability in

the fiscal sphere. In their terminology, public accountability refers to the implementation of

effective, transparent and rule based public financial management. Social accountability, on

the other hand, refers to community-based monitoring of financial management, for example

through legislators, the civil society or the general public.

While focus countries have actively attempted to improve public accountability, those efforts

have been met with a number of challenges. On the one hand, implementing consistent

budgeting procedures and public financial management structures requires human capacity

at the local level. Experience from the focus countries shows that especially skills necessary

for budget execution, such as planning and accounting skills, as well as financial management,

are typically in short supply in local government authorities. In addition, reports on Ethiopia

and Tanzania suggest that there is a lack of capacity to monitor local expenditure. As a result

of such human capacity constraints, LGAs require assistance by experts from higher levels of

governments, therefore undermining local autonomy and reducing the effectiveness of

decentralisation reforms. On the other hand, evidence from Ethiopia suggests that elected

representatives tend to have lower levels of literacy rates compared to full time civils servants.

This significantly constraints their ability to oversee the planning, budgeting and service

delivery process and therefore reduces public accountability (Yilmaz and Venugopal, 2009).

Social accountability faces a similar challenge, as it is unclear whether legislators, the civil

society and the general public have access to the necessary information to scrutinize public

financial management. Evidence from Tanzania suggests that citizens are typically uninformed

about the decision power available to LGAs, let alone their performance (World Bank, 2010).

Similarly, recent survey evidence from Ethiopia shows that 91% of respondents are unaware

of the extent of financial autonomy allocated to their local government. A similar fraction was

unaware of how local expenditure was allocated (FTAPS, 2009). Taken together, this evidence

suggests that local social accountability is significantly constrained by information

asymmetries, hinting to the importance of LGA transparency during and after the

decentralisation process.

5. Conclusion

Decentralisation reforms are typically used as a policy instrument to improve the targeting of

public services and the accountability of public service providers, thereby enhancing the

effectiveness of public sector operations as a whole. This article has argued that there exists

substantial heterogeneity in the extent to which decentralisation reforms allocate de-facto

autonomy to local governments, driven primarily by the design of the reform process.

Evidence from South Africa, Tanzania, Ethiopia and Sierra Leone suggests that decentralisation

reforms maintain strong elements of central regulation and therefore remain incomplete.

Taken together, experience from the focus countries suggests that assuring a commitment to

delegating power at the central level and the availability of capacity to deliver services and

demand accountability at the local level is key for successful decentralisation.

6. Bibliography

Adal, Y., Wube, N., & Cherinet, H. (2005). “Study on the implementation of decentralization

policy in ethiopia.“ Addis Ababa: Oxfam.

Bardhan, P. (2002). “Decentralization of governance and development”. The journal of

economic perspectives, 16(4), 185-205.

Bergvall, D., Charbit, C., Kraan, D. J., & Merk, O. (2006). “Intergovernmental transfers and

decentralised public spending.“ OECD Journal on Budgeting, 5(4), 111.

Beyene, Asmelash. (2000). “Decentralization as a tool for Resolving the Nationality Problem:

the Ethiopian Experience”. Regional Development Dialogue 21(2): 122–147.

Bordignon, M., Manasse, P., & Tabellini, G. (2001). “Optimal regional redistribution under

asymmetric information.“ The American Economic Review, 91(3), 709-723.

Cohen, J. M. (1995). “Capacity building in the public sector: a focused framework for analysis

and action”. International Review of Administrative Sciences, 61(3), 407-422.

Cornes, R. C., & Silva, E. C. (2000). “Local public goods, risk sharing, and private information

in federal systems.“ Journal of Urban Economics, 47(1), 39-60.

Cremer, J., Estache, A. and Seabright, P. (1996). “Decentralizing public services: what can we

learn from the theory of the firm?” Revue d’Economie Politique, 106: 37–60.

Dom, Catherine, and Mohammed Mussa. (2006). “Review of implementation of the

decentralisation policy: A sample survey in six weredas of Amhara Region. “ Report for the

Embassy of Sweden and Amhara National Regional State. Oxford, UK

Edwards, Benjamin, Serdar Yilmaz, and Jamie Boex. (2015) "Decentralization as a Post‐

Conflict Strategy: Local Government Discretion and Accountability in Sierra Leone." Public

Administration and Development 35.1: 46-60.

Elhiraika, A. B. (2007). “Fiscal decentralization and public service delivery in South Africa.“

African Trade Policy Centre (58).

Faguet, J. P. (2014). “Decentralization and governance”. World Development, 53, 2-13.

Faguet, J. P. (2008). “Decentralisation's effects on public investment: evidence and policy

lessons from Bolivia and Colombia.” The Journal of Development Studies, 44(8), 1100-1121.

Fermet-Quinet, Eric, Fatoumata Samake, Yumiko Sakurai. (2008). “PVS Evaluation Report.”

OiE Report 20112008.

Frumence, G., Nyamhanga, T., Mwangu, M., & Hurtig, A. K. (2013). “Challenges to the

implementation of health sector decentralization in Tanzania: experiences from Kongwa

district council. “ Global health action, 6.

Girishankar, N., Degroot, D., & Pillay, T. V. (2006). “Measuring Intergovernmental Fiscal

Performance in South Afirca: Issues in Municipal Grant Monitoring.“ World bank.

Halvorsen Tor and Robert L. Smith. (2005). “Delivery of Health and Education Services in

Ethiopia: The Role of National Institutions.” Presentation at the 8th Annual Norwegian

Education Trust Fund Seminar. October 13th and 14th. Oslo, Norway.

Human Rights Watch, 2010. ““One Hundred Ways of Putting Pressure- Violations of Freedom

of Expression and Association in Ethiopia.” Human Rights Watch: New York.

Kanu, Alhassan. (2009). “Administrative Decentralization: Building the Non-Financial

Capacity of Local Governments” in Yongmei Zhou (ed.) Decentralization, Democracy, and

development Recent Experience from Sierra Leone. World Bank: Washington DC.

Khan, Qaiser M., et al. (2014a). “Decentralization’s effect on education, health and

agriculture outcomes, and on distribution: Evidence from Ethiopia“. In first Annual Bank

Conference on Africa, Paris.

Khan, Qaiser M., et al. (2014b). “Improving basic services for the bottom forty percent:

lessons from Ethiopia.” World Bank Publications.

Koelble, T. A., & Siddle, A. (2014). “Institutional complexity and unanticipated consequences:

the failure of decentralization in South Africa.“ Democratization, 21(6), 1117-1133.

Lockwood, B. (1999). “Inter-regional insurance.“ Journal of Public Economics, 72(1), 1-37.

Oates, Wallace. (1972). “Fiscal Federalism”. New York: Harcourt Brace.

Parliamentary Centre. (2011). “Local Government Budgeting and Implementation in

Tanzania.” Parliamentary Strengthening Program (APSP) for Budget Oversight.

Pausewang, S., Tronvoll, K., & Aalen, L. (2002). “Ethiopia since the Derg. A Decade of

Democratic Pretension and Performance.“ London/New York.

Poeschl, C., & Weingast, B. R. (2013). “The Fiscal Interest Approach: The Design of Tax and

Transfer Systems.“ Mimeo.

Raff, Horst, and Wilson, J. (1997). “Income Redistribution with Well-Informed Local

Governments.” International Tax and Public Finance 4, 407-427.

REPOA (2008). “Local Governance Finances and Service Delivery in Tanzania: A summary of

findings from six councils.” REPOA Working Paper, Dar Es Salaam.

Searle, B. (2009). “Sierra Leone: Issues in Fiscal Decentralization.“ World Bank, AFTPR,

Washington, DC.

Smoke, P. (2003). “Decentralisation in Africa: goals, dimensions, myths and challenges.“

Public administration and Development, 23(1), 7-16.

Smoke, P. (2008). “The evolution of subnational development planning under

decentralization reforms in Kenya and Uganda.“ Planning and Decentralization: Contested

Spaces for Public Action in the Global South, 89-105.

Srivastava, V., & Larizza, M. (2011). “Decentralization in postconflict Sierra Leone: the genie

is out of the bottle”. “Yes Africa Can: Success Stories from a Dynamic Continent”, 141-154.

Tiebout, Charles M. (1956). “A Pure Theory of Local Expenditures.” Journal of Political

Economy 64, 416-424.

Turton, D. (Ed.). (2006). “Ethnic federalism: The Ethiopian experience in comparative

perspective“. James Currey.

USAID. (2010). “Comparative Assessment of Decentralization in Africa: Final Report and

Summary of Findings.” Washington, DC: USAID.

Venugopal, Varsha, and Serdar Yilmaz. (2010). "Decentralization in Tanzania: An assessment

of local government discretion and accountability." Public Administration and Development

30.3: 215-231.

Wittenberg, Martin. (2003). “Decentralisation in South Africa.” Econometric Research

Southern Africa.

World Bank. (2001). “The United Republic of Tanzania: Decentralisation Sector Work.”

Report 22048. World Bank, Africa Regional Office.

World Bank. (2005). “An Independent Review of World Bank Support to Capacity Building in

Africa: The Case of Ethiopia”.

Yilmaz, Serdar, Yakup Beris, and Rodrigo Serrano-Berthet. (2008). “Local government

discretion and accountability: A diagnostic framework for local governance.” World Bank.

Social development department.

Yilmaz, Serdar and Varsha Venugopal (2011). “Obstacles to Decentralization in Ethiopia:

Political Controls versus Discretion and Accountability.”

Jibao, S. S., & Prichard, W. (2013). “Rebuilding Local Government Finance After Conflict: The

political economy of property tax reform in post-conflict Sierra Leone.“ International Centre

for Tax and Development Working Paper, 12.

7. Annex: Focus Country Comparison Based on the 10 Areas of the Zambian

Decentralisation Implementation Plan

Country/