Dialogue as a tool for improving policy implementation: Lessons from Ghana

Implementation research has often tended to focus on what goes wrong or, less frequently, on why success is achieved, rather than identifying tools or instruments to address problems encountered. Yet, identifying tools to avoid typical implementation problems is crucial to successful implementation. Focusing on one such tool, ‘policy dialogue’, and employing a retrospective examination of the implementation process associated with Ghana’s National Health Insurance Scheme (NHIS), we examine the potential that existed to mitigate the risks associated with policy stakeholders’ behaviour during implementation. Policy actor positions in implementation are based on differing values, ideologies, perceptions and power bases, and frequently cause degrees of discordance, as demonstrated in the case of Ghana’s NHIS. Policy dialogue may promote a greater awareness of other actors’ positions and motivations, creating a circumstance in which evidence takes on more significance than it would in more political environments. The study draws extensively from earlier qualitative fieldwork in Ghana, and has been supplemented by desk-based research. Selected key events in the implementation of the NHIS are used to illustrate where and why policy dialogue could have been used to avoid the creation of chokepoints.


Introduction
The article examines a set of events and associated relationships arising between a range of policy actors in a bid to identify circumstances in which a particular device, 'policy dialogue', could be deemed to have had strong potential to avoid chokepoints in the implementation of the National Health Insurance Scheme (NHIS) in Ghana.Chokepoints, in this study, are moments in the implementation process where the programme could not be steered forward due to discordant relationships developing, where stakeholders refused or were unable to adjust their own positions in a manner that avoided disrupting progress with the programme.
With reference to the authors' previous research and knowledge of the history of Ghana's NHIS, this discussion article seeks to further develop the concept of policy dialogue in real-world circumstances.It is frequently observed that policy researchers have been far more successful in theorising and collecting empirical data relating to decisions than in gaining knowledge of resultant implementation processes (Ayee, 1994;Kingdon, 2011).Policy implementation failure, when there are large gaps between objectives and what is actually achieved, or indeed non-implementation, when programmes never reach launch point, cause financial and political failure, and disrupt the lives of people dependent on positive outcomes.While the literature on failure is considerable, there is a gap in terms of practical accounts of how to make implementation effective, particularly in the distinctive context of low-income African countries.In attempting to develop understandings of how policy dialogue can support an effective policy process, this article is unusual in avoiding an examination of a case in which the tool was employed in compliance with the terms of a funding package.Instead of focusing on participation and the results of an actual policy dialogue, the present authors examine a complex implementation case for evidence of where the tool could have made a positive impact on progress.
Five key events are used to illustrate circumstances in which policy dialogue could have made a significant difference in steering the NHIS programme through what became implementation chokepoints.A long view has been taken of what constitutes implementation.In essence, the discussion examines the processes needed to bring the NHIS to operational status after the 2000 election of a government committed to abolishing the existing health financing scheme in favour of an insurance-based system.Other researchers may label some of the five events 'preimplementation'.In examining the potential utility of policy dialogue, a decision was taken to capture the impact of discordant relationships between stakeholders on the eventual roll-out and operation of the NHIS.
In establishing where policy dialogue may have utility as a tool, reference can be made to Hogwood and Gunn's (1984) classic framing of the implementation problem in ten conditions for 'perfect implementation'.Examination of the list helps to identify where policy dialogue has promise.It can be readily seen that this is not a tool that can be expected to address seven of the typical condition-deficient impediments encountered: crippling circumstances external to the organisation; inadequate time and resources; an absence of the required combination of resources; complex, hard-to-negotiate intervening links; indirect relationships between policy action and outcome; problematic dependency relationships; and less than perfect compliance with those in authority (Hogwood and Gunn, 1984).However, the remaining three conditions can arguably be engaged through an effective policy dialogue.Policy dialogue has particular potential with respect to achieving stakeholder understanding of an agreement on objectives, and working out how implementation tasks are specified and planned in the correct sequence.Hogwood and Gunn (1984) also identified perfect communication and co-ordination as a high-priority primary aspiration for policy dialogue.
Policy dialogue provides a framework for improving mutual understanding, identifying priorities, enhancing ownership and participation, finding common ground, building constituencies and resolve for change, and influencing policy (US Aid, 2014).
A wide range of governmental and civil society actors involved in public health and health systems development have increasingly become interested in policy dialogue as a tool for avoidance and/or management of policy discord.Such actors include country-level implementers and decisionmakers at all levels.The World Health Organization (WHO) and other United Nations (UN) agencies, and also funders are prominent advocates (Rajan et al., 2015); Kirigia, et al., 2016; European observatory on Health Systems and Policies, 2019).Yet, limited evidence exists on how policy dialogue has evolved as a concept and tool used for decision-making.This is true, especially, in Africa (including Ghana) and in low-and middle-income countries elsewhere (Dovlo, Nabyonga-Orem, Estrelli, & Mwisongo, 2016).
In democracies, the policy-making process is conventionally shaped by elected officials, administrative agencies and organised interest groups.This is often seen as insufficiently inclusive of communities comprising marginalised lower-income citizens.Their participation, it is argued, needs to be actively secured by public servants and policy makers in order that relevant priorities and rights are asserted in public policy making (Goodin and Dryzek, 2006).Participatory policy making, it is claimed, offers such marginalised communities a platform to self-organise and form alliances with broader interests, to create fairer and better-crafted public policies (Hajer and Wagenaar, 2006).While there are links between the concept and practice of participatory policy making and policy dialogue, the latter, it is argued, should be conceived as primarily a tool that can help deliver supportive inputs to specific policy programmes, rather than a broader democracy-enhancing movement.Arguably, policy dialogues -because they work on the basis of the selective participation of stakeholders -are quite different in terms of what they mean for the broader democratic process.Policy dialogues are not platforms for activists.If policy dialogue is a threat to power structures, then this consists of a rebalancing away from the political way of doing business towards knowledge-based resolutions of problems (Mwisongo, et al., 2016).With an impact focus, policy dialogues cannot allow implementation to become either slower or less wellinformed by evidence, in the way that proponents of participatory decision-making might concede if this was the cost of greater inclusion of marginalised people.Arguably, policy dialogues also need at some level to be government-sanctioned if they are to be effective.While participatory decision-making is an alternative to official politics, or at least a significant adjunct, policy dialogues do not amount to a reinvention of politics, and are not created to build new political communities.

Case context
In assessing when policy dialogue should be considered potentially useful as a chokepoint breaker, it is necessary to consider the context of policy implementation.In this respect, the NHIS implementation history provides rich material from which to learn.Ghana was the first sub-Saharan African country to introduce, through a 2003 Act of Parliament (ACT 650, Amended Act 852), a nationwide NHIS covering both the formal and informal sectors (GoG, 2003(GoG, , 2012)).The NHIS replaced a discredited 'cash and carry system', 1 although the progress of the new programme, inevitably or otherwise, would be marked out with complexity.The basic problems are readily identifiable.The NHIS had to provide financial protection for both the formal and informal sectors through a combination of taxes and annual premium payments (Abiiro & McIntyre, 2013).This implies a scheme with many parts and tasks, implying an implementation process with many potential chokepoints, at which misunderstanding and competing perspectives could occur.
The scheme assures access to basic clinical services for all Ghanaians, regardless of ability to pay (Agyepong & Adjei, 2008).Financing of the scheme had to draw on several sources.It currently consists of: earmarked tax (2.5% value added tax -hereafter referred to as VAT) making up about 70% of the total revenue; 2.5% of the 18.5% of formal sector employees' contribution to a pension fund at the Social Security and National Insurance Trust (SSNIT), which is approximately 18% of the total funding; and premium or annual contributions from informal sector and formal sector employees who do not contribute to SSNIT, which is about 4% of the total funding.Smaller funding sources include donations, gifts, investments, grants, other voluntary contributions, and fees charged by the authority in the performance of its functions, and these sources account for about 8% of the total funding (Alhassan, Nketiah-Amponsah, & Arhinful, 2016;Fusheini, 2016;GoG, 2012).Thus, the Ghana scheme is a combination of a Beveridge tax-funded system and Bismarckian insurance and mutual health organisation (MHO) models.This made for considerable implementation complexity.
Membership is open to all residents of Ghana upon subscription.In theory, and by law, enrolment entitles members to a comprehensive benefits package covering over 95% of both inpatient and outpatient services involving all common illnesses in Ghana.Practically, however, defining what constitutes 'common illnesses' covered by the scheme is problematic; for instance, subscribers often find out that, in most cases, they still have to pay for drugs and only laboratory investigations are covered.The NHIS law requires Ghanaian residents to enrol in one of three insurance schemes: national, private mutual or private commercial.Currently, however, less than 1% of the country's population has private health insurance (Myjoyonline, 2019).The existence of the three schemes could be described as a compromise of some sort during the development of the policy.It also underscores the plurality of the Ghanaian health system, with both the public and private sectors complementing each other, which affords clients a choice in service providers.
Although it is mandatory to enrol in one of the three schemes, this has not resulted in full coverage by the NHIS, because of the presence of a large informal sector and a still somewhat low administrative capability on the part of the National Health Insurance Authority (NHIA) in Ghana (Alhassan et al., 2015;Alhassan et al., 2016).During the policy-making and implementation process, the informal sector was poorly represented.This was because representation was dominated by the community-based health insurance movements, the donor community and the Ministry of Health.This restricted the chances of the informal sector making inputs into designing the system and operational processes, especially during the initial stages of policy implementation (Fusheini, 2013).

Policy dialogue
The concept of policy dialogue is defined to mean: an event, where dialogue takes place around 'a policy question … on which … key documents and international experts … [are brought together] to present recent evidence, as well as relevant case studies from countries that have faced a similar question (EOHSP; Rajan et al., 2015).

Policy dialogue has a distinctive role:
Policy dialogues describe a particular brand of bringing evidence to practice: highly focused, targeting senior policymakers and their top advisers, and marshalling support for key decision points.They aim to offer policymakers in a country or a group of countries a neutral platform to discuss a particular key policy issue on the basis of comparative evidence and sharing experience (European Observatory on Health Systems and Policies, 2019).
Formalised, or state-promoted, policy dialogues should involve policy makers, advocates, other non-governmental stakeholders, other politicians and beneficiaries (Hardee, Feranil, Boezwinkle, & Clark, 2004;USAID, 2014).Policy dialogue enriches policy-and decision-making processes through rounds of evidence-based discussions, workshops and consultations on a particular subject.
A key characteristic of policy dialogue is the involvement of people from different interest groups sitting together to focus on an issue in which they have a mutual interest.It assumes that people in different positions and circumstances will have different perspectives on the same problem, and that they may have access to different information and ideas about the issue (WIEGO, 2013).The best approach for enhancing dialogue, it is argued, is one in which it is closed (open only to those who are invited), where the group is relatively small and where there is little emphasis on formal protocol.It has been observed that a policy dialogue acquires value only by the manner in which it is conducted and formally organised (Dovlo et al., 2016;Rajan et al., 2015;SURE, 2011;Westermann, Verheij, Winkens, Verhulst, & Van Oort, 2013).A policy dialogue, therefore, needs to be well-structured, so that all parties have a chance to contribute by focusing on a limited selection of issues that can realistically be addressed (WIEGO, 2013).
If successful, a policy dialogue will have chosen and secured the active participation of key government and civil society actors who relate to programme-relevant constituencies.Selected participants should be capable of engaging in effective advocacy on behalf of their constituencies, and be inclined to present evidence-informed arguments, while in turn gaining an enhanced knowledge of the policy process.In retrospect, a policy dialogue used at various points in the implementation of the NHIS would have involved all the following stakeholders: the World Bank; the WHO; other international actors in the development community; the GHS; private medical and dental providers; the TUC; the Ghana National Association of Teachers; the Ghana Registered Nurses Association; community-based health insurance associations; the Christian Health Association of Ghana; civil society organisations; researchers; health NGOs; users of services and many more.Indeed, while a majority of the above were part of the implementation process, some of them became, at certain points, either marginalised or ignored, due to different interests and preferences (Fusheini, 2013).
While the 'perfect' policy dialogue should be treated as an ideal or aspiration rather than a commonly observable process, effective policy dialogue is thought to be contingent on the extent to which participants can contribute certain behavioural skills, including strong communication, negotiation, problem-solving and conflict resolution skills (McCullough, 2011;USAID, 2014).
The level of trust generated is also significant, as actors need to enter into open, inclusive and informed discourse.Consensus building on the objectives, purpose and process prior to the dialogue beginning is vital.There is also a need to facilitate easy access to dialogue-relevant information, with impact thought to be dependent on the extent to which relevant evidence from case studies, pilot or demonstration projects is employed (Bowen & Zwi, 2005).Also crucial to the process is appropriate timing of dialogue activities, to ensure that feedback makes a difference to decision-making and implementation.
Moreover, policy dialogue can be seen as a device that restructures decision-making processes.
One way of conceptualising what goes on is to adapt, for the purpose of understanding the policy process, a distinction made by psychologists between information processing that takes place either serially or in parallel streams (Chan & Lam, 2018).Serial processing means that policy decisions are analysed and broken down into sequential parts, where problems can be addressed one by one in a customised way.In parallel processing, by marked contrast, problems are addressed together.
Relationships between problems are recognised and subjected to thinking that can draw on solutions that are dependent on a wider set of beliefs and knowledge than would be drawn on in serial decision processes.This is dependent, however, on including a broad collection of stakeholders in the decision process, and indicates why a tool conceived along policy dialogue terms should be considered.Governments and large-scale organisations tend to default to serial decision-making in certain circumstances, while parallel decision-making is encouraged by policy dialogue.Examples include bringing together stakeholders with domain-specific expertise rather than excluding them until a point in the serial decision process needs their input.Reducing the occurrence of stand-off disagreement by keeping policy actors in a continuous rather than episodic process is another benefit.

Methods
Through examining data generated by an exploratory qualitative case study, this article has drawn on a research-generated, quasi-insider knowledge, of the implementation of the NHIS in Ghana in 2011 and 2012.Qualitative interview-based research best served our interest in the inclusiveness and/or exclusiveness of actors, their respective value positions, power bases, preferences and perceptions, and how these produced consensus and divergent points during the policy engagement.
The study was conducted across four districts (two in the Northern Region and two in the Greater Accra region) using a multi-level approach -with national, regional and district respondents.The respondents comprised various stakeholders: politicians and/or policy elites; district, regional and national health insurance officials and managers; development/donor partners; technocrats; civil society organisation (CSO) members; Trades Union Congress (TUC); professional associations; and medical personnel (service providers) from across the two regions.
A combination of purposive, expert and chain-referral sampling techniques (Biernacki & Waldorf, 1981;Bryman, 2015;Fischer & Strandberg-Larsen, 2016;Luborsky & Rubinstein, 1995) was used following a review of policy documents and stakeholder mapping/analysis.A quasi-insider position was built up through a 'shoe leather'-dependent interviewing fieldwork programme, made possible by judicious use of subject selection through what is sometimes referred to as 'snowball sampling'.The implementation arena for the NHIS was a logistically difficult area to cover.Even Accra itself has transport challenges, due to traffic jams and a poor road network in most parts of the city.A quasi-insider knowledge arguably offers a more rounded appreciation of 'what went on' than that of an actual insider.Emails and phone calls to set up interviews did not yield any results.
Personal visits to the offices and workplaces of the respondents were the only option, and in some cases, it took more than five visits before interviews could be arranged.Travelling between the north and Accra and visiting some of the sites several times helped in building trust, as did using acquaintances such as secretaries and reception desk managers.In some instances, existing oldschool networks and requests to colleagues helped in setting up interviews.This was a particularly useful tool at Parliament House in Accra.
Thirty-three in-depth interviews were conducted in the four case areas over a period of eight months.The interviews were conducted in the English language.The empirical qualitative study supplemented by desk-based research has allowed the cross-referencing of differing accounts emanating from a full range of stakeholders in this discussion paper.The Research Ethics Committee at Ulster University in the UK approved the protocol for the study as well as the Ghana Health Service (GHA).Permissions from the NHIA to visit the selected district offices of the NHIS were secured.All participants in the study provided written informed consent before taking part.

When and why policy dialogue could have helped implement the NHIS
The table below is a summary of key events in the implementation process at which policy dialogue could have facilitated forward movement, avoiding chokepoints.As explained earlier, a long view has been taken of what constitutes implementation in an effort to consider the potential utility of building the programme on the basis of an ongoing policy dialogue between stakeholders.

Key Event
Chokepoint Why could policy dialogue have helped?

Composition of the task force
Different set of values and interests (actor-centred).
Political actors hijacked the policy process.
Technical members marginalised and forced to step down/resign.
By encouraging adherence to principles of mutual respect, accessibility, clarity and transparency (USAID, 2014).

By
providing space for participating policy actors to contribute knowledge, expertise, views and ideas in a setting in which they can anticipate being listened to and appraised by other stakeholders (USAID, 2014).
By giving all parties a chance to contribute on a limited set of issues (WIEGO, 2013).
By reaching agreements or consensus on policy solutions.

Design of the scheme (structure)
Single-payer centralised systems versus multi-payer decentralised system.
Donor partners and actors perceived to oppose actors who are marginalised or ignored by government.
By bringing together policy actors from different positions and circumstances.These actors will have different perspectives on the same problem, with the likelihood of access to different information and ideas about the issue (WIEGO, 2013).
Forceful incorporation of existing community-based schemes into the national system.
By building consensus around objectives, purpose and process of participation.
By identifying groups thought likely to be affected by the policy.
By identifying those stakeholders who can make a significant contribution to the dialogue.

Passage of the bill
Perceived rush in passing the bill under a Certificate of Urgency.

Protests
against specific components of the bill.
Call for broader consultation and consensus building from all stakeholders.
By providing a platform for stakeholders to identify issues, share perspectives, and explore how to achieve common ground and degrees of consensus (Hardee et al., 2004).
By requiring parties involved to have strong communication, negotiation, problem-solving and conflict resolution skills (McCullough, 2011;USAID, 2014).
By involving different interest groups with a mutual but not necessarily common interest.

Funding the scheme
The specific source of fundingrequiring 2.5% of workers to contribute to a pension fund by law without proper and broader consultation.
Contribution/premium payment determination.
By being well-structured so that all parties have a chance to contribute.
By providing easy information access relevant to the dialogue.
Programme roll-out approach 'Big bang' approach versus gradual or staged implementation.

Implementation committees dominated by political associates.
Because policy dialogue is closely associated with the objective of informing implementation with evidence (Bowen & Zwi, 2005).
By selecting those government and civil society actors capable of representing stakeholders through offering evidenceinformed arguments that draw on knowledge of the policy programme.
During the introduction of the NHIS, it was clear there were instances of conflictual behaviour that stalled the smooth implementation of the scheme.We discuss below, with evidence from an extensive fieldwork programme in Ghana, the key events and how policy dialogue could have helped in resolving some of the chokepoints in the implementation process.

Task force composition
The first major chokepoint arose around the composition of the health insurance development task  (Agyepong & Adjei, 2008).The composition of the task force membership implied different sets of values, interests, principles, experience and knowledge, with a consequently high potential for discord.In the absence of an effective resolution mechanism, the differing interests and values of task force members quickly set them against each other, particularly, government representatives against the other actors.This led to the marginalisation of most of the original members and their subsequent resignation (Agyepong & Adjei, 2008).This was confirmed by a CSO participant during the extensive fieldwork: When government saw that we -those who were already on the ground -were not toeing their line, they constituted their own technical team, which excluded all the existing so-called experts in that area then [CSO respondent 1].
By the time of the first draft of the NHIS law, only one member of the original membership remained (Agyepong & Adjei, 2008), as others had either been forced out by government actors or had resigned on principle.According to a representative of organised labour, the politicians assumed that because most task force actors were performing as institutional representatives, 'they were feeding the team with concerns of the various institutions so represented' (TUC respondent 2).Therefore, the political actors ignored inputs considered to be against those of the interests of the ruling political party.This is consistent with Agyepong and Adjei's (2008) observation that trusted political associates of the government narrowed out inconvenient comments, suggestions and criticisms.The policy dialogue principle of consensus or agreement on policy solutions, as well as mutual respect and space for all stakeholders to make their contributions, was missing in the policy-making process.

Drafting of the bill and design structure
If the composition of the task force to draft the health insurance bill set the stage for conflictual and discordant behaviour among the key actors, the design structure of the NHIS was yet another battleground between government actors on the one hand and the other stakeholders on the other.
The debate was about a centralised single-payer system versus decentralised multi-payer autonomous community-based schemes.Key stakeholders, including donor partners -especially, the World Bank, the Danish International Development Agency (DANIDA) and existing community-based schemes -all favoured decentralised multi-payer systems, as evidenced in this interview quote: Most of the donors were quite hostile to the process because they did not want it to go the way it was going.They wanted to maintain community-level activity to the extent that the government decided it was not going to listen or to engage the donors at all, and that they were going to go their own way [World Bank respondent 2].
Government actors, led by the then health minister, however, strongly favoured a centralised single-payer system.The discord was so intense that the government decided not to engage the other stakeholders in any further discussions (Fusheini, 2013), and to choose its own policy path.
The resistance to effectively incorporate the workings of government and civil society actors, who represented relevant constituencies, was a missed opportunity in the policy-making and subsequent implementation process.Formal policy dialogue, incorporating evidence-informed arguments, between all concerned could have proved useful in the policy process (Hardee et al., 2004;USAID, 2014).Consequently, government political actors dominated and easily hijacked the process in favour of pursuing adversarial tactics (Carbone, 2011) at the expense of evidence, leading the programme into a chokepoint at which the cooperation of stakeholders, particularly donors, was withdrawn.

Passage of the bill
The discordance that had been contained within the closed policy cycle, away from the public, emerged during the passage of the bill.Given the lack of stakeholder dialogue over the development of a design for the NHIS programme, a lengthy confrontation ensued.This stage was characterised by agitations and protests by organised labour and other civil society organisations against some aspects of the content of the bill, particularly, decisions regarding the design structure of the scheme (Agyepong & Adjei, 2008;Fusheini, 2013), as discussed earlier.A cross-section of stakeholders and interests registered their protests.Trades unions, in particular, opposed the specific policy directive on the source of funding, where 2.5% of workers' contribution to a pension fund at the SSNIT was to be channelled into the National Health Insurance Fund (NHIF) without consultation and the consent of workers.This led to 'an open confrontation where workers went on the streets, ostensibly to demonstrate against that thing' (TUC, Respondent 3).
Another key disagreement on the passage of the bill centred on the speed with which it went through parliament, being perceived to have been rushed through.This was highlighted when one of the interviewees asserted: The politicians wanted to do it within three months.From the day the bill was passed to the time it was supposed to be implemented was like three months … They [the NPP] still forced the agenda through, saying that there was no time for discussion [CSO respondent 1].
Stakeholders sought further consultations and consensus building with regard to the content of the bill.This view, from a TUC respondent, is particularly pertinent: Our major position on the whole process was that major, well-intended projects and policies fail because not much engagement was allowed, not much consultation was allowed, for the people who were beneficiaries to understand the issues well.So, eventually, the law was passed, even though we protested [TUC respondent 3].
The ruling NPP viewed these attempts, especially, by the opposition National Democratic Congress (NDC), as a strategy to sabotage and/or delay fulfilling its campaign promise to replace the cash and carry system before the next election in 2004.A power struggle ensued between the politicians in the NPP government on the one hand and the technocrats/experts in the Ministry of Health, GHS, civil society organisations (especially, existing community-based mutual health insurance schemes), the NDC and development partners on the other.This came to light during the fieldwork, as some interviewees noted: It was GHS, donor group and workers versus the politicians, who wanted to do it within three months from the day the bill was passed to the time it was supposed to be implemented.It was the debates that forced it and stretched it to even eight months or nine months … The opposition party, which was [the] NDC, took advantage of that and also said, 'Okay, [we] won't even vote, [we] don't support [it]'.They walked out of parliament the day this bill was passed (CSO respondent 1).
Those of us who were kind of technocrats, who were working and had an idea about the nitty-gritty, we actually wanted to ensure that, at the end of the day, we have an insurance scheme that would be very effective and efficient … Politicians also had their own interest because they wanted to meet their political promises (donor partner, respondent 4).
A former health minister gave credence to some of the above views by noting that: The officers or directors in the Ministry of Health (MoH), who were to help us do it [health insurance], came to me and said, 'Sir, it is not possible'.I was the minister and I wanted to get this thing done because my government had promised that they were going to remove cash and carry from the system.So I was not going to leave it to anybody to determine the fate of my government (political actor, respondent 5).
The consequence was the design flaws of the original Act and the need for a subsequent amendment in 2012, as the rush to pass the law had meant that it did not adequately reflect the views of a majority of the stakeholders.As one respondent noted: Today, that is why we are in the process of amending the law again; because if you promulgate a law in 2003, and already by the time it was 2010 you are contemplating of restructuring it, it clearly shows that there are challenges (political actor, respondent 6).
Inevitably, progress was delayed with no policy dialogue or other mechanism to prevent these problems arising.By facilitating discussions amongst stakeholders issues would have been brought into the open, where a prospect existed for different perspectives to be reconciled on common ground.

Programme roll-out
In the absence of a dialogue process for resolving differences, the adopted programme roll-out approach also generated discordance.Since the policy evolved out of a political campaign promise, the government actors favoured a kind of 'big bang' approach, with the goal of nationwide roll-out covering both the formal and informal sectors in a district-wide scheme, to the gradual or phasedin approach favoured by development partners and other stakeholders.
The structure gave us cause to think that perhaps it was quite an ambitious project because [it was] doing health insurance on a national scale, whereas we know that elsewhere [it] is done in sectors, [which] also got us a bit worried.And at a point in time, we were sceptical, because our concern was that if you rush through and you don't succeed, it would put us to a worse situation than even when the health insurance was not in place [TUC, respondent 3].
Ruling political party over-dominance was again observable in the implementation stage.
Government political actors sought to take control of the entire process by determining the composition of implementation committees across the country.
If you were setting up a committee, you will see that the ruling party wants to make its presence [felt] there.[By] making sure that if you are the chairman of a district implementation committee, they were sure that you belong to the party (GHS, respondent 7).
These attempts set the ruling political party class against the technical experts and other stakeholders in the implementation process.It was noted during the extensive fieldwork that informed this paper that although the policy development and implementation process was frequently characterised by mutual stakeholder support, as issues were discussed dispassionately in an open and frank manner, there was no effective mechanism to ensure this.Consequently, at certain points, the unresolved differences in positions led to tensions and breakdown in the roll-out process.

Conclusions
This article has examined a complex case, associated with the NHIS in Ghana, for points at which the employment of policy dialogue could have been beneficial in what became a long implementation process.From the discussion in this study, a number of potentials in utilising a policy dialogue in health policy making and implementation were identified, especially in the political and social context of sub-Saharan Africa.Firstly, policy dialogue can mitigate the risks associated with discordant behaviour on the part of policy stakeholders.Policy dialogue can also provide a platform for evidence building and implementation capacity building.Secondly, efficiency and good governance cannot be assumed.Policy dialogue should be based outside government, to insulate contributors from partisan political pressure.Furthermore, policy dialogue, through bringing witnesses together to exchange positions and views, may provide a more meaningful inquiry process than one confined to parliamentary institutions.It provides a type of transparency in countries in which there are major data gaps, and in which the ability to pursue meaningful scrutiny is consequently constrained.More importantly, policy dialogue can promote a more sophisticated approach to policy development, which draws on the insightful contextual awareness of other actors' positions and motivations.It also provides a mechanism for encouraging actors to be more self-aware about their positions and motivations.
It can make key stakeholders aware of their unconscious biases.
The policy process will typically, to varying degrees, involve alternative policy actor positions based on differing values, ideologies, perceptions and power bases, resulting in varying degrees of discordance.Policy dialogue should not be confused with a political process.Rather than an alternative to politics, in the case examined, policy dialogue could have offered remedies to a particular class of conflict that emerged during implementation.Evidence sharing and mutual comprehension of alternative perspectives, rather than the healing of ideological divisions, is where policy dialogue can be effective.
force set up by the Ministry of Health.Government political actors acted aggressively, especially in relation to the membership of the task force, which was inaugurated in March 2001, two months after the New Patriotic Party (NPP) assumed office.The membership comprised institutions and people with a wide range of expertise and technical knowledge from the Ministry of Health, GHS, Dangme West District Health Directorate and Research Centre and the TUC