Dollar for Dollar?

Accountability and Foreign Aid in Tanzania

Helen Tilley

PhD Candidate, Department of Economics,

School of Oriental and African Studies,
University of London

Paper presented at the Centre for the Study of African Economies 2011
Annual Conference: Economic Development in Africa

Oxford
20th - 22nd March 2011
Contents
1. INTRODUCTION .................................................................................................................... 3
2. DECONSTRUCTING ACCOUNTABILITY ..................................................................................... 4
Accountability as we know it: horizontal and vertical accountability ......................................................... 4
Critiquing traditional accountability.................................................................................................. 7
State and society and the elite ................................................................................................... 8
The elite ........................................................................................................................... 8
The state within society ...................................................................................................... 11
Summary: accountability deconstructed ......................................................................................... 13
3. FOREIGN AID IN TANZANIAN SOCIETY ................................................................................... 16
Aid dependency and modalities ................................................................................................... 17
Conditionalities, negotiation and aid delivery ............................................................................... 19
Foreign policy and aid ............................................................................................................... 19
Tanzanian foreign policy ..................................................................................................... 19
Donors foreign policy......................................................................................................... 22
Power relations and donor-government interaction ........................................................................ 23
Conditionality and power ........................................................................................................ 23
Fieldwork analysis and findings ................................................................................................... 26
Summary ............................................................................................................................... 30
4. CONCLUSIONS: RECONSTRUCTING ACCOUNTABILITY ............................................................ 31
Analysis and interpretation: the accountability framework ................................................................... 31
The conceptual framework applied ............................................................................................... 33
Policy recommendations ............................................................................................................ 34
Suggestions for further research .................................................................................................. 36
BIBLIOGRAPHY ...................................................................................................................... 38

2
1. INTRODUCTION
Accountability is central to both development and to the delivery of aid, and this research explores what
accountability means in the context of foreign aid. A new approach is proposed, based upon insights
from economics, sociology, anthropology and philosophy, that moves beyond the limiting assumptions
constraining the traditional understanding of accountability. It investigates how the emergence,
evolution and sustainability of institutions of domestic accountability are both influenced by, and in turn
constrain donor-government relations. In an analysis of the evolution of general budget support and
political events in Tanzania since 2000, the research presents a new framework for understanding
accountability and the donor-recipient relationship.1

The new type of accountability, relational accountability, is embedded within the personal relations
between agents, which in turn respond to social norms through their interaction with the social
structure. It takes a longer term perspective and is characterised by rules that can be informal and
highly complex, accidental and may have evolved over a period of time. It is a deeper and more durable
concept than the traditional understanding of accountability, procedural accountability. It therefore
provides the boundaries for the operation of the traditional understanding of accountability and when
the demands of donors are discordant with this framework, conflict results.

The analysis will apply a historical institutional perspective to consider the evolution of government and
agencies of the state and to understand how these political and governmental institutions interact with
society. Aid is placed in a broader perspective where it is part of the pattern of relations shaped by the
social context and historical episodes. An analysis of state and society is undertaken to present a
picture of the changing pattern of aid dependency in Tanzania, conditionality and negotiation and the
power relations between donors and Government.

At times it appears as though there are two governments in operation: the government with its external
face for donors and government with its internal face and concerns where the party dominates. It is by
looking at this internal face that the importance of distribution can be seen, whereby the government
spreads out resources to maintain the support of its clients. This is an aspect of accountability which
represents accountability to powerful actors, necessary for the survival of the state. It remains the case
that the private extended family is a primary circle of concern to which payoffs are made, however the
same concept can be applied to the maintenance of political and business alliances which can be
considered to be the public extended family.

Drawing upon fieldwork findings from Tanzania it will be argued that the patrimonial politics of the
domestic sphere cohabit alongside the increasingly formal international arena which uses procedural
accountability as its dominant framework. The neoliberal agenda of donors has been based on a limited
will to understand the complexities of country politics and therefore excludes the complex social context
of the domestic sphere. The focus upon the formal, the contractual, the measurable and the rational is
frequently incompatible with the wider socio-political and economic context.

Section 2 analyses and deconstructs the traditional understanding of accountability, using the
horizontal and vertical framework. The basis for the new approach to accountability, relational
accountability, is developed in a discussion of the relationship between the state and society and the
pivotal role played by the elite. Section 3 considers foreign aid and its associated relations. The

1 This paper is a synopsis of a PhD thesis, submitted in February 2011.

3
patterns of aid inflows are presented and the conditionalities that accompany aid outlined. Foreign
policy from the perspective of donors and the recipient government is discussed, leading to an analysis
of power relations and the associated interactions and conflicts. Section 4 reconstructs accountability,
arguing for the importance of considering the existence and interaction between the procedural and
relational types of accountability. The paper concludes with a call for aid delivery to be structured
around a new understanding of accountability.

2. DECONSTRUCTING ACCOUNTABILITY
This section analyses and deconstructs the traditional understanding of accountability, using the
horizontal and vertical framework. The basis for the new approach to accountability, relational
accountability, is developed in a discussion of the relationship between the state and society and the
pivotal role played by the elite. Firstly traditional, or procedural, accountability will be analysed.

Accountability as we know it: horizontal and vertical accountability

Accountability has been categorised in many ways in the governance, political economy and economics
literature, with its usage evolving over time and becoming popularised in the late 1990s in development
discourse through the good governance agenda (Cornwall 2007: 479). What is notable is the
consistency of the underlying assumptions which have remained fixed upon a notion of what motivates
the principals and agents and in turn what the solutions to these problems are. The traditional
application and use of the term accountability in public management, and applied in international
development, is based on a principal-agent model that focuses on the contractual mechanisms for
regulating behaviour between autonomous parties: henceforth called procedural accountability.

The traditional understanding of accountability, here called procedural accountability, is a concept that
emerges directly from Western culture and is implicitly based on an epistemology whereby actors,
either principals or agents, are rational utility maximisers, transactions are impersonal, rules are formal
and contractual in nature and the pursuit of self interest is assumed to lead to moral hazard in the
absence of full information. The accountability contract arises when a principal exercises authority over
an agents actions, the authority being embodied in a judgement of whether responsibilities have been
met, assessed against objective and externally specified standards (Grant and Keohane 2005). Thus
accountability, in the hard form fills the role of enforcing appropriate behaviour, with the threat of
sanction being necessary to enforce the appropriate behaviour (Schedler 1999; Fox 2007), stemming
from the assumption that in the absence of a threat of sanction the pursuit of self interest would lead to
the misappropriation of resources.

The term accountability and the concept of procedural accountability evokes a certain reassurance as it
implies measurability, for example in terms of financial and governance indicators or national audits;
and by extension that the problem is identifiable and understood. Perhaps even more comforting is the
set of predictable solutions that should be applied: tougher regulation that results in better oversight
and the application of penalties (Weisband and Ebrahim 2007: 1). In turn the drive towards global
accountability involves the placing of a negative value judgement upon practices that do not conform to
such standards. This section will consider the dominance of procedural accountability through an
examination of agencies of restraint and parliamentary democracy (Collier 1999; O'Donnell 1999) or the
horizontal and vertical accountability (H-V accountability) framework, Figure 1.

4
 Figure 1. Horizontal and vertical accountability

Source: Rakner (2005)

The H-V accountability model has been chosen to illustrate this discussion of procedural accountability
as it is the framework within which most of the traditional notions of accountability can be found
ranging from democracy linking state and society to the role of state agencies in accountability. There
have been a number of different types of accountability that have been defined: a comprehensive
categorisation includes administrative, political, managerial, market, legal/judicial, constituency and
professional (Dwivedi and Jabbra 1988). The H-V framework captures these different types within its
structure, as will be elucidated below, however it will be critiqued as it omits other aspects of
accountability, such as the moral category which Dwivedi and Jabbra also noted.

The H-V formulation of accountability involves a power relationship as one group of actors demand
justification for the behaviour of another group and it is often the less powerful (principal) who demands
accountability from the more powerful (agent). In the context of moral hazard, the accountability
relationship aims to support the interests of the less powerful.

Breaking the framework down into its two components, the horizontal and the vertical: horizontal
accountability is the capacity of agencies of restraint (such as law courts, Ombudsmen, audit offices,
electoral tribunals, central banks, media, trade unions, lawyers guilds) to ensure that other branches of
government are answerable for their actions, i.e. restraints imposed by institutions of equal status upon
each other. Vertical accountability refers to the state-citizen relationship, or the relationship between
unequals, and the holding to account of public officials through actions of democracy such as elections,
civil society and the media or making elected officials answerable to the ballot box (ODonnell, 1994:
61).

Whilst H-V accountability is a useful model for understanding the traditional (procedural) approach
towards accountability, it only allows a partial understanding as it excludes the informal, the personal
and the unspecified and unwritten aspects of behaviour and the functioning of power. These other
elements will be captured in the concept of relational accountability, to be arrived at through a brief
deconstruction of the assumptions underlying procedural accountability.

Collusion blurs the lines between who is the principal and who is the agent, presenting enforcement
difficulties (Schmitter 1999; Castellani and Debrun 2005: 19). It was argued that the framework is too
simplistic. Whilst it can be extended to recursive accountability, where A is accountable to B who is
5
accountable to C, who is accountable to A (Schedler 1999: 26), it struggles with the possibility of
dynamic recursive accountability and overlapping accountabilities. Not only may state agencies collude
with each other, but alliances may be dynamic and overlapping, rather than simply static.

Vertical accountability, the relationship between unequals, includes democratic accountability, taxation
and international accountability; issues of power and information asymmetries, and the realities of
social and political structures. The concept of democratic accountability and, therefore by extension,
vertical accountability has been widely criticised in the literature for being over simplified and for not
capturing the complexities of reality. As for horizontal accountability, elements such as time
inconsistency, information asymmetry and power relations can be developed as nuances of vertical
accountability to allow a fuller model of traditional accountability to be presented. Information
asymmetry highlights the case where political parties have incentives to limit the range of policy choices
presented to voters; and the importance of inter-electoral political activities stress the limitations of only
considering voting behaviour. Despite these extensions, the political reality in developing countries, of
clientelism and patrimonial politics, remains unaddressed. The nature of such politics is that ties of
solidarity result in leaders never being separated from their supporters and actions being based on a
continual calculus of patrimonial reciprocity (Chabal and Daloz 1999: 39).

In developing Europe, notable periods of significant social change were often the catalyst for the
emergence of a social contract: resulting in negotiation in Britain and resistance in France. The social
contract requires that the taxation system adequately represents the values of a proportion of the
population, the size of which is determined by the ranking of taxation within the basket of issues voted
for, therefore depending upon both theories of taxation and democracy.

Many of todays developing countries have legislatures which appear to be representative, however
underneath this external face of representation, there are powerful interest groups operating serving
well established political settlements and clientelistic interests. Therefore, moving beyond the scope of
procedural accountability, pertinent issues include the balance between powerful groups with
allegiances to the state; the extent of bias and capture involved and the role that such issues play in the
accountability context in todays developing countries.

Collective action rests upon the assumption that there are collective identities that emerge from the
bottom and that there is a desire for collectivity. However studies found that collective identities in
modern Sub-Saharan Africa are not likely to be used, as exit dominates voice (Hyden 1980; Isaacman
1996). Collective action, is not simply the rulers versus the ruled. Rather, one person can move from
dominating in one situation to being dominated in another. In addition short term interests may differ
from long term interests, introducing time inconsistency. A circle of trust, or a large and dynamic
support web is the most likely result, reflecting a plurality of different identities (Chabal and Daloz 1999;
Kelsall 2004). Rather than the homogeneous economic actor with a pre determined utility function, that
is implied by procedural accountability, the reality in Tanzania and many other developing countries is
that of a fragmentation of identities where each individual has a variety of different identities that include
the religious, ethnic, geographical, economic, occupational and aspirational (Kelsall 2004). These are
complexities that require examination.

If theories of state creation are followed, the predominantly indirect nature of taxation in Sub Saharan
Africa weakens the vertical accountability relationship, as defined by the H-V accountability framework,
as the tax burden is not felt by citizens. The tax structure in Sub Saharan Africa is structurally different
from that in Europe and it not only matters what the tax levels are, but also how the tax level has been
chosen, how it is imposed (factors related to the strength of the state) and how the funds are used (Bird
et al. 2004). Taxation will not be pursued in this paper as it was not found to have played a key role in
accountability relationship as in Tanzania the tax base is insufficiently broad and there is a large
6
informal economy, such that the majority of people are not taxpayers.2 Furthermore, the electoral
mechanism of societal accountability does not fully function as there is a lack of real policy choices for
the electorate. The boundaries between the public and the private are merged such that there is no
clear distinction between the elected and civil society, supporting Gramscis (1971: 253) notion of
political society, where the state and society are intertwined and both rule and govern. Accountability
in the sense of the pure state creation model is not applicable to Tanzania due to the pervasive
influences of a complex web of relational accountability relationships and a wider circle of trust which is
maintained.

The underlying assumptions of public management permeate the procedural accountability model and
include the assumption of rational principals and agents who are maximising their own individual
interests, which linearly aggregate to an outcome for society as a whole. In response to this, contractual
mechanisms for regulating behaviour between autonomous parties are employed, leading to the
assumption that accountability is measurable, identifiable, and it allows the adoption of predictable
solutions. Behaviour and results that fall outside this understanding have a negative value judgement
upon them and are considered to be detrimental. Issues that are missing from the H-V accountability
model, and from procedural accountability, include the complexity of the network of accountability and
the personal nature of the accountability present within patronage relationships. These weaknesses
point to a limited explanatory power resulting from the restrictive assumptions of the procedural notion
of accountability, highlighting the need to reconsider how accountability functions.

Turning to pursue these criticisms through an analysis of the concepts implicit in the traditional notion of
accountability. The assumptions underlying the procedural understanding of accountability will be
deconstructed and a more complex picture presented that extends the meaning of accountability
beyond that implied through the limited notion of procedural accountability. It picks up the issue of
morals, noted as an omission of the H-V framework by Dwivedi and Jabbra (1988). It also examines the
implicit role of power in the H-V model and the role of the elite in society.

The section below will develop the notion of relational accountability, a complementary concept that
when considered alongside procedural accountability allows a full understanding of how accountability
operates in practice. It will discuss the limitations of the concept of rationality, and considering the
emergence of norms of trust and the psychological basis to decision making. This allows an
understanding of accountability to emerge that is less procedural in nature and more complex, with
overlapping and potentially contradictory responsibilities and allegiances.

Critiquing traditional accountability

The assumptions and the theoretical roots of accountability can be located at different levels of
analysis: ranging from the individual, to the institutional, and in turn to the structural. At the level of the
individual it must be asked: are individuals motivated by rational optimisations or do altruism and
morality influence behaviour? If there are influences upon behaviour beyond maximisation, where do
such influences come from? Furthermore, who are the individuals or agents that are motivated and
what is their influence upon units of economic organisation? What is the interaction between these

2
Tanzania has seen a shift towards income tax since the mid 1990s, with corporate income tax reducing from 20 percent of
total domestic revenue to 10 percent by the mid 2000s and personal income tax increasing from 8 percent to 16 percent.
However the dominance of taxation revenue amongst large taxpayers persists: accounting for 67 percent of total revenue
(World Bank 2007: 30).

7
different factors: between the individual agent and between the economic unit, organisation or
institution; and temporally, between the behavioural outcome and the influence upon future behaviour?

Due to their simplification, the assumptions that underpin the procedural notion of accountability have
only limited application in reality. Each asumption will be briefly summarised in turn. For rationality what
matters is not whether the behaviour of individuals or institutions is rational or not, but what is meant by
rationality, in other words what is included in the preference function and the extent to which bounded
rationality captures the necessarily elements. If concepts drawn from sociology and cultural
anthropology such as culture, social structure and ideological power are included then a broader
perspective of accountability can be formed to include elements that are informal and appear to be
inconsistent and irrational if restricted to the narrower preference function. The broadening of the
preference function beyond rationality removes its usefulness for the traditional rational choice theories
and it is no longer sufficiently simple to allow its use as a testable concept. Rather it becomes complex,
iterative, recursive and dynamic characteristics that are found in reality.

Agency and structure are seen as closely interacting. The recognition of a greater role for action,
agency and motivation permits a greater range of responses to stimuli that demands an alternative
model of accountability. Moreover, the interaction between norms and action removes the simplicity of
the rationality of actors embodied within procedural accountability. Agency is therefore a relative
concept, determined by the capacity of actors to act independently of constraints that may exist in the
form of social norms (Gulrajani 2010: 10). This implies a reflexive component of action whereby an
agent monitors their action and provides reasons for such action, or rationalises their action; and this
reflexivity occurs within the social setting. These elements will now be explored in a discussion of what
is meant by the state, society and the elite.

State and society and the elite

The discussion of the concepts of the state and society highlight the risk of imposing a dichotomy
between the two which is a false interpretation of a more fluid and complex reality. It does however
illustrate the tensions that may exist as the state operates within society, and from this the functions
and limitations of the state can be identified (Migdal 1988). The elite will now be discussed, followed by
a consideration of how the state and society are interrelated.

The elite

The elite evokes the personal through the notion of culture, influence and co-optation. It presents a
dynamic interface between agency and structure, through which the personal interacts with institutions.
Different agents and different groups at different times exert different influences, and therefore the elite
is fluid, mirroring the dynamism of social relations.

Lachman defines elites by their control over organisational apparatuses for appropriating resources
from non-elitesconstrained primarily by coexisting elites.. (1990: 398). The elite is central to this
analysis as conceptually it builds in conflict between competing groups. Methodologically, it brings
together the structural (macro), the institutionalist (meso), and the agent (micro).3 This is achieved
through firstly, providing an analytical interface between society and the state. Secondly, it consolidates
the role of agency firmly within culture, thus allowing both the existence of and also explicitly

3 The focus here is upon the national, to inform the national level case study of Tanzania, however in consideration of the
international aid context some extensions beyond the national will be made.

8
acknowledging an interface between the actor, the role of norms and the place of these within the
social structure. In turn, dependency
dependency and power relations can be considered. The concept of the elite
allows the existence of multiple elites that are in continual conflict with each other, moving away from a
unified elite to an emphasis upon factions within the elite (Lachmann 1990: 401). 401) Traditional Marxist
analysis is limited by the centrality of a unified ruling class, placed in opposition to the rest of society.4

The elite as a tool in historical analysis was first examined by Mosca and Pareto who considered elites
to exist only temporarily and to be challenged or decay in the longer run (1896, 1901; cited in
Lachmann 1990: 398, 400).. C. Wright Mills allowed for the permanence of of elites through combining
structural and institutional factors with the social and the personal, extending the earlier analyses and
drawing upon Webers bureaucratic authority. Mills also observed the fluidity of the elite, and that the
nature of the powerer elite is that within it there is a good deal of shifting about. This resulted from the
combination of personal friendships with active co-optation,
co optation, invoking relational characteristics (1956:
287). Recent work defines the elite as the powerful decision makers; the small number of people
controlling the key material, symbolic and political resources,
resources occupying commanding
commanding positions
within key institutions (Reis and Moore 2005: 2) or those with the power and influence to determine
which policies are implemented, when and how as well as the rules of the game game (Kalebe-Nyamongo
2009: 2).

Figure 2. The structure of the elite: an adaptation of Mills 1956

Source: Author

Mills identified three tiers of the elite: i) the inner core; ii) the outermost fringes; and iii) the tier below the
fringes, or the middle level of power. The inner core comprises the decision makers who rotate between
commanding roles at the top of organisations and therefore transcends transcends the particularity of interest of
any one of the milieux (1956: 288). The outermost fringes comprise those
he institutional milieux count and
those who count
who are taken into decisive account when decisions decisions are being made, even if they are not personally
present. This tier is more fluid than the core. Finally the middle level of power consists of pressure
groups who are not vested in the elite, regional, state and local interests and the rank and file who must m
sometimes be taken into account, handled, cajoled and raised into the higher circles if support is
4 Applying a theoretical classification, this is an adaptation of the analytical Marxist position as whilst incorporating both
structural Marxism and political conflict theory it builds in a theory of elite
el confrontation.

9
needed to achieve a specific end (1956: 288-293). These three tiers can be simplified to two levels: A)
the interior - the inner core and the fringes, contesting power at the top levels of society and where the
state class is located; and B) the exterior - the middle level of power providing the interrelationship
between the elite and the rest of society, Figure 2.

Moving beyond Mills, the analysis of the contestation between elites can be aided by invoking the
notion of factions, supported by an informal network. Indeed, the elite is not considered to be
homogeneous and consensual, but rather a heterogeneous network built around conflict and
contestation where different interests come together and compete with one another. This reflects the
dynamism within society that is a continual test of the legitimisation of the ruling elite and captures the
different interests described by Reis and Moore (2005) and Dogan (2003) such as business, politics,
media and landowners. Elite conflict is focused around the competition for revenue extraction from non-
elites where one elite challenges another through attempts to appropriate the revenue sources of the
competitor elite faction. In resisting such challenges to their autonomy, an elite faction may prevent
changes in the relations of production, and through this can have a direct impact upon the rest of
society (Lachmann 1990: 403, 412).

The dynamic nature of this competition is emphasised through the elite network: the structure within
which actors move between different elite interests, interconnecting the different factions. The network
therefore cuts across formal or informal interests or organisations ranging from state institutions such
as the civil service, political parties and parliament; to NGOs and CSOs; to the private sector. This
concept is emphasised as it is likely that individual actors are influential within more than one of these
categories. Where the network is particularly strong a coalition results, bringing different factions
together. This can be seen in Tanzania where the bringing together of the political and business elites
increases the importance of the private sector in the patronage relationship (Kelsall 2002: 610). Such
coalitions may be formal or informal, and also temporary or permanent (Leftwich 2010), however their
strength in relation to the strength of the leadership determines the degree of centralisation of power
and rent seeking. Chabal and Daloz (1999: 27) consider the connections between high and low politics:
between political elites and their clientelistic networks. They see these relationships as central in
determining the mechanisms of power and social control. Developing a circle of trust through a large
support network is particularly important in the absence of an autonomous and impartial state (Chabal
and Daloz 1999: 27), and has implications for the centralisation of power. The effectiveness of
centralised power depends on the balance between the strength of the competing factions in relation to
the leaderships ability to directly centralise power.

What role do elite confrontations play in social change or stability? The specific nature of the elite is of
course a direct function of the political economy of each country, however a common thread in the
literature is that the support of the elite is necessary for the maintenance of political power (North et al.
2007; Lindemann and Putzel 2008). However the elite can be seen as the manifestation of political
power and when the latent concept of power is applied, ideology is central to the continued
maintenance of that power. The elite dominate through groups that serve well-established interests and
the elite bargain is the outcome that provides the winning coalition with incentives to be peaceful by
limiting access to rents to those within the coalition (North et al. 2007). Lachman concluded that in the
countries he studied, elite conflict rarely achieved the intended results, however [it] is the bright thread
of agency that propelled structural changes in all situations (2000: 231).5

5Lachmans case study countries were England, France, Spain and the Netherlands. Whilst differing in social context from
Sub-Saharan Africa, he provides applicable theoretical extensions.

10
The ruling elite maintains support through patronage payments, underwritten by the latent operation of
power such that this system of support becomes institutionalised and accepted: the main interface
between the two levels of the elite (A and B, Figure 7). These payments to obtain the support of the rest
of society require a coalition between the political and the business elites as politics and business
linkages are used to identify the opportunities and siphon off the rents to allow redistribution to a wider
group. This requires off-budget redistribution through patron-client networks to achieve political
stabilization. The corrupt exchange here involves politicianstransferring resources to powerful clients
and receiving, in exchange, their political support. (Khan 2004: 14).

Acemoglu, Ticchi and Vindigni (2007) develop a model that considers why many democracies pursue
pro-elite policies and why inefficient states can persist. Consideration of this is relevant for the
discussion of how aid impacts upon accountability as it helps to address the issue of accountability and
the form it can take in a democracy, through a procedural accountability lens. The premise of the model
is that an inefficient state structure creates more rents for bureaucrats and therefore elites can use
patronage and capture democratic politics. Such rents may support the party who will maintain the
inefficient structure: by extension an inefficient state is more likely to arise when there is greater income
and wealth inequality.

In summary, drawing from these writings, the working definition of the elite that will be applied here
follows the visual representation in Figure 2. It is based on Mills interplay of the institutional and
structural with the personal and social and therefore groups, or factions, within the elite are
characterised by consensus and cohesion through cooptation and common values (1956). However,
this consensus and cohesion exists only within groups as there is continual contestation between the
factions within the inner core of the elite, resulting in fluidity as actors move between factions and
between positions. These factions function within a network where coalitions may result to serve
particular interests. Power is important to understand the elite, and elites can be considered to be
defined in terms of power (Lachmann 1990: 403), drawing upon Lukes (2005) latent power for its
reproduction. This has been linked to inequality and lower levels of economic competitiveness. An
inefficient state structure creates rents allowing elites to use patronage to maintain their support (Khan
2004; Acemoglu et al. 2007).

What does this conceptualisation of the elite imply for accountability? This notion of the elite and their
role in society does not fit well with the traditional procedural notion of accountability which provides no
space for differential power of both the tiers within the elite and the factions within the inner core. In
contrast, procedural and traditional accountability is based on the implicit assumption of equality across
actors and society is simply an aggregation of individual preferences, and therefore highly impersonal
in nature. In contrast to this, the elite invokes the personal through the notion of culture, influence and
cooptation, and an interface between agency and structure. It also provides a comprehension of how a
shared awareness can facilitate a common understanding beyond the national level, within the
international development circle, operating through the informal elite network.

The state within society

The concept of the elite is central in an analysis of the relationship between the state and society as it
permeates both the state and society, facilitating consideration of the interaction between the two, and
in turn the mechanisms through which power is contested. This emphasis on the interrelationship
between the two is opposed to the distinction between the state and society that has been retained in
some of the literature since the work of Hobbes and Locke (Migdal 1988:17).

11
The model of the state that will be applied here is one in which the state comprises of numerous
agencies whose control and reach is a function of the balance of power within a society, operating
through the elite. There is no strong line of divide between state and society as there are allegiances
and interests that refute any absolute distinction between the two. The party and the political leadership
are seamlessly merged with the structures of power and legitimacy in society that are dominated by the
elite. Bringing in the elite allows the interplay of the institutional and the structural with the personal and
social, whose elite factions compete and in doing so co-opt actors in society. Ideological power
achieves consensus and cohesion through co-optation and patronage. The continual contestation
amongst the factions within the inner core of the elite functions through a dynamic network as actors
move between factions and between positions. The shadow state stresses the role of patronage,
informal connections, networks and the elite and stresses the importance of the personal for the
effectiveness of the state (Reno 2000). Corruption is necessary for predation to operate efficiently and it
is therefore normalised.

This view of the state and society is influenced by Marxists who perceive the state and society to be
integrated and furthermore perceive societal institutions to be ideological apparatus of the state from
which it cannot be separated (Althusser and Gramsci). This was stressed by Gramsci who noted that in
certain states the head of state is the same as the head of the political party and maintains the balance
of power within civil society through its hegemonic function and influence across different groups. This
presents a view of the state, or political society as intertwined with civil society and thus both ruling
and governing (Gramsci 1971: 253). Gramsci drew upon Hegels conception of the parties and
associations of society as the private woof of the state such that the state governed with consent
through hegemony, protected by the armour of coercion, and stressed that the state reaches beyond
government to the private apparatus ofcivil society (Gramsci 1971: 259, 261, 263).6 This consent is
related to the discussion above of the coalition of elites as the state manufactures such consent by
actively encouraging and co-opting private elites.

This juxtaposition of the state versus society has been critiqued by Migdal (2001), and other recent
authors uphold the integration of the two (Gupta 1995; Chabal and Daloz 1999). Migdal considered the
state to be a public construct which is a function of the position of an individual; and considering that
the fallacy of such a construct is highlighted once the non-unitary characteristics of the state are taken
into account. Migdals recent work on the state (2001) places it firmly within society and highlights the
paradox that is implicit within his definition of the state: it is a contradictory agency that acts against
itself. The paradox concerns the corporate image of a whole and unified state; in contrast to the
reinforcing and contradictory practices and alliances of its disparate parts. (Migdal 2001: 22).

Applying the state capabilities of Migdal, the engagement or disengagement limits the extent to which
the state can both penetrate society and therefore can regulate social relations. This may be due to
resistance, mutual disengagement, some aspects of society being beyond the influence of the state, or
a combination of these reasons. Migdal presents the state as a mlange whereby heterogeneous and
autonomous groups have control, in contrast to other theories where the state is seen as dichotomous,
such as in models of the centre-periphery form. This perspective therefore limits the extent of control of

6 In the Prison Notebooks, Gramsci drew upon but also critiqued the work of both Hegel and Marx. It should be noted that
not all Marxist writers concur with the integration of state and society, for instance Bayart (1986: 112) considers the
relationship between the state and society in terms of civil society, distinguishing between society evolving in conflict with
the state and the state being set up against society, argued to be the case in post-colonial Africa. Here the state, along with
the dominant social groups, aims to control society, and it is thus dominated by a well networked elite. This research takes a
more nuanced view arguing that the state comprises elite factions whose reach extends into society through networks
characterised by patronage relations.

12
the state as here control can be distributed amongst a number of groups which may themselves be in
conflict with one another: for example the church, ethnic groups, villages, languages, etc (Migdal 1988:
28). Indeed, such a web-like model of society is closer to that observed in Africa as identified by Iliffe
(1971) and Kelsall (2004). This shares some similarities with the form used by Chandhoke (2003) who
describes the state being represented by many agencies and organisations and sharing power with
sub-national and partnership organisations including international agencies.7

Social control is based on compliance, participation and legitimacy (Migdal 1988: 32) and the
relationship between state and society can be viewed as being based on these three functions. The
three functions can be considered as a tension between state and society, with the state placed within
society (Migdal 1988; Migdal 2001). Compliance is achieved by the perpetuation of the dominant
ideology of the elite, and where this fails agencies such as the police are used to obtain conformance to
the rules of the state. Participation here is not of the type typically denoted in PRSPs, which refers to
citizen voice, but rather refers to voluntary use of state institutions, which increases the strength of the
state as actors are organised in their participation. Such organisation can range from the need to obtain
licences to the cooperative structures for selling their agricultural produce. Legitimation can be seen as
the successful outcome of bargaining for power as compliance and participation have been achieved
and the role and authority of the state has been accepted and has the approval of the people. In short,
the greater the extent of compliance, participation and legitimacy, the greater the control of the state.

This theory cannot fully explain the mechanism through which legitimacy is created and sustained, nor
does it give sufficient emphasis to the role of the elite and conflict. Therefore it is supplemented by the
shadow state and the elite. The individual and personal rule of the shadow state relies on patronage,
informal connections, networks and the elite. In the shadow state corruption is necessary for predation
to operate efficiently and it is therefore normalised, although not necessarily explicit. Patronage is a key
mechanism through which power is exercised and insecurity is created to ensure a reliance upon the
distribution of rents, which in turn are used to restore an element of stability such that the incumbent
retains power (Reno 2000). Bringing in the elite allows the interplay of the institutional and structural
with the personal and social and consensus and cohesion are induced through co-optation. Continual
contestation amongst the factions within the inner core of the elite functions through a network of
connections as actors move between factions and between positions.

Summary: accountability deconstructed

It is by going beyond the methodological individualism and rationality of economics and through
applying the work of sociologists, anthropologists and philosophers that the limitations of the procedural
understanding and application of accountability are drawn out.

This discussion has pointed towards a more complex underpinning to what accountability may mean
and the limitations of restricting understanding to the procedural form of accountability, pointed to by a
number of writers (Bovens 1998; Eyben 2007; Ebrahim 2009; Gulrajani 2010). What emerges from this
analysis is the need for an additional form of accountability that considers the complex, highly
personalised, and often contradictory political and social relations, that are based on implicit, informal
and accidental rules. This broader and more culturally relevant understanding of accountability is
termed relational accountability.

7 Chandhokes (2003) concept does however differ in that it focuses on the increasing decentralisation and pluralisation of
the state and increasing network governance.

13
Relational accountability has been developed from a multidisciplinary understanding of: culture; the
interface between agency and structure; the critique of rationality; and the positioning of the state in
society and the role of the elite. It is defined as the accountability that is embodied within the personal
relations between agents, those which in turn respond to social norms through their interaction with the
social structure. The elite is at the core of the analysis due to the pivotal role that the group plays in
influencing policies and decisions, linking the state and society. It is fluid and the reach of accountability
from this core stretches out in a web-like structure capturing actors as they function within a multitude of
different social roles within the elite network. Relational accountability depends on the process of power
that is at play in all relations and is subtly demonstrated through language. This type of accountability
allows the role of obligations and patronage to be analysed in a reciprocal relationship that is influenced
by social norms, culture and power.

Williamson presented a tiered framework for social analysis distinguishing between the micro theories
(the neoclassical methodological individualist approach at the bottom level) and the macro theories
(institutional and structural approaches at the top level), Figure 3. The top level (Level 1) is the focus of
the work of economic historians and sociologists and is the location of norms, traditional and embedded
elements such as cognitive, cultural, structural and political factors. Much social analysis takes these
factors as exogenous as they can remain constant, changing across centuries or millennia. This
captures the relational accountability characteristics of longer term change, the importance of the wider
social system and its nuances, and the informal role of customs. The levels below are where procedural
accountability can be found as rules become formalised within a contractual basis. The second and
third levels are the focus of new institutional economics, with Level 2 concerning the institutional
environment where formal rules and regulations are located, such as property rights, legislation and
constitutions. The legislative structure of Level 2 determines the structure and operation of the state
agencies including the executive, the parliament and the legislature. Level 3 is the location of the
operational rules of governance structures and the location of governance as an effort to install order to
minimise conflict and to result in mutual gains. This is grounded in transaction costs, with a focus on
contracts and the transactions that subsequently occur. This focuses upon the impact of governance
structures on incentives and, due to the emphasis on the contract, changes take place alongside
contract renewals at a frequency of between a year and a decade. Level 4 is the focus of neoclassical
economics, underpinned by optimisation and the maximisation of surplus based on calculations of
marginal cost and benefit. The firm, aggregated into the production function, and the consumer are the
units of analysis within which continual adjustments to price and quantity take place. This micro level of
analysis also is where agency theory is located, involving ex ante incentive alignment and efficient risk
bearing, rather than ex post governance. (Williamson 2000: 600).

The higher levels of the analysis impose restrictions upon the lower levels such that changes in Levels
3 and 4, can only occur within the boundaries dictated by Levels 1 and 2. In the short term change is
restricted to those micro elements such as price and quantity; in the medium term change at the level of
governance institutions within Level 3 can occur, for instance as contracts are renewed; however only
the longer term permits institutional adjustments to state institutions. In turn all of these changes are
framed within the cognitive, cultural and normative rules of society captured within Level 1. The
restrictions that the higher levels impose upon the lower levels are such that the particular cultural
setting restricts the range of governance institutions that can be accepted within a particular society and
in turn dictates those that may successfully function. Where a change is imposed that does not accord
with the higher level structures a discord results in either: i) the alien element being rejected, after a
period of conflict and tension; or ii) an adjustment to the higher structures to incorporate the new
element. This accords with the etymology of hierarchy that is noted by Simon whereby within a formal
system each tier is a subordinate subsystem situated within and responsive to the authority of a higher
tier (Simon 1962: 468).
14
In Figure 3 the two forms of accountability are placed within Williamsons framework. Relational
accountability sits at the highest tier, Level 1, and therefore presents the boundaries for the operation of
procedural accountability, which sits below, in Levels 2 to 4. The context of procedural accountability is
immediate and specific, whereas in contrast that of relational accountability relates to the wider social
system. Procedural accountability has a short term time frame and focuses on the immediate
consequences of actions, whereas relational accountability takes a longer term perspective, such that
immediate repayment for social favours is not a necessity. The rules underpinning procedural
accountability are formalised and simplified in contracts; relational accountability is characterised by
rules that can be informal and highly complex, accidental and may have evolved over a period of time.

Figure 3. Procedural and relational accountability within Williamsons framework

RELATIONAL ACCOUNTABILITY
Context: wide; long term
Rules: informal; implicit; accidental; complex
Power: a process; unlimited; fluid; latent; ideological
Behaviour: appears irrational from procedural
perspective; not measurable

PROCEDURAL ACCOUNTABILITY
Context: specific and close; short term and
immediate consequences of actions
Rules: formal; contractual; simplified
Power: contractual; limited; with principal or agent
Behaviour: rational; measurable

Source: Williamson (2000: 597)

Finally and resulting from these characteristics, the mode of behaviour of procedural accountability is
rational and measurable. When viewed from the perspective of procedural accountability, the behaviour
arising from relational concerns would appear to be irrational as it responds to wider social norms and
longer term accountability relations.
15
Considering the epistemological roots of procedural accountability and relational accountability,
procedural accountability is rooted within economics and draws upon rational public choice models,
serving a functional need to achieve the rationalist standards of effectiveness, efficiency, lawfulness or
greater productivity. (Gulrajani 2010: 8). Relational accountability expands the epistemological
boundaries and moves towards and beyond bounded rationality, which allows for the influence of social
norms and a psychological basis to decision making.

These characteristics of relational accountability address the shortcomings of procedural accountability

and in doing so present complimentary descriptors. Relational accountability therefore provides the
framework within which procedural accountability can function. If an occurrence arises through
procedural accountability that does not fit within the boundaries of the higher level relational
accountability, discord will result. This can lead to conflict in the form of game playing and the tension
between the need for adjustment could result in either the rejection of the procedural accountability
demand or a shift in the relational accountability part of the framework.

Relational accountability is subject to some of the critiques that have been levied at procedural
accountability. Whilst procedural accountability is based on methodological individualism, relational
accountability is also an actor-centric approach, however this is modified as the actor is firmly placed
within structures and there is continual feedback and a multitude of crossovers within and between the
different social structures and the individual. Combining procedural accountability and relational
accountability risks presenting a model of accountability that, whilst more applicable, is so complex that
its representation is elusive. However the objective of this paper is to provide specific examples that
illustrate how procedural accountability and relational accountability are interrelated and how they co-
exist.

In this paper the model of accountability will focus upon the central government public sector and how
the public sector interacts with the international development community through the delivery of budget
support. In the international aid environment there is a meeting of actors from different cultures yet
operating within a non-Western societal context. In analysing this, Section 3 will apply the model to the
aid context in Tanzania. Here the interaction and mutual dependency of the two types of accountability
can be seen: where procedural accountability is imposed as a tool by which donors understand the
development context, embodied in processes; however its operation is tempered by the relational
accountability context and methods and therefore it will be argued that a relational lens must also be
adopted to enable a full understanding.

3. FOREIGN AID IN TANZANIAN SOCIETY

The pattern and delivery of foreign aid will be briefly outlined, considering the changing patterns of aid
to Tanzania in the context of foreign aid relations.8 An analysis of the associated power relations and
credibility issues in terms of the interactions between donors and the government will be presented. It
will look at donor - government relations through the lens of procedural and relational accountability
analysing conditionality in Tanzania. Drawing upon fieldwork findings the application of the relational
notion of power will place foreign aid in the context of the wider system, briefly considering how it is
influenced by domestic political relations.
8 The data presented are as recent as possible, however due to the prevalence of data and budget planning and
implementation weaknesses, to ensure accuracy some older data are used and where possible secondary analyses are
referenced.

16
Aid dependency and modalities

Aid can be considered from two perspectives: the narrow perspective where it is a financial flow with
accompanying policy conditionalities; or from a broader perspective where it is taken to be greater than
the sum of financial flows and associated conditionalities and is seen as a pattern of social relations
shaped by context and history, drawing upon the relational approach to accountability. The latter
perspective will be applied in this paper. However to outline the scale of aid and to appreciate its impact
on the public finances of Tanzania the fiscal context along with the pattern and type of aid flows will be
briefly outlined. Along with a consideration of foreign policy objectives, this will allow an assessment of
the extent of and the changing pattern of aid dependency in Tanzania, and a consideration of the power
relations between donors and the government.
Aid inflows play a key role in both fiscal sustainability and the delivery of public services in Tanzania
and through the latter it is argued that the direct impact upon the population can be felt.9 Grants amount
to 28.6 percent of total expenditure, with 14 percent delivered through the general budget support and
basket fund modalities (IMF 2007: 29). Total aid inflows (grants plus loans) as a percentage of
government expenditure, have risen fairly rapidly from 28.8 percent in 2000/01 to 42 percent in
2007/08, making Tanzania one of the largest recipients of aid in Sub-Saharan Africa (General Budget
Support Donors 2007; Government of Tanzania 2008). However since that peak, aid has been
financing a declining share of the budget, and is expected to finance 28 percent of the budget in
2010/11.
Aid by delivery mechanism is shown in Figure 4 and the increasing share of budget support until
2003/04 is apparent. In more recent years there has been a change in this trend as donors have
exercised more caution, and have reduced their general budget support (GBS) allocations, Figure 5.
Placing aid in the context of domestic revenue collections and focusing upon the period from 2000
onwards, Figure 5 shows how tax revenue is almost four times greater that GBS during the decade.
Tax revenue financed an average of 56 percent of total expenditure, whereas GBS financed an
average of 15 percent. Development expenditure is the proxy for project inflows as, whilst it includes a
small domestically financed component such as matching grants or counterpart funding (which varies
by project), it is dominated by externally financed projects which comprised between 62 and 81 percent
of all development expenditure during the decade.10 Total foreign financed expenditure, projects plus
programmes, amounted to an average of 41 percent of total expenditure.

9 This has an accountability influence, however due to the necessity of imposing limitations upon the scope of the research
this has not been pursued. This is discussed further in Section 4.

10This is symptomatic of a weakness in the budget classification system as such a division is misleading and the
development and recurrent budget should be fully integrated.

17
 Figure 4. Mix of Aid Instruments 1999/00 to 2003/4

100%
90%
80%
% of Total Aid

70%
Projects
60%
50% Basket Funds

40% HIPC
30% Budget Support
20%
10%
0%
1999/00 2000/01 2001/02 2002/03 2003/04

Source: External Finance Department, Ministry of Finance, cited in (Booth et al. 2004: 16)

Figure 5. Tax revenue and GBS as a percentage of total expenditure*

70%

60%

50%

40% GBS

Tax revenue
30%
Development
20% expenditure
Total foreign financed
(proxy)
10%

0%
01/02

02/03

03/04

04/05

05/06

06/07

07/08

08/09

09/10

10/11

Source: Authors calculations based on Ministry of Finance and Economic Affairs data compiled from different
sources
* The GBS disbursement for 2009/10 excludes emergency funding of US$191m related to the global recession in
order to determine the trend in GBS.

Figure 6 shows the rate of growth of GBS, tax revenue and total expenditure. In 2007/08 GBS peaked
at Tsh 963 billion, reduced slightly to Tsh 941 billion in 2008/09 and increased slightly to Tsh 955 billion
in 2009/10 once the emergency funding to counter the effects of the global recession from the EC
(EUR14.8m) and the World Bank (US$170m) is removed. The trend decline resumed with Tsh 821.6
billion expected for 2010/11.

18
 Figure 6. Tax revenue, GBS and total expenditure, Tsh bn*

10150

8150

6150 GBS
Tax revenue
4150
Total expenditure
2150

150

Source: Authors calculations based on Ministry of Finance and Economic Affairs data compiled from different
sources
* The GBS disbursement for 2009/10 excludes emergency funding of US$191m related to the global recession in
order to determine the trend in GBS.

Conditionalities, negotiation and aid delivery

GBS has evolved since 2000 to be based on three main contractual mechanisms: i) a joint
memorandum between donors and government (the Partnership Framework Memorandum); ii) a
monitoring framework (the Performance Assessment Framework, PAF); and iii) individual donor
agreements with government. Each of these reflects the foreign policy objectives of donors, often
implicitly, and plays a key part in the negotiations around the delivery of aid. The examination of the
management of aid provides insights for the tensions between procedural accountability and relational
accountability. The frameworks are procedural in nature and structure, although their operation in
reality displays relational characteristics. It will now be considered how issues of power and foreign
policy aims are reflected in these agreements.

Foreign policy and aid

This section will discuss the foreign policy of Tanzania and her donors, with a focus upon the GBS
group. These two influences will then be brought together in a discussion of power, considering donors
in the Tanzanian context and the extent of their direct and indirect influence upon policy.
Tanzanian foreign policy

Post independence, the strongly nationalist and independent stance pursued by Nyerere resulted in
relations with foreign governments weakening, although there was a notably close relationship which
was maintained with China. The environment after the war with Uganda was one of regional tensions
and worsening relations with Kenya culminating in the collapse of the East African Community (EAC) in
1977 (Berg-Schlosser and Siegler 1990: 92). As discussed above the economic decline of the 1970s
resulted in a shortage of foreign currency and increasing (and reluctant) reliance on foreign credit.

19
Disagreement on the timing and extent of the reform measures to be taken and the social and political
consequences arose with the IMF, and the Tanzanian government was reluctant to accept the
conditionality of an extreme and immediate devaluation. The resultant break with the IMF in 1979
caused other foreign funding to cease, by virtue of the IMFs gatekeeper role (Berg-Schlosser and
Siegler 1990: 93).

Mwinyis presidency saw the Economic Recovery Programme (ERP) agreed in 1986, which required
devaluations in instalments and significant structural adjustments (Berg-Schlosser and Siegler 1990:
93). Subsequently aid began to return to Tanzania. After the collapse of communism in Eastern Europe
there was increasing focus and accompanying pressure from donors on democratisation and
accountability in Africa, reflected in the aid to Tanzania which was accompanied by discussions of
democratisation. It should not however be overlooked that this mirrored the domestic debate and
demands for democratisation (Shivji 2006: 8), ahead of the 1992 move towards multi-party democracy
and Mkapas era of international repoliticisation.

The return of aid should not be interpreted as extending to a positive relationship between Tanzania
and her donors, rather as Helleiner (2002: 251) notes the aid relationship was in a perilous state. The
response of Tanzanian leaders was more measured than that of Kenya as Nyerere maintained close
relations with many European governments. Choosing not to enter into direct confrontations with
overseas governments, Tanzania raised the issue of democracy at a meeting of the Organisation of
African Unity in 1990, with a resolution resulting. Hyden summarises the content:
African governments do take democracy seriously and want to be given the opportunity to
prove their commitment to this idea without being forced to do so by external political
conditionalities. (Hyden 1994: 97)

Mkapas first term as President commenced in 1995 in the context of a lack of trust between donors
and the government, based on perceptions of: a poor administration; corruption; lack of democracy;
budget mismanagement; and the prickly personality of the Finance Minister (Helleiner 2002: 251).
This environment was soon to change. Mkapas second term as President (until 2005) saw a stronger
pursuit of neoliberalism; the promotion of globalisation and entrepreneurship; and a close allegiance
between both donors and government on one hand and between the political elite and the business
elite on the other. These allegiances capture the importance of the characteristics of relational
accountability, including the personal and informal networks of support.

The aid relationship during this period was initially based on the relationship between the government
and the Nordic group of countries, with the World Bank and the UK coming on board a few years later.
The Helleiner Report of 1995 (Helleiner et al. 2002) represented a declaration to get the aid
relationship back on track and made recommendations to improve relations between donors and
government. One recommendation included the formation of an Independent Monitoring Group (IMG)
to monitor progress in aid relations. The report identified the capacity of the government as the key
constraint and pointed to the need to improve partnership and to develop greater local ownership and
trust between the government and donors. Gould (2005: 6) identifies this as a liberal project of state-
making where there is reliance instead of coercion. From then onwards the emphasis has been on the
embedding of an explicit performance focussed relationship based on a transparent monitoring and
evaluation (M&E) framework (Gore 2000: 795). This is characteristic of an increase in procedural
accountability characterised by the increasingly formalised approach of donors.

Helleiner (2002: 257) notes the movement towards a more genuine dialogue between the government
and donors since the 1995 report, with Tanzania being cited as a showcase for aid and donor-
government relations. Seeing the other side of this, Shivji (2006: 11) notes that Mkapa saw the
economy through the eyes of the IMF or World Bank, and that increasingly legitimacy was sought from
20
donors, rather than from the people. This is not to deny the improvements in the financial control
systems and other important reforms that were attained. It is necessary, however, to look beyond the
immediate to tell the story from the perspective of relational accountability and to question what
genuine dialogue is and what happens below the surface.

It is argued here that it is not as simple as an endorsement of neoliberalism. Rather the government is
fully aware of the external face it has to present in order to obtain foreign aid, and that real legitimacy is
sought from domestic, not international sources. The issue is the extent to which the government, or
President specifically, can be perceived as seeking legitimacy from donors and still be accepted
domestically. Mkapa was skilled at achieving both, and it could be argued that he built his domestic
legitimacy on his reformist policies which supported the private sector and in turn the gains of economic
growth were experienced by the influential middle classes. Therefore in the context of the accountability
framework he had fostered procedural accountability through his compliance with the policies of the
donors but also in co-opting the private sector into the elite he facilitated the redistribution of resources
necessary to buy and maintain the support for the ruling party, encouraging relational accountability.

The close relationship that was fostered between donors and government under Mkapas second term
is seen by some as a conditionality of objectives rather than a natural consensus (Harrison and Mulley
2007: 9). This reflects the perspective that a game was being played and the Tanzanian government
was facilitating the emergence of the allegiance with donors. Harrison (2001) presents a picture of what
he calls post conditionality reforms where the distinction between the external donor and the internal
politics of African states is no longer valid, arguing that instead both work together intimately and build
a set of mutual dependencies. It is argued here that the mutual dependencies involve a pulling in
different directions in pursuit of opposing interests in terms of the implementation of some reforms,
although there is a coincidence of interests to disburse aid.

To use the analogy of Kelsall (2002) such an argument states that debate is acted out on the air
conditioned veranda and that an external consensus is built. However, the rooms where the
governments position is formed are smoke-filled, such that complex interests involving relationships
between the state and wider society, the extended family and the public-private interface are captured.
Supportive of Kelsalls view, it is considered here that the apparent disappearance of the dichotomy
between the interests of donors and government between 2000 and 2005 is an illusion based on a
small elite group in government becoming apt at interacting with donors to reap their own gains,
whether directly or indirectly. Nevertheless they still have to maintain their circle of trust and to feed
their children or extended family or sphere of influence.11 This small group who interact with donors are
effectively put on display, whereas the real politics take place behind the shop windows, in the smoke-
filled rooms of the CCM (Kelsall 2002).

The approach of Mkapas two terms differs from that of Kikwetes first term, whose re-domesticisation
has involved a retreat from the sphere of traditional donors and a more focussed international policy.
This enabled the more active pursuit of a domestic agenda allowing closer attention to domestic
coalitions, which in turn had an impact upon the relationship with budget support donors. It may be that
an increased liberation from the need to service clientelistic networks in his second term will provide
Kikwete with the freedom to pursue more reformist and outward looking policies.

11 Kelsall (2008) notes and draws upon the analogy with food and children that is often found in Tanzanian society.

21
Donors foreign policy

Whilst each donor has specific foreign policy objectives that result in different styles of aid delivery and
specific interests dominating, there are some common pursuits. One foreign policy objective is that of
stability such that disturbances or violence are avoided as they risk disrupting economic interests. A
second is the interest in protecting and enhancing coalitions of investment that are in the donors
interests. The separation of such political concerns from the disbursement of aid varies amongst the
GBS donors. At one extreme, the World Bank has a policy of clear separation, whereas some bilateral
donors closely link the two. The incentives of the donors to disburse aid identified by Kanbur (2000: 6)
include such broader economic issues but also can include more personal incentives. He identifies four
motivations: avoidance of economic chaos risked by non receipt of aid inflows and a negative impact
upon the poor; political clientelism; risk of non servicing of debt; career interests of the aid agency
personnel and the maintenance of the reputation of the agency.

Whilst the higher level political interests are implicitly connected through aid being related to influence,
the explicit aim of foreign aid amongst the group of GBS donors is to pursue policy objectives in line
with both poverty reduction and the international aid agenda, and in doing so ensuring that increasing
amounts of aid are disbursed in a harmonised manner. This agenda has involved the movement
towards the more formalised nature of the aid relationship, along the lines outlined in the Paris
Declaration and in the Accra Agenda for Action, firmly located within the procedural tiers of Williamsons
hierarchy and which takes no account of the higher level influences of relational accountability. An
example of this is the Accra website that in a slightly paternalistic tone, states that Developing
countries are committing to take control of their own futures12 and pursues the same results-focused
and alignment issues as the Paris Declaration such as: ownership; country systems; untying aid; South-
South cooperation; transparency and accountability; conditionality and predictability (Eurodad 2008).
The emerging, and increasingly mainstream, governance discourse has probed the edges of these
pursuits of the international aid agenda with its indication of the importance of politics and questioning
of what underlies formal institutions. However its perceptions of rent seeking, clientelism and
associated characteristics as pathologies persist (Unsworth 2010: 6). The two core foreign policy
objectives of security and domestic investment highlighted above remain unreconciled with the rhetoric
of aid.

The distinction between GBS and non-GBS donors has been explicitly made here. Although this
research focuses on GBS donors, the influence of non-GBS donors is of interest as it has an impact
upon the engagement of GBS donors. The increasing international influence of China and the rise of
private sector financing have increased the availability of foreign aid and investment from non-GBS
donors. This can be argued to have had an impact upon the response of the Tanzanian government to
GBS donors and therefore is increasingly beginning to impact upon the behaviour of GBS donors. This
can be seen in the increasing interest of GBS donors in Chinas strategy, exemplified by the UK
governments aim to develop a partnership with China to deliver aid to Tanzania (Hawksley and
Mikwambe 2010).13

In summary Tanzanias foreign policy relations with her donors has clearly reflected domestic interests.
The internal and external demands for democratisation during Mwinyis presidency saw increasing

12

http://www.accrahlf.net/WBSITE/EXTERNAL/ACCRAEXT/0,,menuPK:64861886~pagePK:4705384~piPK:4705403~theSiteP
K:4700791,00.html

13 The interest of GBS donors in China was witnessed by the interest generated in the work of a visiting Chinese academic
to Dar es Salaam in June 2010.

22
procedural accountability and started the process of opening up to democracy. Mkapas era of
international re-politicisation endorsed neoliberalism and the continuing opening up to multiparty politics
legitimised CCM and maintained the relational links between business and politics. There was a
movement towards self-reliance instead of coercion as liberal state-building involved increased
legitimacy being sought from donors combined with increased procedural accountability which in turn
allowed donors to fulfil their interests to disburse increased amounts of aid. Kikwetes stronger domestic
focus, at the expense of the traditional donors, was arguably necessitated by the need to juggle
domestic networks of support.

After 2005 the relational accountability interests were increasingly focussed upon as those of
procedural accountability faded, so to what extent does this more explicit focus upon domestic
coalitions impact upon the relationship with budget support donors? The extent to which donors directly
or indirectly influence policy and therefore in turn the extent of domestic policy space, autonomy or
ownership will be considered in the following section on power.

Power relations and donor-government interaction

This section on power will discuss the power relations between donors and government, as represented
in their interactions around the disbursement of GBS and the review of progress in the Performance
Assessment Framework (PAF). In doing so it will attempt to draw out occurrences of the procedural and
relational understanding of power: procedural power being a limited resource that is contractual and
rests with the principal or the agent; whereas relational power is a process that latently underpins all
interactions, and is unlimited and fluid. The sections below will discuss the nature of interaction between
donors and government in the context of conditionality and power.

The question of government leadership and ownership has been continually raised and debated in
donor circles. The second generation Poverty Reduction Strategy Papers (PRSPs) were developed to
increase government ownership, yet they embody an externally developed neoliberal agenda. There are
two issues of primary concern which determine the extent of recipient government ownership: firstly the
balance of power between donors and government; and secondly the role of different groups within the
recipient government.

The PAF as the framework of conditionality for budget support is procedural in nature as it detracts from
the substantive and often political issues affecting development and in doing so restricts the involvement
of donors in policy dialogue to technical issues, even in the higher level forums. The level of detail
embodied in the PAF, which has contained more than 80 actions, has resulted in higher transaction
costs, intrusion which reduces national ownership, and a technocratic focus which militates against
political input (Booth et al. 2004: 49). The procedural nature of the instrument is accompanied by a
relational style of implementation, highlighting the conflict between and co-existence of the procedural
and relational frameworks. There is often collaboration between donors and government, and between
the Ministry of Finance and sector ministries to achieve satisfactory ratings despite performance being
poor. Further negotiation and strategising takes place as the working groups can update and propose
new indicators for the following year.

Conditionality and power

Models of conditionality address: the extent of the divergence or convergence of views between donors
and recipients (White and Morrissey 1997); short versus long term impacts of conditionality (Coate and
Morris 1996). Game theory models highlight the nature and possible outcomes of the bargaining

23
process (Mosley, Harrigan et al 1995; Kanbur 2000; Svensson 2000), and the representativeness of
government and the impact of policy reversibility (White and Morrissey 1997).

Applying some of these models to Tanzania, firstly the objectives of conditionality will be considered.
The six objectives of conditionality identified by Collier and Gunning will be applied to aid programmes
in Tanzania: 1) paternalism donors believing they know what is best for the recipient; 2) bribery
persuading the recipient to change their behaviour; 3) restraint recipient agrees to reforms and
conditionalities prevent policy reversal; 4) signalling that the reform programme is genuine; 5)
monitoring to ensure aid is having the intended effects; and 6) supportive provided needed
financing for reforms (Collier and Gunning 1996 cited in White and Morrissey 1997: 498). The analysis
of the Tanzanian programme is shown in Table 1, although a few caveats are warranted: in many
cases it is difficult to distinguish between paternalism and bribery and to do so requires further
consideration of the motivations behind the intentions to bribe the government to change their
behaviour. Consideration of the supportive aspect of aid requires an extent of abstraction from the other
objectives as almost all aid could be argued to be supportive. However the question is the extent to
which this dominates as an objective.

It is possible to identify examples that fall into the six different objectives of conditionality, drawing upon
the IMFs PRGF programme, the ECs Millennium Contract for budget support, and the World Banks
PRSC. A number of the objectives refer to the monitoring of performance using the PAF such as
monitoring and signalling through outcome indicators, and paternalism and bribery in relation to the
variable tranches. Technical assistance is the clearest example of the supportive objective.

Table 1. The objectives of conditionality in aid programmes in Tanzania

Examples
Objectives IMF PRGF EC variable tranche Millennium contract World Bank PRSC
Paternalism Letter of Intent the The selection of sectors for the variable tranches and the
IMF imposes criteria setting of indicators and targets was done in a joint donor-
that it considers to be government working group (2003/04). However due to
appropriate. capacity constraints in government the donors were
dominant.
Bribery Letter of Intent a Variable tranche The inclusion of
focus on certain To improve and focus on performance in certain areas. environment in the PAF.
reforms / areas Prior action
disbursements.
Restraint Inflation
Signalling Gatekeeper for GBS Outcome indicators in the PAF. Outcome indicators in
donors (condition in the PAF.
PAF).
Monitoring IMF PRGF missions. Ex-post monitoring of performance against indicators. Monitoring of
performance in PAF
with focus on prior
actions
Supportive TA support provided TA support provided in key policy areas where common TA support provided in
in key policy areas interests key policy areas where
through independent common interests
EastAfritac
Source: Authors application of Collier and Gunnings (1996) framework

To consider what type of conditionality exists in Tanzania a generalisation will be made across all
donors in Tanzania to apply the framework developed by White and Morrissey (1997).14 They outline

14There has been a broad movement towards ex post conditionality, as disbursements are made on the basis of policy
implementation or performance in outcome indicators. However ex ante conditionality is arguably present and the dialogue
and negotiation around the inclusion of policies in the PAF feeds into the allocation and disbursement of GBS funds.

24
four different cases of ex ante conditionality with full information of each actors choices available ex
ante but the recipient being allowed to renege on their agreement. Each donor, D, desires policy
reform, P; and each recipient, R, desires financing, F and as they each have to negotiate different
combinations of financing and reform their utility functions are U = U(F,P). Therefore (URF > 0) and
(UDP > 0). The four cases are as follows:

Case 1: UDP > 0 UDF > 0 URP > 0 URF > 0

Case 2: UDP > 0 UDF > 0 URP < 0 URF > 0
Case 3: UDP > 0 UDF < 0 URP > 0 URF > 0
Case 4: UDP > 0 UDF < 0 URP < 0 URF > 0

In summary: case 1 does not present any restrictions; in case 2, the recipient is resistant to policy
reform; in case 3 the donor is resistant to delivering finance; and in case 4 these two resistances are
combined. In Tanzania since 2000, cases 3 and 4 can be ruled out as there is very little resistance to
disbursing funds. This is based on a generalisation ruling out specific and unusual cases, for example
the case of 2007 where there was a delayed notice of GBS disbursements due to the dispute
concerning the External Payment Account corruption case.

Figure 7. Case 2 conflict over preference for reform

In applying this framework the case in Tanzania relates to the extent to which the government resists
policy reform. Case 2 therefore provides the most relevant analysis as it considers the outcome to be
determined by bargaining, which depends on how the recipient can play the game of agreeing to the
tighter conditionality, knowing that slippage will result. This models a conflict in the preference for
reform as (URP < 0), given the budget constraint P*F*, Figure 7. Due to this constraint the recipients
utility curves are rotated by 90 degrees and utility is maximised along URMAX at point c and the donors
utility is maximised at point a. Therefore the final outcome will be somewhere between a and e and will

25
be a function of bargaining between R and D with the donor using tighter conditionality to move towards
point a, through a rotation towards P=CTF. The recipient can play the game of agreeing to the tighter
conditionality, knowing that slippage will result. The slippage will move the outcome towards point d in
a single period game however in a repeated game it once again presents the credibility dilemma as the
donor can increase both UR and UD by disbursing anyway, moving up to point d.

The model describes how the recipient can use slippage as a strategy, agreeing to a tighter
conditionality than they are willing to implement and it raises the issue of the credibility of conditionality
in a repeated game. As White and Morrissey highlight, slippage is the recipients strategy, presenting a
credibility dilemma to donors of whether to punish slippage in accordance with the conditionality
agreement or whether to disburse anyway and increase utility. This can be assessed through
considering how the PAF has been used.

The coverage of the PAF broadened the content of the policy discussions and became the tool through
which donors monitored and in some cases attempted to influence policy. The stated aim of the PAF to
prioritise existing targets and measures and link their achievement to the provision of budgetary
support (United Republic of Tanzania 2001: 3), underestimates its usage as the donor agenda has
been pushed forward through some PAF actions. This leads the government to retain space in policy
formulation and reform in three ways. Firstly through maintaining a strategic ambiguity whereby some
room to manoeuvre is retained through extent of ambiguity in reform commitments which creates the
space for later reneging upon agreements and pursuing policies which may be in conflict with the
assumptions underlying the commitments (Richey 1999). Secondly through slippage in implementation;
or finally through policy reversal. It should be noted that these are ex-post responses, and take place in
the context of limited capacity in government. Where government capacity is greater a preferred
response would be ex-ante, staying a step ahead of donors and either being able to make policy
proposals, or to make clear responses to donors proposals. It is through these routes that the influence
of the PAF is restrained.

Fieldwork analysis and findings

Fieldwork results interpreted accountability in foreign aid and the relationship between donors and
government to address the research question: what are the accountability relationships involving
general budget support and public financial management in Tanzania? In attempting to uncover what
these accountability relationships are and how they have an impact, the research question was
organised around three hypotheses:
H1: The higher the budget support inflows the higher accountability to donors.
H2: The higher the number of policy conditions the higher the accountability to donors.
H3: The higher the accountability to donors the lower societal accountability.

Each hypothesis was considered through the lens of procedural accountability, analysing the identified
independent and dependent variables, and was subsequently critiqued through taking a relational
accountability perspective.

The analysis of procedural accountability focused on the contractual mechanisms for regulating
behaviour between the donors delivering general budget support and the government representatives.
It considered how these functioned and how they influenced disbursement outcomes. In turn the
relational accountability analysis considered the relations between those involved in the delivery of
general budget support and how they operate within the wider socio-political system. It assessed how
the contractual mechanisms functioned in reality and placed procedural systems in the societal context.
26
It was found that the traditional understanding of accountability, procedural accountability, presented a
limited understanding of reform challenges and tensions between donors and government.

The apparently procedural mechanisms and measures, for example external audit, the PAF and public
financial management indicators such as the Public Expenditure and Financial Accountability (PEFA),
rely upon relational elements for their implementation, however such relational aspects are overlooked
when the procedural lens is looked through. In contrast the analysis of relational accountability stressed
the role of personal connections that function through networks and presented a picture of the
interconnections between the different networks and positions that people hold, confirming the complex
web of identities. This highlighted how procedural mechanisms may need to be adjusted to respond to
relational accountability demands.

The procedural accountability analysis, using proxies for accountability, found that accountability to
donors was partially influenced by the amount of GBS disbursed as the largest donors received the
greatest accountability response from government. However the relationship has neither been linear
nor constant over time and accountability is found to be episodic and greatly influenced by the
relational attributes of individual personalities and reinforced by a thought collective operating amongst
donors. Conditionality was found to have no accountability impact, but it was either the focus of
negotiation and collusion between donors and government or used as a tool to retrieve some policy
space.

The procedural accountability analysis found societal accountability to be weak in the analysis of H3
and there was no evidence of it being influenced by accountability to donors, despite donors having an
increased focus on domestic accountability from 2006 onwards. This is a particularly important
question in understanding societal accountability. So how do the two types of accountability operate
together? Procedural accountability operates within the relational context. Applying Williamsons
framework, relational accountability exists within Level 4, providing the societal context for procedural
accountability, in the lower tiers.

The government has a predisposition towards relational accountability and the donors towards
procedural accountability and respectively these are the dominant approaches of each group. Amongst
donors, however, the employment of either relational or procedural accountability in their interactions
with government is not static, but is fluid and changeable depending on the specific context and
interests that need to be served. For instance, in terms of GBS, procedural accountability facilitates the
disbursement of funds, as it provides the necessary evidence to underpin the contracts between
donors and government, although relational accountability characteristics are displayed, as some
negotiation and manoeuvrability is needed to interpret and apply the indicators.

The stated aim of the GBS donors is to pursue policy objectives in line with both poverty reduction and
the international aid agenda, and the policy constraints of the latter has greatly influenced the behaviour
of donors during the last decade. There is a dichotomy between the informal nature of the relationships
in-country and the more formalised direction in which aid reforms move, guided by the international aid
agenda. Evidence from amongst the donor group in Tanzania points to the increasing formalisation of
the donor-government relationship and deterioration in the nature of the relationship in terms of access
to information and cordiality of relations. At one level this has been encouraged by the Tanzanian
government with the adoption of quiet times and support for the more formal working practices of the
GBS group of donors however such examples clearly represent the possibility of reduced transaction
costs for government. Other examples of increased formalisation have been less popular in
government circles, including the PAF monitoring framework, the variable tranche of budget support
and evaluations such as the PEFA and the Country Financial Accountability Assessment (CFAA). Such

27
frameworks or reviews fall firmly within the procedural accountability category. This operates alongside
relational accountability which predominates in the implementation of the agreements.

Both groups are able to employ behaviours that fit with the other form of accountability when
necessary, although the dominant frameworks are reverted to in order to understand the environment.
It is through donors reverting to their dominant procedural framework that they look above the surface
and avoid the reality of how state and society functions in Tanzania.15 This is inappropriate to the
Tanzanian context where the need to maintain the support of the winning coalition in the context of a
limited access order economy requires that redistribution takes place along patron-client networks
(Khan 2004; North et al. 2007; Lindemann and Putzel 2008). This results in the government being
accountable to interests other than donors and this accountability operating within a relationally
dominant framework. Kelsall (2008: 15) cites anecdotal evidence that points towards the continuing
intervention of senior officials, ministers and presidents in day to day affairs, such that formal rules are
frequently pushed aside. In the context of weak administration and inadequate capacity that prevents
the following of rules, such an approach can be more effective.

Considering Tanzanias changing aid modalities and the good governance agenda, what are the
implications are for the maintenance of political alliances? As a greater share of aid funds are
channelled through budget support the question is raised as to what impact this has on the payments
that local politicians can make to their constituents to secure their support. How do the benefits evolve
in response to these changes? For example, benefits in kind such as clean audits being rewarded by
feasts (Kelsall 2008: 16). This leads us to the often cited issue of workshops: the associated benefits of
per diems, meals and transport provision and the frustration of donors that the provision of such
benefits is the only way to ensure attendance. The civil service reform component of the good
governance agenda has focussed on increasing salaries for public servants and the removal of per
diems. This has been pursued without consideration of the wider social context and the blurred
interface of the public and the private. The extended family, at the core of African society is often the
recipient of increased salaries of public servants and it is only through benefits such as meals, transport
and per diems that the civil servants themselves can retain the benefits and therefore may be motivated
to improve their performance (Kelsall 2008: 18). This is yet another example of the imposition of a
procedural approach to reforms that is in conflict with the more pervasive relational context where the
extended family or circle of trust dominate and personal networks are key to the functioning of the state.

As the framework that donors use is procedurally dominant, this results in them overlooking the
importance of networks for the government, in terms of it ensuring that it maintains support, a support
that is central in maintaining a stable socio-political environment. Whilst stability is an important foreign
policy objective of donors, the importance of relational accountability in achieving this is purposefully
overlooked in the preference for procedural dominance in the implementation of aid.

The different frames of reference of donors and government inevitably results in instances of conflict in
dialogue and reform and during the last half of the decade the procedurally-dominant approach of
donors has clashed with the highly personalised domestic relational accountability context and there
has been a deterioration of trust on both sides. The Tanzanian government has been able to use their
relationally-dominant approach to gain ex-post policy space, in terms of the use of the PAF through
game playing around conditionality. This occurs in the context of limited capacity within government
and allows the interests of powerful groups to be served and stability to be maintained. Game playing

15A distinction should be made between what donors will formally express whilst representing their organisations and the
understanding of some individuals which goes much further and understands the limits to the procedurally dominant
approach. This critique of the approach of donors is based on what they are able to formally express and therefore it may
not capture the more in-depth understanding of some.

28
can also take the form of collusion between donors and government to accord with the dominant
strategy to disburse, facilitated by a shift towards a more relational mode of operation.

In the context of a strategic game, the use of GBS is one of the few tools that donors have to send
signals to the government. The reduced acceptance in Tanzania of the neo-liberal consensus that has
been forged by President Mkapa was apparent in calls for a reduced dependence upon donors and the
ability to pursue policies of Tanzanias own choosing, vocalised by politicians from all parties and
academics. For the Tanzanian government, slow progress disguised other interests and allegiances,
and whilst it can be argued to be a display of government leadership, the real motivating factor is the
need to maintain accountability to internal party factions and to ensure that the interests of influential
members of the elite are protected.

The role of different groups within the Tanzanian government should be highlighted to be aware not just
of the operation of different interests, but of conflict of interests and to stress that the government
should not be viewed as a homogeneous entity. Costello (1996: 137) notes the differing extent of
influence of donors in the case of the Ministry of Agriculture and Livestock (Kilimo) and the National
Development Cooperation (NDC) and finds that in Kilimo foreign aid was accompanied by carrots and
sticks to influence policy. For the NDC this was less the case, pointing to a variance of views across
different government agencies in the context of their retention of autonomy due to the failure of the
state to define its role.

The above discussion of negotiation between donors and government is based on the view that any
leadership is not broad-based but is focussed around a small cooperative elite centred in the Ministry of
Finance. Eyben (2007) referred to this as group think and Fleck (1979) and Douglas (1986) as a
thought collective that has a specific thought style whereby a mutually reinforcing perspective and
dialogue is embedded amongst a small group of donors and a small group of government officials.
Harrison and Mulley (2007: 19) identify two characteristics of the absence of a broad leadership base:
the rise in the small group of officials who directly benefit from aid (from the perks of workshops and
international travel) and with whom the donors feel most comfortable working; and the cultural effect of
aid on Tanzanian governance which is largely self reinforcing and exists to ensure its own reproduction.
This was specifically a characteristic of the neoliberal agenda of Mkapa.

It is from within this elite group of officials that aid policies and development management discourse
that will appeal to donors is predicted (Harrison and Mulley 2007: 24). It is interesting to note that the
Public Expenditure Review (PER) sector evaluations are supported by the Ministry of Finance but are
resisted by some sectors and seen within the sectors as representing an increased work burden which
does not feed into the budget negotiations. This highlights the differing views held within government
and the lack of representativeness of the donor-friendly group of senior officials within the Ministry of
Finance who are in the elite group and help to maintain the illusion of consensus and mutual
dependency with donors. Bigsten (2001: 327) has gone further to suggest that the policymakers
anticipate which policies would appeal to donors and create their own policies on this premise.

This section has presented an initial review of the context of foreign aid in Tanzania: attempting to draw
out the power relations underlying the negotiations and the incentives for policy involvement. The PAF
emerged so that donors could be actively involved in the prioritisation of policy measures, using the
carrot of financing as an incentive; however the discussion of the use of the PAF demonstrates that
some policy space is effectively negotiated by the government. Whilst the PAF is firmly rooted in
procedural accountability and this is increasingly the dominant approach of donors, an element of
relational accountability does exist as negotiations and collaboration take place between donors and
government. In many cases, despite conflict being present, the incentives of donors to disburse
dominate. This interaction between procedural and relational accountability is important and it is only
29
through the consideration of how both function together that an understanding and a complete picture of
how accountability works in Tanzania can be generated. Relational accountability is frequently either not
considered at all or is perceived as representing negative influences that should be eliminated. Rarely is
its centrality to the effective functioning of the Tanzanian state recognised. Overlooking the relational
aspect of accountability is detrimental to the understanding of how state and society works. This is not to
present relational accountability as a catch-all that can explain all behaviours, rather it is presented as a
new lens through which to look at the aid relationship, a lens that is more firmly rooted in the wider
cultural and historical context.

Summary
This section has analysed the provision of aid, specifically GBS in Tanzania and has placed it in a
broader perspective where it is part of the pattern of relations shaped by the social context and
historical episodes. This section has built upon the state and society analysis undertaken in Section 2
to present an examination of the changing pattern of aid dependency in Tanzania, conditionality and
negotiation and the power relations between donors and government. The discussion has pointed
towards the coexistence of the procedural and relational aspects of accountability, and has highlighted
cases where the increasingly formal procedural elements are not aligned with the more pervasive and
higher level relational structures.

The recent history of foreign policy and relationships with donors presents a picture of underlying
tensions, although these have frequently not been allowed to impact upon external relations. Initially
Nyerere, and then later presidents, have trodden a careful path around issues that have had the
potential to present challenges to the foreign aid relationship. It is at times when challenges have arisen
that the stance of governments and donors and the style of their interactions is most interesting.

The rise in the administrative style of governance became dominant during Mwinyis presidency and
was developed further under Mkapa after 1995 as a close relationship was fostered between the
international agencies and a small group of senior government officials. The reformist presidency of
Mkapa arguably marked the start of increasing procedural accountability, with an increasing focus on
improving public auditing and taxation collections. As aid inflows to Tanzania increased, there were
increasing demands in terms of the resources required to manage and report upon funds and to
maintain a dialogue with donors. This saw the emergence of a core group of senior officials,
concentrated in the Ministry of Finance and the Presidents Office, who played the role of trusted
individuals, to which much of the donor confidence in Tanzania can be attributed. Kikwetes more
focussed international policy and the retreat from the sphere of traditional donors to respond to
domestic coalitions during his first term has been one of the factors in the deterioration of the
relationship with donors.

These changes in Tanzanias foreign policy relations with her donors have in common the consistent
reflection of domestic interests. The international demands to open up to democracy during Mwinyis
presidency coincided with the start of multi-party democracy. Mkapas era of international re-
politicisation further legitimised CCM and co-opted the business elite into the core of the state elite. The
recent move towards a more explicit domestically focused policy pursued by Kikwete was influenced by
the need to respond to domestic clients who are less reformist and more socialist in their outlook than
those to whom Mkapa was responding.

Negotiation with donors has been discussed in terms of conditionalities and the extent of divergence or
convergence of views between donors and government. The annual assessment of the PAF monitoring
framework provides the negotiating forum between government and donors and therefore is the arena
30
in which power is demonstrated and exercised. The PAF itself is procedural in nature although is
accompanied by a relational style of implementation, which disguised other interests and allegiances,
maintaining accountability to internal party factions and ensuring that the interests of influential actors
were protected. As White and Morrissey highlight, slippage is the recipients strategy, presenting a
credibility dilemma for donors of whether to punish slippage in accordance with the conditionality
agreement or whether to disburse anyway and increase their utility. This is supported by Richeys
concept of strategic ambiguity which the government aims to maintain in its negotiation with donors
(Richey 1999).

Fieldwork found that the increasingly formalised international aid environment of the Paris Declaration
and the Accra Agenda for Action is juxtaposed to the relational context that frames the dances between
donors and government. However government has to juggle its external face for donors and its internal
face where it maintains the support of its clients through the distribution of resources. This is an aspect
of relational accountability which represents accountability to powerful actors (including elite groups),
necessary for the survival of the state. It remains the case that the private extended family is a primary
circle of concern to which payoffs are made, however the same concept can be applied to the
maintenance of political and business alliances which can be considered to be the public extended
family. Reiterating the argument of Kelsall (2002), there is little distinction between the public and the
private such that public office is often used for private benefit and from this it can be considered that the
public extended family and the private extended family merge.

The domestic sphere, which is dominated by relational accountability characteristics, is accompanied

by an increasingly formal international arena, dominated by procedural accountability. Donors rarely
look below the surface to understand the complex domestic social context. Whilst procedural and
relational accountability can and do work together, conflict results when the formal, the contractual and
the measurable lens of procedural accountability is looked through in an attempt to understand the
wider socio-political and economic context. Aid most effectively supports reform in the rare instances of
a coincidence between donor and recipient interests such that the procedural accountability
requirements of the conditionality do not conflict with the higher level relational context.

4. CONCLUSIONS: RECONSTRUCTING ACCOUNTABILITY

This section revisits the conceptual framework, summarising the new theory of accountability, as the
understanding of accountability is reconstructed. It reviews the political economy of Tanzania in the
context of the new framework, from a domestic perspective drawing upon the findings from fieldwork in
Tanzania. It argues for both procedural and relational accountability to be considered together, as they
closely interact in the delivery of aid. The policy recommendations that emerge from the research are
discussed. These are followed by suggestions for further research, which involve both an extension of
the current analysis to examine its conjectures in greater detail and an exploration of areas that were
precluded due to the need to limit the scope of the research. Firstly the conceptual framework is
summarised.

Analysis and interpretation: the accountability framework

A new theory of accountability has been formulated by recognising the limitations of the traditional
(procedural) form of accountability, pointed to by a number of writers (Bovens 1998; Eyben 2007;
Ebrahim 2009; Gulrajani 2010). The procedural understanding of accountability was extended to
embrace concepts drawn from sociology, philosophy and anthropology such as culture and social

31
structure. Such influences, within the narrower procedural accountability can result in behaviour that
appears to be inconsistent and irrational. Expanding the frame beyond the restrictions of procedural
accountability, the influences become complex, iterative, recursive and dynamic characteristics that
are found in reality.

A comparison of the characteristics of the two types of accountability highlights such differences.
Firstly, the context of procedural accountability is immediate and specific, resulting in a focus upon the
short term and immediate consequences of actions. In contrast that of relational accountability relates
to the wider social system and takes a long term time frame whereby reciprocity may not be immediate.
This wider perspective results in the rules of relational accountability being complex and informal,
possibly evolving over a period of time and therefore being highly embedded in terms of culture and
structure. Juxtaposed to this, the rules underpinning procedural accountability are formalised, specific
and often simplified in contracts. These characteristics feed into the nature of power and the
behavioural outcomes. Through the procedural accountability lens power is a contractual resource that
is limited, and rests with either the principal or the agent. In contrast, relational power is fluid, unlimited
and latently underpins all interactions. Finally, the mode of behaviour of procedural accountability is
rational and measurable but when viewed from the perspective of procedural accountability, behaviour
arising from relational concerns appears irrational as it responds to wider social norms and longer term
accountability relations.

These characteristics of relational accountability are located within the highest tier of Williamsons
social hierarchy and provide the boundaries within which procedural accountability can operate, located
at the lower levels. Specifically, procedural accountability resides within Levels 2 to 4 of Williamsons
hierarchy and exists within the broader concept of relational accountability, which has a deeper and
more durable structure and is located at Level 1, as illustrated in Figure 4. If an occurrence arises
through procedural accountability that does not fit within the boundaries of the higher level relational
accountability, discord will result. This can lead to conflict in the form of game playing and the rejection
of the procedural accountability demand or a shift in the relational accountability framework, seen in the
analyses of GBS conditionalities in Section 3.

The interaction and mutual dependency of the two types of accountability is key. Procedural
accountability is imposed as a tool by which donors understand the development context, embodied in
processes; however its operation is tempered by the relational accountability context and methods. This
can be emphasised by considering how the two may operate in isolation. If the procedural
understanding is applied in isolation there is no analysis of the highly context specific and complex
cultural factors that underlie the relationships around which accountability is formed. This has the result
that it is not possible to understand the motivations and the incentives of actors and in turn to
understand their responses to different scenarios a typical frustration amongst donors. In turn, if only
the relational accountability framework is applied the ability to comprehend both shorter term progress
and closer accountability, and to frame our immediate context is diminished a limitation that can be
found amongst recipient civil servants. By extension, the two types of accountability are
complementary. As discussed above, they can however also be in conflict if the demands of procedural
accountability push outside the framework of relational accountability. This results in either the
resolution of the conflict at the procedural level, such that the demands of one party are refuted; or it
may result in the procedural demands being accommodated by an evolution in the relational
accountability characteristics. Such an adjustment in the higher level relational characteristics is,
however, unusual as these are the highly embedded attributes such as social norms, customs and
other complex informal institutions, which as Williamson (2000) outlined, can take millennia to change.

This paper has presented a perspective on the elite and power which underpinned the concept of
relational accountability as it helped to describe how networks are structured. The elite is pivotal in
32
formulating policies and taking decisions, and links the state and society as it reaches out in a web-like
structure capturing actors as they function within a multitude of different social roles within different
networks. As elite factions compete and co-opt actors from society, the elite brings in an interplay
between the institutional and the structural and the personal and social. The party and the political
leadership are intertwined within these structures of power and legitimacy. The elite in Tanzania is
comprised of factions, in competition with each other, not highly centralised and based upon clientelistic
relations. The operation of clientelistic relations based on informal rules depend upon retaining the
network of support of the elite, embedding actors and providing the knowledge and the respect that
facilitates the fulfilment of their designated roles (Evans et al. 1996; Kelsall et al. 2010: 23). Such
networks are key to the functioning of the state.

The paper has argued that this structure is reflected in often apparent contradiction in the way that
people behave as the flexibility in responses and roles in response to the complex reality on the ground
does not easily fit into the procedural notion of accountability. Specifically, the notion of the elite and
their role in society does not fit well with the traditional procedural notion of accountability which
provides no space for the differential influence, through power, of different groups. The use of the term
structure is not to imply that it is constant or fixed, rather as different agents and different groups exert
different influences at different times, the elite is fluid, mirroring the dynamism of social relations. This
allows a movement away from the assumption embodied within procedural accountability that there is
equality across actors and that society is simply an aggregation of individual preferences, and therefore
highly impersonal in nature.

The conceptual framework applied

The different periods of Tanzanias post independence history have been discussed and, whilst longer
term trends and differences have been highlighted, the main analysis focused upon the period since
2000. This period in itself captures two different leaderships and changes in the aid relationship which
highlight the inappropriateness of procedural accountability and the need for this new model with its
extension to relational accountability.

The decade since 2000, has broadly constituted the second term of Mkapas reformist presidency,
during which a close relationship with the IFIs being nurtured; and the first term of Kikwetes presidency
that saw an increased domestic focus with a reinstatement of the reach of the state and the
maintenance of legitimacy through increased voice. Mkapas second term saw the co-option of the
business elite into the political elite, enhancing the ability of the party to redistribute resources to buy
and maintain the support of the wider elite. However Kikwetes first term highlighted the factional nature
of the elites inner core as within a multitude of networks coalitions are serving particular interests and
continual contestation results in fluidity as actors move between factions and between positions and the
public and the private are merged. It was against this background that the interaction between
international donors and the Tanzanian government was analysed.

Section 3 placed aid in the context of the relations shaped by the social context and episodes in
Tanzanias political and economic history. Discussing cases where the increasingly formal procedural
elements are not aligned with the more pervasive and higher level relational structures, the analysis of
power highlighted the coexistence of the procedural and relational aspects of accountability. It
recognised the recent increasing tensions in the aid relationship with GBS donors, but alongside this
the dance that has been skilfully played by the Tanzanian leadership who, following Nyereres example,
pursued domestic interests through a non-confrontational foreign policy. Mkapas reformist presidency
developed a close relationship with donors that was focused on a small group of senior government
officials and coincided with the start of increasing procedural accountability in the international aid
33
agenda. Increasing aid flows cohabited with increasing demands to report upon funds and to formalise
dialogue with donors. The more domestically focused policy pursued by Kikwete was influenced by the
need to respond to domestic clients as he retreated from the sphere of traditional donors. In the context
of the continued formalisation and result orientation of the international aid environment, tensions arose
with the predominantly relational domestic context.

The section argued that the clientelistic politics of the domestic sphere, characterised by relational
accountability, cohabits with the international, which has become increasingly formalised and is
characterised by procedural accountability. The need for donors to respond to the demands of their
headquarters reflecting the international aid agenda has, despite the calls for increasing country
ownership, resulted in a limited will to understand the complexities of country politics. The result has
been the exclusion of the complex domestic sphere and a focus upon the formal, the contractual, the
measurable and the shorter term.

It is where there is a coincidence of interests between the donor and the recipient that aid most
effectively supports reform, such that the procedural accountability requirements of the conditionality
mechanisms do not conflict with the higher level relational context. It was acknowledged in Section 3
that if this were the case conditionality would not be necessary. The extent of divergence or
convergence of views between donors and government was examined in the nature of conditionalities
and the supporting frameworks. The interplay between procedural and relational accountability is
demonstrated in the reliance of procedural monitoring frameworks, such as the PAF, upon relational
characteristics for their implementation.

Policy recommendations
This section makes a number of policy recommendations that are based on the incorporation of the
new model of accountability, focussing on an adjustment by donors. The recommendations are
grouped around two key issues: i) adjusting the approach of donors to take into account the constraints
associated with relational concerns; and ii) to deconstruct and to recognise the accountability
constraints upon donors. These recommendations support those emerging from recent governance
literature that focuses on the relational and argues for a movement away from an understanding of aid
and development that is rooted in Western experience (Gulrajani 2010; Unsworth 2010).

The recommendations made here have the potential to reduce the conflict between donors and
government and also to enhance the effectiveness of donor support, with the scope for beneficial
effects upon reforms and development. Whilst this research has been based on a case study of
Tanzania, these recommendations are generalised to other countries that are highly aid dependent as
the model of accountability is considered to be applicable to many countries where aid supports a large
proportion of the budget, proposed below as a suggestion for further research.

Turning to discuss each recommendation in turn, firstly, the adoption of the new model of accountability
requires an adjustment in the approach of donor agencies to take into account the different frames of
reference that arise due to constraints imposed by relational accountability concerns. This should be
based on greater attention to the political economy context of the recipient country. The outcome of
such a shift could be expected to be seen in terms of both the content of aid policies, the operation of
the delivery modalities and the style of interaction with recipient governments. This, however, requires a
fundamental adjustment in both the individual thought process of each donor agency and in the thought
collective that they comprise. This involves a movement from the analysis of development problems in
terms of deficiencies when compared to this ideal [Western] model, based on a procedural
understanding (Unsworth 2010: 1).
34
A central concept in adopting a relational understanding is that of the elite and their influence upon, or
the constraints that they may present for development, discussed in this section building upon the
model presented in Section 3. Whilst there are characteristics of the elite that cannot be influenced by
donors and foreign aid, it is important to understand those areas where elite interests positively impact
upon development and how such positive outcomes can be supported by donors. There may also be
openings for donors to influence elite interests at the margins which would require identification.

Defining what is meant by positive outcomes through an application of the relational perspective to aid
policy and how such outcomes could be influenced by donors overlaps with the work of research
programmes such as the Africa Power and Politics Programme and the Centre for the Future State.16
Part of these considerations involves working with different partners, identifying the nature of the
relationship that sits close to the interface between the state and the business elite, and the potential
that this may have for development outcomes.

The movement towards a balance between relational and procedural approaches and the shift in the
approach that is subsequently required from donors has other implications. In particular, the need for
donors to accept that their influence upon development outcomes is less than they currently assume.
This requires a shift from donors attempting to drive the path of development and assuming a
responsibility for delivering on the Millennium Development Goals (Unsworth 2010: 76) to identifying
where they might be able to intercept with domestic circumstances and to support specific
developments.

The second group of recommendations concern deconstructing the accountability constraints upon
donors to consider what these means in practice. The adoption of the Paris Declaration as a theology
has constrained the behaviour of GBS donors such that the increase in group think has resulted in a
trend to increasingly focus upon procedural accountability. This is evident in an increasingly formalised
dialogue structure and increased results orientation, and a deteriorating relationship with government
as the scope for accepting the importance of relational accountability has diminished. The
understanding of results and monitoring needs to be adjusted such that there is a movement away from
a short term focus and a reconsideration of what relational accountability means for development.

In addition to these international trends, donors, as representatives of elected governments, have to

respond to their headquarters, which in turn are responsive to their own constituents and elite networks.
The incentive structure in donor agencies is such that employees have to respond to headquarters and
this can result in the pursuit of policies and a style of delivery both of which are procedural in nature,
taking little account of the recipient domestic context.

The necessity of limiting the scope of this research has, in particular, limited the application of the
analysis to policy implementation, and has resulted in the restriction of the research to the development
of a new framework for understanding accountability. Further work to develop the conclusions in a
direction that makes them more applicable for policy is presented in the following section.

16
Africa Power and Politics Programme http://www.institutions-africa.org/ and the Centre for the Future State
http://www2.ids.ac.uk/futurestate/

35
Suggestions for further research

The suggestions for further research arise both from the discussion and conclusions of this paper and
the boundaries to the investigation. Possibilities for further research with respect to both of these will be
presented in turn. Firstly, drawing from the research findings further research into the nature of the new
accountability model is suggested, followed by further analysis into the relationship between foreign
policy and aid strategy, and also into the nature of relational accountability and how it changes over
time.

As discussed, combining procedural accountability and relational accountability risks a model of

accountability that is so complex that its representation becomes elusive. Efforts have been made to
avoid formulating an overly complex model of accountability by detailed analyses of the different
components of the framework. Further research could pursue this. Indeed, the new model of
accountability has not, in this thesis, been extended to consider its implications for development
potential and policy. This could be extended in many directions, and perhaps most urgently into an
enquiry of what this means for economic growth and development. For instance it could examine what
the relational accountability context implies for the provision of public goods, building on the work of
Khan and Gray (2006) that looks at the tensions between the power of factions and the leadership in
the competition for rents. Concerning further research to enhance the relationship between aid and
development, questions that could be asked include which industries are most closely connected to the
state, who are the members of the elite and how do their interests overlap with developmental concerns
such as technological adaptation and innovation. Concerning inequality and poverty reduction, how
narrow is the elite and to what extent does its persistence depend upon networks that reach into wider
society. Further research into the factional nature of the elite, the extent of the competitive threat and
the subsequent time horizons require consideration to understand the extent to which rent seeking
could support development.

As discussed above, the incentives produced by the objectives of foreign policy could be explored. One
foreign policy objective is that of stability such that disturbances or violence are avoided as they risk
disrupting economic interests. A second is the interest in protecting and enhancing coalitions of
investment that are in the donors interests. How donors could respond in the context of the constraints
posed by accountability to their taxpayers could be pursued in an analysis of alternative approaches to
the good governance agenda and policies towards corruption.

This research may also be extended to examine the broader and changing context in which foreign aid
is provided, the increasing international influence of China and how the rise of private sector financing
has increased the availability of foreign aid and investment from non-GBS donors. It could be argued
that this has had an impact upon the response of the Tanzanian government to GBS donors, and
therefore in turn may impact upon the behaviour of GBS donors. Such an exploration could provide a
different perspective upon the aid relationship between GBS donors and the Tanzanian government. To
further examine the implications of this research for policy, a closer comparison of the strategies of
individual donors could be undertaken. For instance, it is interesting to note how some donors, notably
USAID, have consistently pursued an independent and unharmonised approach and an examination of
their incentives and an understanding of the incentives for doing so could be enlightening.

The non-linear trajectory of change has been alluded to in this research, and deserves further
exploration within this new theory. In this regard, an interesting and important extension could be the

36
outcomes of conflict over time and the implications for adjustments within the framework. It has been
argued that the characteristics found within the relational tier are the most resistant to change and
therefore the most likely response to conflict is through adjustments in the procedural levels. It would be
interesting to test this assertion through the examination of specific instances of conflict, following them
through to analyse the trajectory of change.

Turning to the suggestions for further research arising from the boundaries of this research, the
constraints imposed, whilst at times frustrating, have resulted in a wealth of suggestions for further
research.

This research has focused upon accountability within central government, thus overlooking key
accountability relationships at the local level. To establish a more complete understanding of
accountability, it is necessary to extend the analysis beyond the scope of this research into the sphere
of local government. It is here that the impact of accountability relationships upon the lives of the actors
in society can be more keenly felt. The examination of accountability institutions at the central
government level is also restricted to those that are most closely related to budget support and
therefore this has precluded examination of all institutions, excluding for instance the legal and judicial
systems. The new accountability framework could be used to explore how these institutions function.

The research avoided making an assessment of the successes and failures of accountability, for
example when accountability fails or when reforms have not succeeded. This new perspective of
accountability could be further explored in its application to an assessment of the effectiveness of the
two types of accountability, considering accountability weaknesses, examining the reasons for failure
and how it could be prevented.

Finally, and perhaps most importantly, if this research is to have implications beyond Tanzania, is an
investigation to see whether the new theory of accountability can be applied to other countries. It would
be interesting to investigate what are the specific characteristics that determine the applicability of such
a framework and to identify the boundaries of its explanatory power. Interesting case studies could be
Ghana, by virtue of its highly competitive multi-party politics; and Rwanda by virtue of its stable elite,
centralised rent seeking and the implications of this for its longer term vision whereby clientelistic
politics may be compatible with developmental outcomes.

37
BIBLIOGRAPHY
Acemoglu, D., D. Ticchi, et al. (2007). Emergence and Persistence of Inefficient States.

Bayart, J.-F. (1986). Civil Society in Africa. Political Domination in Africa. P. Chabal.
Cambridge, Cambridge University Press: 109-125.

Berg-Schlosser, D. and R. Siegler (1990). Political stability and development : a comparative

analysis of Kenya, Tanzania and Uganda. London, Rienner.

Bigsten, A. (2001). Tanzania. Aid and reform in Africa : lessons from ten case studies. S.
Devarajan, D. R. Dollar and T. Holmgren. Washington D.C., World Bank.

Bird, R., J. Martinez-Vazquez, et al. (2004). Societal Institutions and Tax Effort in Developing
Countries. International Studies Program Working Paper Series,
International Studies Program, Andrew Young School of Policy Studies, Georgia State
University.

Booth, D., A. Lawson, et al. (2004). Joint Evaluation of General Budget Support Tanzania
1995-2004: Phase 2 Report - preliminary assessment of efficiency & effectiveness of budget
support and recommendations for improvements Dar es Salaam, Daima Associates Limited
Overseas Development Institute.

Bovens, M. (1998). The Quest for Responsibility: Accountability and Citizenship in Complex
Organisations. Cambridge, Cambridge University Press.

Castellani, F. and X. Debrun (2005). Central Bank Independence and the Design of Fiscal
Institutions. IMF Working Paper. Washington DC, IMF. 205.

Chabal, P. and J.-P. Daloz (1999). Africa Works: Disorder as a Political Instrument. Oxford,
James Currey.

Chandhoke, N. (2003). "Governance and the Pluralisation of the State: Implications for
Democratic Citizenship." Economic and Political Weekly: 2957-68.

Collier, P. (1999). Learning from Failure: the International Financial Institutions as Agencies
of Restraint in Africa. The Self-Restraining State: Power and Accountability in New
Democracies A. Schedler, L. Diamond and M. F. Plattner. Boulder, CO, Lynne Rienner
Publishers: 313-330.

Cornwall, A. (2007). "Buzzwords and fuzzwords: deconstructing development discourse."

Development in Practice 17(4-5): 471-484.

Costello, M. J. (1996). "Administration Triumphs over Politics: The Transformation of the

Tanzanian State." African Studies Review 39(1): 123-148.

Dogan, M. (2003). "Introduction: Diversity of Elite Configurations and Clusters of Power."

Comparative Sociology 2(1): 1-15.

Douglas, M. (1986). How Institutions Think. London, Routledge.

Dwivedi, O. P. and J. G. Jabbra (1988). Public Service Responsibility and Accountability.

Public service accountability: A comparative perspective. J. G. Jabbra and O. P. Dwivedi.
West Hartford, CT, Kumarian Press: 1-16.

38
Ebrahim, A. (2009). "Placing the Normative Logics of Accountability in "Thick" Perspective."
American Behavioral Scientist 52(6): 885-904.

Eurodad. (2008). "Eurodad analysis of the outcome of Accra." Retrieved 17 October 2008,
from http://www.eurodad.org/whatsnew/articles.aspx?id=2822.

Evans, P. B., W. F. Lam, et al. (1996). "Government Action, Social Capital and
Development: Creating Synergy across the Public-Private Divide." World Development
24(6): 1119-132.

Eyben, R. (2007). Power, Mutual Accountability and Responsibility in the Practice of

International Aid: A Relational Approach. Working Paper. Brighton, Institute of Development
Studies.

Fleck, L. (1979). Genesis and Development of a Scientific Fact. Chicago, University of

Chicago Press.

Fox, J. (2007). "The uncertain relationship between transparency and accountability."

Development in Practice 17(4-5): 663-671.

General Budget Support Donors (2007). Rapid Budget Analysis. Dar es Salaam, GBS
donors.

Gore, C. (2000). "The Rise and Fall of the Washington Consensus as a Paradigm for
Developing Countries." World Development 28(5): 789-804.

Gould, J. (2005). Timing, Scale and Style: capacity as governmentality in Tanzania. The Aid
Effect: Giving and Governing in International Development. D. Mosse and D. Lewis. London,
Pluto Press.

Government of Tanzania (2008). Speech by the Minister for Finance and Economic Affairs
Honourable Mustafa Haidi Mkulo (MP), Introducing to the National Assembly, the Estimates
of Government Revenue and Expenditure for the Financial Year 2008/09 on 12th June,
2008. Dodoma, Ministry of Finance and Economic Affairs: 50.

Gramsci (1971). Selections from the Prison Notebooks of Antonio Gramsci. London,
Lawrence and Wishart.

Grant, R. W. and R. O. Keohane (2005). "Accountability and Abuses of Power in World

Politics." The American Political Science Review 99(1): 29-43.

Gulrajani, N. (2010). "Challenging Global Accountability: The Intersection of Contracts and

Culture in the World Bank." GEG Working Paper (56).

Gupta, A. (1995). "Blurred Boundaries: The Discourse of Corruption, the Culture of Politics,
and the Imagined State." American Ethnologist 22(2): 375-402.

Harrison, G. (2001). "Post-Conditionality Politics and Administrative Reform: Reflections on

the Cases of Uganda and Tanzania." Development & Change 32(4): 657-679.

Harrison, G. and S. Mulley (2007). Tanzania: A Genuine Case of Recipient Leadership in the
Aid System? Working Papers. Oxford, The Global Economic Governance Programme.

Hawksley, H. and Mikwambe. (2010). "UK seeks China aid partnership in Africa." Retrieved
6/10/10, from http://www.bbc.co.uk/news/world-africa-11444441.

39
Helleiner, G. (2002). "Local Ownership and Donor Performance Monitoring: New Aid
Relationships in Tanzania?" Journal of Human Development 3(2): 251-261.

Helleiner, G., T. Killick, et al. (2002). Development Cooperation Issues Between Tanzania
and its Aid Donors. NEPAD at Country Level - Changing Aid Relationships in Tanzania. S.
Wangwe. Dar es Salaam, Mkuki na Nyota.

Hyden, G. (1980). Beyond Ujamaa in Tanzania : underdevelopment and an uncaptured

peasantry. London Heinemann Educational.

Hyden, G. (1994). Party, state and civil society: control versus openness. Beyond capitalism
vs. socialism in Kenya and Tanzania. J. D. Barkan. London, Lynne Rienner.

Iliffe, J. (1971). Agricultural change in modern Tanganyika : an outline history. Nairobi East
African Publishing House.

IMF (2007). Article IV Consultation. IMF Country Report. Washington, D.C, IMF. 246.

Isaacman, A. (1996). Cotton is the Mother of Poverty: Peasants, Work and Rural Struggle in
Colonial Mozambique, 1938-1961. Portsmouth, Heinemann.

Kalebe-Nyamongo, C. (2009). Self-interest vs. Altruism: Malawian elites and poverty

reduction. DSA Annual Conference 2009: Contemporary Crises and New Opportunities.
Ulster, DSA.

Kanbur, R. (2000). Aid, Conditionality and Debt in Africa. Foreign Aid and Development:
Lessons Learnt and Directions for the Future. F. Tarp, Routledge.

Kelsall, T. (2002). "Shop windows and smoke-filled rooms: governance and the re-
politicisation of Tanzania." Journal of Modern African Studies 40(4): 597-619.

Kelsall, T. (2004). Contentious Politics, Local Governance and the Self. Goteborg, Nordiska
Afrikainstitutet. 129.

Kelsall, T. (2008). Going with the grain in African development? Africa Power and Politics
Programme. London, ODI.

Kelsall, T., D. Booth, et al. (2010). Developmental patrimonialism? Questioning the

orthodoxy on political governance and economic progress in Africa. Africa Power and
Politics Programme. London, ODI. 9.

Khan, M. H. (2004). Corruption, Governance and Economic Development. The New

Development Economics. Jomo K.S and F. B. New Delhi and London, Tulika Press and Zed
Press.

Lachman, R. (2000). Capitalists in Spite of Themselves: Elite Conflict and Economic

Transitions in Early Modern Europe. Oxford, Oxford University Press.

Lachmann, R. (1990). "Class Formation without Class Struggle: An Elite Conflict Theory of
the Transition to Capitalism." American Sociological Review 55(3): 398-414.

Leftwich, A. (2010). "Beyond Institutions. Rethinking the role of leaders, elites and coalitions
in the institutional formation of developmental states and strategies " Forum for Development
Studies 37(1).

40
Lindemann, S. and J. Putzel. (2008). "State Resilience in Tanzania." Draft. Retrieved 1
September 2009, from
http://www.crisisstates.com/download/seminars/Putzel.Lindemann.Tanzania.Apr30.pdf.

Lukes, S. (2005). Power: A radical view. London, Palgrave Macmillan.

Migdal, J. S. (1988). Strong societies and weak states : state-society relations and state
capabilities in the Third World Chichester Princeton University Press.

Migdal, J. S. (2001). State in society: studying how states and societies transform and
constitute one another Cambridge Cambridge University Press.

Mills, C. W. (1956). The Power Elite. New York, Oxford University Press.

North, D., J. Wallis, et al. (2007). Limited Access Orders in the Developing World. Policy
Research Working Paper. Washington D.C., World Bank.

O'Donnell, G. (1999). Horizontal Accountability in New Democracies. The self-restraining

state : power and accountability in new democracies A. Schedler, L. Diamond and M. F.
Plattner. London, Lynne Rienner Publishers, Inc: 29-51.

Rakner, L. (2005). External accountability and PFM: Reflections from recent studies in
Malawi, Uganda and Tanzania. CAPE workshop on Aid, Budgets and Accountability.
London.

Reis, E. and M. Moore (2005). Elites, Perceptions and Poverties. Elite Perceptions of
Poverty and Inequality. E. Reis and M. Moore. London, New York, Zed Books: 1-25.

Reno, W. (2000). "Clandestine Economies, Violence and States in Africa." Journal of

International Affairs 53(2): 433.

Richey, L. (1999). "Family Planning and the Politics of Population in Tanzania: International
to Local Discourse." The Journal of Modern African Studies 37(3): 457-487.

Schedler, A. (1999). Conceptualizing Accountability. The self-restraining state : power and

accountability in new democracies A. Schedler, L. Diamond and M. F. Plattner. London,
Lynne Rienner Publishers, Inc.: 13-28.

Schmitter, P. (1999). Comments on O'Donnell: The Limits of Horizontal Accountability. The

Self Restraining State: Power and Politics in New Democracies. A. Schedler, L. Diamond
and M. F. Plattner. London, Lynne Rienner Publishers Inc.: 59-62.

Shivji, I. G. (2006). Let the people speak: Tanzania down the road to neo-liberalism. Dakar.

Simon, H. A. (1962). "The Architecture of Complexity." Proceedings of the American

Philosophical Society 106(6): 467-482.

United Republic of Tanzania (2001). Memorandum of Understanding on Performance

Assessment Framework: Poverty Reduction Budget Support Monitoring Framework 2001-
2004. Dar es Salaam, Ministry of Finance.

Unsworth, S. (2010). An upside down view of governance. Brighton, Institute of

Development Studies.

41
Weisband, E. and A. Ebrahim (2007). Introduction: forging global accountabilities. Global
Accountabilities: Participation, Pluralism, and Public Ethics. A. Ebrahim and E. Weisband.
Cambridge, Cambridge University Press: 1-24.

White, H. and O. Morrissey (1997). "Conditionality When Donor and Recipient Preferences
Vary." Journal of International Development 9(4): 497-505.

Williamson, O. E. (2000). "The New Institutional Economics: Taking Stock, Looking Ahead."
Journal of Economic Literature 38(3): 595-613.

World Bank (2007). Implementation Completion and Results Report on a Credit to the United
Republic of Tanzania for a Tax Administration Project, World Bank: 49.

42