Decentralization and Pro-poor Participation in Ghana: Unmasking the Barriers to Inclusive Grassroots Development

Donors have strongly advocated decentralization on the grounds that it broadens the participation of citizens in development processes, thereby increasing government responsiveness to their needs. Although there have been studies seeking to establish the veracity of this claim, they remain weak on two fronts. One, while wealth differentials affect citizen participation, these studies approach citizens as a homogenous group. Two, participation is mostly viewed narrowly—participation in elections. Drawing on empirical data from Ghana’s decentralization reform, this paper addresses these gaps by questioning how pro-poor citizen participation in decentralized development planning has been. I argue that although Ghana’s decentralization was propagated on championing pro-poor grassroots participation in governance and development, in reality, participation is elitist and has failed to reflect the voices of the poor. My findings demonstrate that the participation structures and processes used in local development planning are unfavourable to the poor in many respects. Elite and representative participation is promoted to the detriment of broad citizen participation. This exclusion is resulting in surging apathy of citizens towards the district assemblies—institutions hitherto trusted as their development champions. This paper concludes that the poor will remain voiceless despite decentralization unless the structural barriers to their participation are tackled in decentralization design and praxis.


Introduction
Decentralization has been broadly embraced by donors and governments of the developing world since the 1980s as a reform that could propel grassroots development. Citizens' participation in governance and development processes is said to improve under decentralized governance (World Bank 2001;Faguet 2004;Faguet and Shami 2022;Blair 2000;Debrah 2014;Bardhan 2002;Ribot 1999;Jütting et al. 2004;Manor 1999). This makes the provision of development services more responsive and reflective of citizens' needs within a decentralized governance system (Campell 2003;Crook and Manor 1998;Dufhues et al. 2015;Fatke 2016; Baral and Heinen 2007). However, some studies find no strong evidence of decentralization enhancing participation (Andrews and De Vries 2007;Francis and James 2003;Phago and Molosi-France 2018;Devas and Grant 2003); they assert that local residents are mostly not involved in participatory processes and that, even when they are included, their influence over decentralized decision-making processes is marginal and tokenistic. More importantly, studies on decentralization and participation over the years have largely overlooked one crucial issue regarding participation. That is, many of the assessments on decentralization and participation (such as Caparas and Agrawal 2016;Waheduzzaman and As-Saber 2015;Ishii 2017;Ryan 2012) have always framed local people as a homogenous group; hence, such assessments have often failed to ascertain whether decentralization has truly provided spaces for the poor to effectively participate. This presents a critical gap in the assessment of decentralized participation because involvement of citizens is often not differentiated among the various strata of citizens to consider the participation of the poor.
To address this gap, this paper uses Ghana's decentralization reform to question the participation claims of decentralization with a particular focus on ascertaining how pro-poor citizens' participation in local development planning has been. By assessing pro-poor participation in local development planning, I will be able to ascertain how pro-poor services delivered within decentralization framework are likely to be. This is because participation in many cases is not an end in itself but a means to ensuring that development interventions are delivered to reflect the needs of local citizens, particularly the poor and vulnerable groups (Jütting et al. 2004;Francis and James 2003). This paper contributes to the existing knowledge by reconciling and bridging the stance of the two strands of literatures, that is decentralization improves citizen participation, and decentralization does not improve citizen participation; by firmly pointing out that while decentralization may provide spaces for local citizens to participate in decision-making, the local citizens who are mostly involved are not the poor but the elite in society. This paper therefore contributes to a growing new body of knowledge by drawing attention to the structural barriers that normally work against the participation of the poor even under decentralized governance systems. This has the effect of diminishing the voices of the poor in local development decisions and making development interventions not pro-poor.
In this paper, I conceptualize decentralization following Crook and Manor (1998) and Oxhorn et al. (2004) as the transfer of power to different sub-national levels of government by the central government. The remainder of this paper proceeds in four parts. In "Decentralization and Participation: a Brief Overview of Theoretical and Empirical Literature", I present an overview of the theoretical and empirical literature on decentralization and participation. This is necessary to situate the discussion in the right context. I conclude the section with a brief discussion of the participation ethos of Ghana's decentralization. This lays the foundation for a richer understanding of the empirical section. In "Research Design and Methodology", I then outline the methodology employed in this inquiry. This is followed by "Results and Discussions: Decentralization and Citizen Participation in Practice", framed into three sub-sections, where I draw on the empirical data. The paper concludes by re-echoing how the poor have remained voiceless and powerless despite decentralization. I then point to ways the poor could regain their lost voices.

Decentralization and Participation: a Brief Overview of Theoretical and Empirical Literature
My main objective here is not to give a comprehensive account but to highlight the theoretical debates underpinning decentralization and participation and how this informs practice. Democratic participation has often been the strongest theoretical basis for decentralization (Stoyan and Niedzwiecki 2018;Hyden 2017). There is an enduring history of theoretical discussions by Alexis de Tocqueville, Stuart Mills, and Dahl among many others frontally linking decentralization to democracy (Ribot 2002;Matthews 2018;Houdret and Harnisch 2019). Some political scientists and governance theorists equate decentralization to democracy, due to the participation that decentralized governance engenders. Hart (1972, 606) for instance firmly argues that: "… 'decentralization' equals 'democracy.' Once organizations are decentralized, all of the positive benefits associated with democracy will supposedly come to pass…citizen groups would be able to participate directly with them in the making of policy decisions." Thus, decentralization makes direct participation of citizens in democratic governance feasible.
By decentralizing power and resources to subnational units of government, citizens are offered the opportunity to interact face-to-face with important policymakers to shape local decisions that directly affect them (Arkorful et al. 2021;Faguet and Shami 2022). Pateman (1970) and Elliott et al. (2018) assert that citizens are more concerned about decisions closer to them than remote decisions at the national level. This suggests that decentralization deepens democracy and makes it more meaningful by creating avenues for citizens to directly participate in decisions that are pertinent to them. This is consistent with Dahl's (1967) proposition that for direct participation of citizens to be effective and realistic, cities need to be smaller to ensure that people are able to express their views and feel part of the democratic governance process. This again makes decentralization an effective tool for democratic participation.
The result of the direct participation windows offered through decentralization is that the needs of the citizenry within each jurisdiction are known. This makes subnational governments more efficient in directly responding to the needs of citizens (Ribot et al. 2006;Ryan 2012;Be-ere 2022;Robinson 2007). A key test of efficiency is whether governments are able to provide public goods that meet the preferences of the citizens (Smith 1985;Grazzi and Jaramillo 2015). Walsh (1996) and del Granado et al. (2018) consider the issue of allocative efficiency as one of the most compelling grounds for implementing decentralization reforms. Beetham (1996) and Levitas (2017) observe that government responsiveness encapsulates the incorporation of citizens' views into policy formulation and implementation. On this basis, decentralization is said to improve the responsiveness of local governments to public needs by directly engaging citizens in the decision-making process (Bardhan 2019;Kauzya 2007). On efficiency, public choice theory suggests that decentralization leads to the creation of smaller jurisdictions that are suitable for local bureaucrats to consult the people so that the officials will have the right information about the needs within their jurisdictions. In fact, decentralization has received a lot of support on the basis that, if it could respond to citizens' needs, then it could as well propel development that addresses poverty and advances the cause of the poor (World Bank 2001;Bawole 2017;Crook 2003;Boret et al. 2021;Crawford and Hartmann 2008;Blair 2000).
On the whole, the theoretical discussions suggest that, by providing spaces for direct participation, decentralization remedies the ailments of representative democracy and effectively connect participation, decentralization, and efficiency (Andrews and De Vries 2007;Masvaure 2016). Hart (1972) and Maksimovska and Stojkov (2019) explain that decentralization not only ensures that decisions are reached democratically but that it offers citizens a double gain of both participation and responsive policy outputs. In the ensuing discussion, I seek to ascertain how these theories are manifested in decentralization practice in developing countries.
From the theoretical discussions, there is a strong consensus on the potency of decentralization to deepen participation of citizens in governance and decision-making processes. However, in practice, this is hardly the case because the evidence to ground these well-argued theories is unconvincing and lacks unanimity. The World Bank has argued forcefully in its 2001 and 2004 world development reports that decentralization gives voice to and creates a conducive climate for the participation of local people. Manor (2011) and Faguet and Shami (2022) endorsed the view that decentralization enhances citizen participation. However, the majority of the studies have been unable to produce unambiguous and conclusive evidence of improved participation in the practice of decentralization in developing countries.
For instance, Von Braun and Grote (2002) failed to find clear evidence of decentralization enhancing participation in China, India, Ghana, and Egypt; they observed that citizen participation is not guaranteed unless fiscal decentralization is preceded by political and administrative decentralization in these countries. Adams and Taabazuing (2015) also point to limited citizens' participation in Ghana's decentralization. Similarly, Bulut and Abdow (2018) reported low citizen participation under decentralization in Kenya. Devas and Grant (2003) also observed a weak connection between decentralization and participation in Uganda and Kenya, noting that, it is naïve to assume that decentralization will automatically lead to decisions reflecting local needs. This is because citizen participation 1 3 has often not yielded concrete development benefits to communities (Saito 2008;Tapscott 2008) In both Uganda and Kenya, citizen involvement in decentralized governance was reduced to participation in periodic local elections only. Local people are not consulted about resource allocation in between elections. These findings are strongly corroborated by assessments in Vietnam, Malawi, Bangladesh, Peru, and Zimbabwe as pointed out by Fritzen (2006), Tambulasi and Kayuni (2007) Panday (2011;, McNulty and Garcia (2019), and Masvaure (2016) respectively. The paucity of strong empirical evidence of improved participation under decentralization has led Andrews and De Vries (2007, 447) to conclude that "the expectation that decentralization will enhance participation seems to be based mostly on 'faith' instead of empirical evidence." This is because their investigation in four countries has also recorded no improved participation.
Two key books on the subject authored by Manor (1998) andCampbell (2003) have however reported enhanced participation under decentralization. Crook and Manor (1998) revealed that, participation in India (state of Karnataka), Bangladesh, Ghana, and Côte d'Ivoire has improved due to decentralization reforms. However, Bergh (2004) questioned this finding, arguing that participation was defined very narrowly, that is participation in local elections. But decentralization tends to perform poorly in countries where citizen participation is merely limited to elections (Jütting et al., 2004;Masvaure 2016). Participation of the people in the election of local leaders alone is not enough and does not empower the majority of the people (Crook and Sverrisson 2001;Dufhues et al. 2015). Campbell (2003) on the other hand alluded to increased grassroots participation in leading cities in Latin America under decentralization. However, Campbell's findings cannot be generalized across Latin America since only a few successful urban municipalities "leading cities" out of the over 14,000 local government units in the region were selected and examined (Gilbert 2004).
Overall, the evidence of decentralization propelling participation is not only limited and unconvincing, but it is inconclusive in two respects. One, while local citizens are mostly not homogenous in terms of wealth, most discussions of citizens' participation lump all grassroots people together (Botchway 2001;Nair 2016). However, the evidence firmly demonstrates that, there are more structural barriers facing the poor than the non-poor with respect to participation (La Ferrara 2002; Caínzos and Voces 2010;McNulty 2015). Yet, this has largely been overlooked by the literature on the subject. And two, while there are some studies on participatory budgeting (particularly in Porto Alegre, Brazil), a greater majority of studies on participation in decentralized governance have mostly looked at citizen participation from a minimalist point of view-participation in local elections. For instance, a key advocate of decentralization, the World Bank subscribes to this popular view. The World Bank (2001, 108) maintains that "one direct way of ensuring participation is to hold regular elections for local government." This paper addresses these critical gaps and contributes to a growing new body of literature, by moving beyond participation in local elections and homogeneous community participation to specifically assess how favourable decentralized participation structures and processes have been to the poor in Ghana in local development planning decisions.

The Participation Ethos in Ghana's Decentralization
Ghana's current decentralization program was introduced in 1988 by the Rawlingsled Provisional National Defence Council (PNDC) military regime. The reform was extensive and promised to give power to the people (Debrah 2014;Alam 2011;Ayee 1996). In furtherance of the agenda to devolve power to the grassroots, the PNDC enacted the Local Government Law-PNDC Law 207, and departed from past reforms by grounding decentralization laws in Ghana's 1992 Constitution. Since 1993, successive governments have made incremental administrative changes and policy reviews to decentralization. However, potentially transformational changes to decentralization since 1993 have been elusive and have not gone beyond political parties' manifesto and campaign promises (Abdulai 2017;Hoffman and Metzroth 2010).
Ghana's decentralization reform was anchored on championing participatory governance and development at the subnational level. The participation principle of decentralization was trumpeted into the hearts of Ghanaians by the government, with the assurance that the reform will genuinely give "power to the people" (Ayee 2012, 2;Alam 2011, 147;Crook 1999, 118). To actualize this, the country's, 1992 Constitution provides in Article 240[2e] that people of a particular local government area must be given the opportunity to participate effectively in their governance. Similarly, the Local Governance Act of 2016 (Act 936, hereinafter referred to as the Act), the law that governs decentralization and local governance, has explicit provisions on citizen participation. For instance, Section 83[1b] of the Act directs the district planning authority to ensure that district development plans are prepared with the full participation of local communities. To ensure that plans reflect communities' needs, the Act (Section 87) recommends that development plans be prepared below the district level, what is termed sub-district/local action plans. The participation of citizens is required before a proposed district development plan and subdistrict plan can be adopted. Section 88 of the Act directs that public hearings be organized by the district planning authority to consider the views of citizens.
A cursory reading of these provisions on participation in the law appears to promote broad citizenry involvement. A deeper analysis of these provisions however reveals that citizens have been conceived as a homogenous group and that, as a result, the participation of communities is the main focus of all citizens' engagements. These laws have in effect underrated the unequal barriers confronting the poor and the non-poor in the communities with respect to participation (La Ferrara 2002;Caínzos and Voces 2010;McNulty 2015). Although Section 48[e] of the Act encourages district assemblies to promote the participation of marginalized groups in public life, the participation being referred to here typically relates to the involvement of excluded groups in elected/public offices.
I need to emphasize that I have focused on participation in development planning because it is through the local planning process that citizens' development needs are expressed. And this forms the basis for the provision of services to address the needs of the poor. Again, participation in local elections in Ghana is less fruitful because the most powerful person who is the political and administrative head of the district assembly is not elected but appointed by the president, making him/her only accountable to the centre (Crawford 2008;Yeboah-Assiamah 2016;Abdulai 2017). Assembly members who are less influential are rather elected. I have outlined the de jure participation framework in the legal regime of Ghana's decentralization to serve as the basis for questioning how these dictates are being applied de facto in the districts' planning.
This assessment uses Ghana's decentralization as a case on three main grounds: firstly, Ghana's decentralization is seen as one of the comprehensive and leading decentralization reforms in the African continent and is therefore considered a potential model for Africa (Crawford 2009;Ayee 2012;Debrah 2014). Secondly, Ghana's decentralization has been implemented for over three decades. As such, it has matured enough for its impact on participatory planning to be assessed. This is because a period of 10-15 years is required for decentralization programs to mature for their impact to be assessed (Crook and Sverrisson 2001). And lastly, Ghana's decentralization is solidly anchored on championing grassroots participation and development. Therefore, it is the right decentralization reform or case to assess.

Research Design and Methodology
This paper draws on qualitative and quantitative data from a broader study on decentralization and pro-poor development in Ghana. A combination of qualitative and quantitative data in a single study minimizes the weaknesses of using one method and ensures a more holistic inquiry (Creswell 2014;Bryman 1992). Again, Smoke (2015) maintains that mixed methods are well suited for studying decentralization. Fieldwork was conducted in Ghana in 2019 in two districts-Nadowli-Kaleo District from the north and Atwima Kwanwoma District from the south, and in the national capital (Accra). However, in this paper, I relied largely on data from the two districts for my analysis, since in Accra I did not collect data from districts but from national-level officials. And national-level interviews are less useful in this paper. 1 These districts were purposively selected to reflect the diversity between the north and the south of Ghana. Figures 1 and 2 show the maps of Nadowli-Kaleo District and Atwima Kwanwoma District respectively. 2 For the qualitative data, purposive sampling was used to select 38 interview participants at three levels. These were community/opinion leaders, district officials, and national-level officials. Since the surveys and community-level interviews were conducted in the same communities, opinion leaders were identified by asking survey respondents to name people in their communities who are influential and are playing various leadership roles. Table 1 gives more details.
1 Only one quote was cited from a national-level official in this paper. 2 These districts are further divided into sub-districts structures. In both Nadowli-Kaleo and Atwima Kwanwoma districts, the sub-district structures present are the area councils. Several communities are grouped together to form one area council. Currently, there are seven area councils in Nadowli-Kaleo District and two area councils in the Atwima Kwanwoma District. Thus, in terms of scales, districts are the largest units, followed by area councils (sub-district structures) with the smallest units being the communities. (2023)  Quantitative data was collected through surveys at the community level only (in the 14 communities where community-level interviews were conducted). The surveys were intended to broadly capture the perceptions of the ordinary citizens, particularly the poor across both districts about how pro-poor participation in decentralized planning has been. The study districts-Nadowli-Kaleo District and Atwima Kwanwoma District, are from regions with multidimensional poverty incidence of 65.5% and 31.1% respectively 3 (GSS 2020). It is important to clarify that, the significant differences in poverty levels in the regions from which the two study districts were selected are not unusual; in general, there are geographic differences in poverty levels in the country with northern districts being generally poorer than southern districts as a longstanding legacy of colonial and post-colonial capital investments (Abdulai et al. 2018;Plange 1979). And these levels of poverty are relatively reflective of districts in general in the north and south relatively. These districts are therefore not strange districts. They were selected to represent the geographical, political, and socio-economic differences between the north and the south of Ghana.

Studies in Comparative International Development
Simple random sampling was used to calculate the households to be included based on a sample frame from the Ghana Statistical Service (GSS). This was followed by quota sampling, where samples were allocated to communities based on their population weight. In the Nadowli-Kaleo District, the sizes of the communities from which respondents were sampled range from 92 to 802 households. While the sizes of the communities surveyed in Atwima Kwanwoma District vary from 357 to 1,668 households. The use of quotas was to ensure high representativeness of the sample to the entire study population (Nardi 2018). The Nadowli-Kaleo District has 10,179 households whiles Atwima Kwanwoma District has 20,734 households (GSS 2014a, b). In an ideal situation, the sampled households for this study should have been 370 households in Nadowli-Kaleo District and 377 households in Atwima Kwanwoma District (if sample size calculation was the only consideration). 4 However, these are relatively large samples for the study. Most importantly, owing to the difficulty of collecting data in rural areas, which is time intensive, and the resource constraints on my part (as a research student then), coupled with the fact that qualitative interviews were conducted alongside the household surveys, one-fifth (74 households) of the calculated 370 households were sampled and surveyed in Nadowli-Kaleo District. And another one-fifth (75 households) of the computed 377 households were sampled and surveyed in the Atwima Kwanwoma District. In the end, I randomly sampled 74 households out of the 10,179 households in the Nadowli-Kaleo District, and 75 households out of the 20,734 households in the Atwima Kwanwoma District. Overall, a total of 149 households were sampled and surveyed in both districts. In fact, Punch (2003) and Denscombe (2014) maintain that students should use survey samples that are realistic taking into consideration their time and resource constraints.
To ensure that the voices of the poor were truly captured in the surveys, I prioritized the participation of poor and vulnerable households in the selected communities in the surveys. The poor and vulnerable households were easy to identify because they were already selected through a rigorous national poverty benchmark and were benefiting from a national anti-poverty cash transfer program for the poor and the vulnerable, called the Livelihood Empowerment Against Poverty (LEAP). At the time I conducted the surveys, 3,745 households in Nadowli-Kaleo District and 525 households in Atwima Kwanwoma District were benefiting from the LEAP program (NKDA 2020; AKDA 2019). In the end, 75% of the survey participants from both districts were recruited from households benefiting from the LEAP program.
I orally conducted all the surveys in person using the respective local languages of both districts, since I can speak both languages. The technique of personally presenting questionnaires face to face has the advantage of reducing non-response rate as the researcher can keep contacting people until the required number of respondents has been achieved (Denscombe 2014). In this regard, survey participants who were contacted but did not respond to the survey were replaced with new participants in the same community or district. In both districts, I contacted a total of 173 survey participants in order to achieve the targeted responses from 149 participants. The adoption of this approach ensured that all my questionnaires were responded to.
Qualitative data was analysed using content analysis while quantitative data was coded into SPSS and analysed. To disguise the identity of interview participants, pseudonyms were used to identify each participant. So, participants' real names were never used. The analysis in this paper triangulates and weaves these multiple sources of data together to strengthen the reliability and trustworthiness of the findings Creswell and Poth 2018).

Results and Discussions: Decentralization and Citizen Participation in Practice
In this section, I discuss the empirical findings regarding how citizen participation in local development planning is playing out on the grounds. To address the neglected but critical issue of how favourable the participation structures are, for the poor to be effectively engaged in local development planning, my findings are unpacked under three headings: direct community participation, representative participation, and how pro-poor are the participation structures and processes?

Direct Community Participation
I argue in this section that direct participation of grassroots citizens in local development planning is extremely limited in both districts. As I have stated earlier, citizen participation in decentralized governance and development is ingrained in Ghana's laws. This is envisaged to practically manifest through community participation in local elections and decision-making, specifically in district development planning. This paper focuses on participation in development planning, which is critical to improving citizens' lives but has been given less attention in mainstream discussions on decentralization as I have pointed out in the literature section. To assess how citizens' voices have been reflected in planning in Ghana's district assemblies, I engaged 16 district officials in Nadowli-Kaleo and Atwima Kwanwoma districts to ascertain how communities are involved in the preparation of the districts' mediumterm development plans. These are plans that capture the development needs and aspirations of citizens in the medium term, usually, a 4-year plan.
In responding to this, majority of the officials (11) articulated that citizens have always been consulted at the community level, where community fora are organized to broadly assess the development needs of the people for incorporation into the plan. The view of Ben, a 56-year-old official of Atwima Kwanwoma District underscored this: We first start with the preparation of the medium-term development plan. That is where the community engagements begin from. We go to each community and have stakeholder meetings about their needs and aspirations. All information is gathered, and the district planning unit collates it and forms the basis of the development plan.
However, an aspect of Ben's pronouncement is inconsistent with the National Development Planning Commission's (NDPC) guideline for organizing communitylevel planning. This is because community-level fora are supposed to be opened to all community members and not stakeholders as Ben suggested. To achieve this, the NDPC recommends wide publication of community-level fora in the local media to attract many community members to attend. When I probed further seeking to establish the veracity of the organization of community-level fora, some officials made statements that question the truism of their broad involvement of citizens at the community level. I will return to this issue after reflecting on the citizens' voices.
Community leaders (referred to in this paper as opinion leaders) and ordinary local citizens in both districts have largely dismissed suggestions by officials that community-level fora were broadly organized to assess citizens' needs for planning purposes. 5 An overwhelming majority of the opinion leaders (13) maintained that their views have never been sought at the community level for the purpose of the plan preparation. The perspective shared by Charles, a 55-year-old teacher and an opinion leader of Nadowli-Kaleo District, substantiates the wide rejection of the officials' position, "For the district assembly to come to the community and organize a meeting, to find out the problems of the people, is not common." Analogous sentiments were shared by other opinion leaders. To further demonstrate the broad nature of the refutation of the officials' claims, 74.3% of ordinary citizens of the Nadowli-Kaleo District and 84% in the Atwima Kwanwoma District have affirmed that they have never been involved in any community-level planning consultation by their respective assemblies. In the Nadowli-Kaleo District, the remaining 18.9% and 6.8% of the survey participations however said they had low involvement, and high involvement in their district planning respectively. Similarly, 10.7% and 5.3% of survey respondents in Atwima Kwanwoma District maintained that they had low involvement, and high involvement in their district planning respectively. Ofei-Aboagye (2011) and Abdul-Rahaman and Adusah-Karikari (2019) back the views of the majority citizens, arguing that involvement of grassroots people in local planning in Ghana is extremely limited at best and virtually non-existent in many districts.
As I have signalled earlier, some views shared by many officials concerning community-level participation are inconsistent and contestable. In fact, while these views were aimed at evidencing citizens' involvement, they tend rather to corroborate the above position of the citizens. At the end, officials were unable to provide reliable evidence of such community-level planning meetings. In fact, some officials appear to concede, noting that they do not go to all communities due to logistical and human resource constraints. For instance, Amina, a 46-year-old official of Nadowli-Kaleo District stressed that, "It is not everybody we target, we are constrained in terms of resources. So, we cannot go to every community." This makes sense, as it would be idealistic to expect all communities to be consulted. In fact, the NDPC (2013) recommends in its guideline that communitylevel fora should be organized in at least half of all communities in the districts. However, beyond issuing guidelines, the NDPC does not monitor to ensure that citizens are truly involved in the plan preparation (Mensah 2005). Interestingly, officials do not usually admit the assemblies' inability to undertake community-level planning in all communities upfront until I notify them that citizens have widely rejected their claims. To counteract this, some officials responded by outrightly dismissing citizens' assessments, labelling such views as coming from people with short memories. But the ubiquitous nature of citizens' rebuttal of the bureaucrats' claims raises serious doubts about suggestions that communities were broadly consulted. Again, some officials argued that although they were unable to organize community-level planning fora due to human resource and financial limitations, they observed that assembly members were always tasked to list ten needs each of their respective electoral areas for incorporation into the plan.
However, development planning, as Thompson (2006) asserts, is a highly technical and scientific process. Therefore, it is not simplistic, a mere listing of ten needs of each electoral area by the assembly members. It is rather a collective process involving the grassroots people (Muluk et al. 2019). What makes this kind of planning process baffling is that communities have different needs. Some communities have more challenges and development needs than others. The listing of an equal number of development needs for each electoral area cannot be sincere development planning. If local planning is approached this way, such a plan is unlikely to truly represent the needs of the people. This could be equated to a mundane exercise to meet the NDPC's requirements, since every district is obligated to prepare a medium-term plan. In any case, the assembly members are not trained development planners to be able to take the people through the technical processes of needs assessment and their prioritization. Beyond these, assembly members also lack the resources to hold meetings in several communities in their electoral areas. And without such meetings, the ten needs the assembly member may list, are unlikely to reflect the actual needs of the different communities in the electoral area (assembly members may only have a broad sense of the needs of their electoral areas). Assembly members lobby for support for their electoral areas. However, planning and lobbying are totally different. More importantly, the core principle of Ghana's decentralized planning is to directly engage the citizenry (Ofei-Aboagye 2011; Nyendu 2012).
The district assemblies face systemic challenges in terms of resources as they are only given 5% of national revenue from the centre and are empowered to collect only low-yielding taxes locally (Nkrumah 2000;Abdulai, 2017). Hence, with these challenges, the district bureaucrats have been unable to directly consult and broadly involve citizens in the local development planning processes. Therefore, the bureaucrats need to openly highlight the constraints making it practically difficult to directly involve citizens in the districts' decision-making endeavours instead of making inadequately persuasive arguments to justify the assemblies' extremely low involvement of the grassroots people in the districts' development planning process. I was therefore not bewildered to discover the inconsistencies in their claims on this issue. It is little wonder such claims were broadly refuted by a significant majority of the citizens in those districts.
Decentralized planning was intended to promote participatory planning, but this is being impeded by the exclusion of communities in the medium-term planning (Boachie-Danquah 2011; Ofei-Aboagye 2011). Interestingly, national-level officials who are responsible for shaping decentralization policy are not oblivious to the fact that the districts' medium-term plans are not reflecting the needs of the communities and the citizens. This is because Frank, a 57-year-old senior national-level bureaucrat, has cast serious doubt on the planning practices of the district assemblies, noting that most of those plans do not emanate from the grassroots and are merely desk work, saying, "Implementation is bad, and whether those plans have even been prepared to reflect what is on the grounds, or they are desk work. Most of them are desk work." The statement of this bureaucrat is significant because he is an insider and was a development planning officer with several assemblies for three nearly decades. His statement further solidifies the position of the grassroots people and the opinion leaders of the extremely limited involvement of citizens in the districts' development planning and decision-making process.
The extremely limited nature of the involvement of citizens in the communitylevel planning is further epitomized by the fact that only 23% of survey respondents in Nadowli-Kaleo District and 9.3% in Atwima Kwanwoma District have affirmed that their participation in community-level decision-making under the district assembly system has improved as compared to the pre-decentralization era, that is the period before 1988. The remaining 77% of respondents in Nadowli-Kaleo District and 91.7% in Atwima Kwanwoma District have asserted that their participation level either has remained at the same level, prior to decentralization (before 1988) and currently, or has worsened. Specifically, an average of 41.6% of the respondents in both districts are of the perception that their participation in planning and local decision-making under a decentralized system has deteriorated as compared to their participation in the pre-decentralization era. While the perception of deterioration in citizen participation was articulated by less than one-half of the respondents, it is concerning that some citizens would even perceive a deterioration in their participation under a reform that was instituted to improve citizens' participation. The district assemblies' unresponsiveness to citizens' development needs has led to discontent (Crawford 2008;Fridy and Myers 2019), and this could be fuelling citizens' perception of a deterioration in their participation under decentralization. Bawole (2017) argues that even in instances where limited community participation was found in Ghana, it was tokenistic, mainly intended to meet the requirements of decentralization laws and policy guidelines and not to truly involve the people in district planning and decision-making. This reinforced my earlier argument that in a local planning system where each assembly member is tasked by the planning authority of the district assembly to present ten needs of their respective electoral areas, cannot be sincere planning but only superficial, merely used as a "tick box" measure.
Some opinion leaders pointed to the lack of development interventions in their communities as further evidence that citizens' views have rarely been reflected in previous plans. They observed that if citizens were indeed consulted, some of those needs would have possibly been addressed. Tapscott (2008) however postulates that citizen participation has often not yielded concrete development benefits to communities. Despite this, even if citizens had participated with little benefit, it would have been better than this situation where many citizens feel alienated and abandoned by their assemblies. Kofi, a 65-year-old trader and an opinion leader of Atwima Kwanwoma District lamented that "I do not even think we are on their list of communities in the district." This is because his community has mostly not been involved in the district's decision-making process, so the community feels cut off from their local government. In fact, Saito (2008) acknowledges that a participation system with weak development dividends is better than no participation at all.
Decentralized participation is not limited to attendance at planning meetings at the community level. However, the depth of citizens' participation in the planning meetings cannot be assessed without the meetings being organized at the community level in the first place. Community-level meetings are the first step to opening spaces for deeper citizens' participation, yet this foundational requirement has not even been met by the district assemblies in Ghana as communities have largely been excluded in decentralized planning and decision-making processes. A parallel trend is observed in decentralized participation in other African countries. As Devas and Grant (2003) have observed that citizens' voices have not always been reflected in local prioritization of development and resource allocation under decentralization in Uganda and Kenya. Improved community participation could enhance local government performance (Robinson 2007). However, the two district assemblies in Ghana and many other developing countries are yet to make community participation a reality. Thus, Andrews and De Vries (2007) stress that the expectation that decentralization would enhance participation is merely based on faith, rather than facts.

Representative Participation
Citizen participation at the community level portends a greater chance of capturing more voices of the ordinary citizens into the plan. However, the previous section has established that this rarely happens in practice. Other participation structures exist beyond the community level. These are normally the sub-district structures. In Nadowli-Kaleo and Atwima Kwanwoma districts, these are area councils and district level/offices. Participation above the community level is usually for selected community notables and local elites, but district officials normally refer to these people as stakeholders/representatives. These local elites in the communities are typically the well-educated, well-off, chiefs, queen mothers, assembly members, and unit committee members among others. 6 Again, these local elites are normally selected to participate due to their prominence in the communities. I argue that while there is a recognition of some participation at the area councils and district levels, participation at these levels is exclusionary and limited to a few handpicked local elites. Over three-quarters of the officials of both districts reflected that, area council and district-level participation are not intended for all community members. They argue that community representatives have always been selected to participate. Ahmed, a 48-year-old official of Nadowli-Kaleo district, endorsed this view, noting that: Community-level engagements are opened to everybody, then representatives are selected from the community level to participate in the area councils. Again, representatives are selected from the area councils to participate in the district level. This arrangement makes some people feel excluded from the decision-making process because sometimes, their views only end at the community level.
If the community-level engagements were actually held, then it would have been reasonable to then select representatives from the community level to participate at the area councils and the district levels. Essentially, Ahmed's point is, if the community-level engagements were done, and is opened to all citizens, it would have still involved the ordinary citizens in a very limited way since they only participate at the basic level where key decisions are not taken. In this situation where community-level participation is severely limited, the citizens are worse off since they have extremely narrow opportunity to articulate their views in the medium-term planning process at all. The closeness of government structures to the people in the local communities by virtue of decentralization was anticipated to give voice to citizens who are mostly excluded (Crook and Sverrisson 2001;Crook 2003;Arkorful et al. 2021).
Opinion leaders' recognition of the existence of participation structures at the area council and district level is low. Approximately, one-third are aware of this. Those with knowledge of this upper-level participation structure were unequivocal in their discussions that the opportunity to participate at these levels was reserved for a few privileged elites and notables in the communities. In fact, they consider it as a rare honour to be invited to participate in those fora. In this situation, the plan that will be prepared is highly unlikely to echo the needs of the ordinary citizens. It is more likely to be elitist and largely mirror the development aspirations of the privileged few who were handpicked to participate in the plan preparation engagements. In this case, decentralized decision-making may not benefit the majority at the grassroots (Rees and Hossain 2010;Ribot 2002). District bureaucrats usually justify their use of selected representatives on the basis that those stakeholders were expected to organize community-level fora to gather the needs of communities for discussion at the area council. Again, after the meeting, they were supposed to organize another meeting in their respective communities and inform residents of the decisions made.
However, an opinion leader debunked such expectation as utopian, since those representatives are not provided with funding by the assemblies to organize such fora that require community-level participants to be refreshed with drinks. Certainly, this is a cost an unpaid representative cannot shoulder. Since the representatives are aware (based on previous experience) that community members will mostly be expecting refreshment if they organize community-level engagements, the representatives do not always organize the meetings in the communities to share with the ordinary people the issues discussed at the area councils and district offices. In fact, Charles, a 55-year-old opinion leader, has confirmed that the representatives do not usually organize those meetings, noting that, "When they [representatives] are going for the meeting they will never ask of the views of the community; and after the meeting at the district, they do not come back and share the information with the community." In this case, there are no opportunities for citizens to participate at the community level. While it may be understandable that resource constraints make it difficult for the assemblies to organize community-level planning fora, it is intriguing that the bureaucrats would then expect the unpaid representatives to organize community meetings in their respective communities with their own resources.
The officials have even admitted that their handpicked representatives hardly brief their communities. For instance, Amina, a 46-year-old district official conceded that, "When they go back, they have to organize their people and give them the information; but I doubt if some do that." Even if they were able to relay the information, it would have still been a minimalist way of involving the people in the districts' planning. Gaber (2019) argues that such an approach to citizen participation ignores citizens' experiences; participation in this regard is merely a public relation gimmick. This kind of participation where decisions are taken, and people are only informed afterwards, is described by Sherry Arnstein (1969) in her influential article, as participation by informing; it is tokenistic and one of the weakest forms of participation because it encourages a one-way flow of information from the representatives to citizens, it provides superficial information and discourages questions. In this kind of participation, citizens have limited or no opportunity to influence policy or program design. Even though this kind of participation is weak, it is hardly followed by the representatives due to the constraints I have highlighted earlier. Khan (2013) and Steiner (2007) argue that the involvement of the people using their representatives is a weak form of participation. Direct citizens' participation is considered the most effective way of involving them in the decision-making processes (Andrews and De Vries 2007;Crook and Sverrisson 2001). This is the kind of participation decentralized governance seeks to engender (Masvaure 2016;Faguet and Shami 2022). However, it appears structural resource constraints have made it challenging for the Nadowli-Kaleo and Atwima Kwanwoma District Assemblies to directly involve citizens in the planning and decision-making processes. Consequently, the assemblies' severely limited use of community-level planning and the prioritization of local elites in planning is an ineffective way of involving the grassroots people in local-level development. This is not only an aberration of the constitutional injunction but also an affront to the much-publicized pro-poor participation mantra of Ghana's decentralization (Yeboah-Assiamah 2016;Haynes 1991). Khan (2013) posits that citizen participation positively enhances their interest in local governance processes.
From the foregoing discussions, since community-level participation is extremely weak, and the area councils and district-level participation also exclude the ordinary citizens, therefore, any development plan that emanates from such largely exclusionary processes will be driven by the views of the district officials and local elites. However, this is not the kind of participation decentralization seeks to foster. This is because many decentralization scholars have argued that decentralized governance is greatly weakened if local elites capture subnational participation processes, resources, or power (Johnson 2003;Crook and Sverrisson 2001;Cheema 2007;Khan 2013;Blair 2000, Elliott et al. 2018. My findings amply suggest that the districts' development planning has fundamentally been captured by local elites and champions under the guise of representative participation that significantly excludes the masses at the community level in planning and critical decision-making in the assemblies. In this situation, Ghana's decentralized planning is failing to reflect the aspirations of the masses at the local level as it was initially trumpeted. This form of participation cannot mirror the views of the grassroots people, and this cannot be practised in a responsive decentralized governance system. This does not also empower the grassroots people as trumpeted by Ghana's decentralization which explicitly promised to give "power to the people" (Ayee 2012, 2;Alam 2011, 147;Crook 1999, 118). Ghana's decentralization was intended to bring government structures closer to the people so that grassroots residents can participate directly in decisions that affect them. The structures of participation designed to encourage representative decision-making work against the stated intentions of the country's decentralization (Be-ere 2021). International Development (2023) 58:280-307 How Pro-poor Are the Participation Structures and Processes?

Studies in Comparative
The structures and processes of participation employed in the district development planning mostly determine whether the voices of the poor will be amplified in the plan or not. I argue in this section that Ghana's decentralization laws frame citizens as a homogenous group and that this has engendered the use of generic participation structures that failed to adequately reflect the voices of the poor in the development plans. My engagement with officials from both districts to ascertain how the poor have been involved in the local development plan preparation has revealed disappointing but unsurprising responses. None of the officials could point to a deliberate measure adopted to improve the participation of the poor in the plan preparation process. Most bureaucrats underrated and at times dismissed questions about targeting of the poor for inclusive and effective participation by resorting to these phrases: "we do not discriminate" "generally we are all poor" "that is why we have their representatives". Ironically, when local elites are selected by the officials to participate at the area councils and district offices, the officials do not normally consider this as discriminatory. These elites are nicely christened by the bureaucrats as stakeholders/representatives. It is important for me to state that while the bureaucrats' statements regarding targeting of the poor in the participation processes are worrying, they were not without basis. This is because the Ghanaian Constitution and the local governance Act only limit participation to communities. These laws consider citizens homogenous in terms of wealth (Botchway 2001). Hence, once communities participate, it is assumed that the views of every citizen will be reflected in the plan. However, there are more structural barriers facing the poor than the non-poor with regard to participation (La Ferrara 2002;Caínzos and Voces 2010).
The generic approach of Ghana's laws to participation has made district officials perceive targeting of the poor for effective participation as discriminatory. Yet, Singh (2003) asserts that interventions targeted at the general population tend to have very limited impact on the poor. Hence, he recommends targeting of the poor as an effective way to making participation and development interventions more pro-poor. Similarly, Kauzya (2007) argues that since some groups require special encouragement to participate, homogenous participation structures should not be used. This is because a generic approach to participation in planning is unlikely to yield development interventions that truly address the multiple deprivations confronting the poor.
What is more concerning is that, even the generic community participation hardly happens. However, three-quarters of the district officials from both districts contended that community participation reflects the voices of the poor adequately. They argued that all classes of people can gather in the community forum and express their views without anyone being discriminated against. For instance, Ali, a 47-yearold official of Nadowli-Kaleo District supported this view by, arguing that: At the preparation of the plan, we go to each of the communities. And all classes of people: students, girls, women, the aged are supposed to sit at the community level and then they are guided to come out with their needs. Some opinion leaders tend to agree with the officials that community-level meetings would have provided an opportunity for the poor to participate. But they were quick to proclaim that, such meetings were not organized at the community level. James, a 68-year-old retired teacher and an opinion leader of Nadowli-Kaleo District, questions how the poor could know what is transpiring at the district assembly without community-level engagements, "It would have been easier for the poor to participate if the meeting were called at the community level. How will the poor sit in his dark room and know what is happening at the assembly?" Similarly, Luke, a 64-year-old retired teacher of Atwima Kwanwoma District said there were no community-level participation channels, and that the poor are not heard, "It will be difficult for the poor to express their views because we do not have such meetings that the poor could participate. Their inputs are not taken, the poor are not heard." These opinions are broadly consistent with the views expressed by ordinary citizens in the surveys. A massive 90.5% of respondents in Nadowli-Kaleo district and 85.3% in Atwima Kwanwoma district stated that the immensely limited existence of participatory structures at the community level has made it extremely difficult for the ordinary citizen, particularly the poor to meaningfully participate in the district assemblies' planning processes. These categorical views from citizens point to the exclusion of the masses in critical development decisions. Bulut and Abdow (2018) found community participation weak in Kenya's counties due to high illiteracy among citizens. However, in Ghana, grassroots participation was pathetic not because of illiteracy but due to the failure of both district assemblies to provide favourable participation opportunities to the poor at the community level (Abdul-Rahaman and Adusah-Karikari 2019). As I have argued, a generic community-level participation is unsuitable to truly reflect the voices of the poor and disadvantaged groups. Yet, some of the poor feel that in the absence of well-tailored fora that specifically target them, homogenous community-level engagement was better than the current situation where communities have virtually no platform to voice their development needs to their planning authorities. Joel, a 50-year-old health worker and an opinion leader of Nadowli-Kaleo District, argues that "They [officials] need to involve the poor, if they do not involve them, how will they know their problems?" It is important to clarify that there is no aversion to identifying people as being poor in Ghana for participatory planning, decision-making, and other development purposes.
It would be difficult for the district officials to truly know the problems of the poor without broadly involving them in any of the planning and decision-making structures (Adams and Taabazuing 2015). For a development plan to be responsive to the needs of the poor, the district planners must plan with the poor, and not plan for them. Arnstein (1969) observes that even when citizens are involved in the planning process, if their role is peripheral, they are still being planned for. Therefore, in this situation where the poor are rarely involved in local planning, it is not likely they will benefit from services emanating from a plan whose preparation they have not participated in.
The exclusion of the poor in the districts' participatory channels is leading to their apathy towards the district assemblies. It is not surprising when one of the opinion leaders-a 73-year-old pensioner of Atwima Kwanwoma District, discloses that, "The problem is that the poor do not benefit a lot from what is going on. So, the poor are left behind, and they do not take an interest in what the assembly does." This signals disillusionment of the poor towards the district assembly. While this is worrying, it is not unexpected because the poor have not benefited from decentralization as it was initially preached (Crook 2003). The local people see the status quo being maintained where they continue to remain voiceless and left behind. Citizens become disenchanted when their local governments cannot improve their lives (Tapscott 2008;McNulty and Garcia 2019;Smoke 2003). This appears to be the case in Ghana, exemplified in the two districts. In fact, Fridy and Myers (2019) confirm citizens' disillusionment towards the assemblies, as they no longer consider the assemblies as institutions that can address their development needs.
What is troubling is that when assemblies seek broad citizens' participation in a forum, such engagements are always intended to sensitize the people on the need to pay local taxes and not to discuss planning/development issues. Sam's (a 49-year-old official of Nadowli-Kaleo District) statement supports this assertion, he stated that, "When it is about revenue mobilisation, the assembly officials will ask the assembly members to bring all the people to the area council, so that they will explain to them the need to pay taxes" The officials seem more interested in getting the people to pay taxes than involving them in the planning processes. Payment of taxes is important for mobilizing resources for development. Despite this, if citizens feel excluded from participating in local development and governance processes, they will be demotivated from honouring their tax obligations (Steiner 2007;Smoke 2003;Bardhan 2019). In Brazil, as citizens were allowed to participate in the budgetary processes in Porto Alegre, they were motivated to pay their taxes resulting in a 50% rise in the revenue of the municipality (Cheema 2007). This suggests that citizens' participation in planning/decision-making should be prioritized, since this has a concomitant effect on their willingness to pay taxes.
Decentralization in South Africa and Rwanda has improved citizens' participation in development planning and decision-making; thus, placing communities' development destinies in their hands 7 (Kauzya 2007). Ghana's decentralization was also intended to enhance the popular participation of citizens in local development and governance (Crook 1999;Ghana 1992). However, my findings have laid bare the fact that in practice, this is hardly the case. In this situation, the poor who are mostly found at the community level, have been excluded in local planning and decisionmaking endeavours. Therefore, their development needs/aspirations are unlikely to be captured by a plan whose preparation process has largely eliminated and silenced their voices. The unfavourable decision-making and planning processes have left the poor behind in a state of powerlessness despite decentralization. Other levels of participation exist at the area councils and district levels, but these are not opened to the public or the poor to participate in. Local elites are handpicked to participate. This exclusion of the poor, who are mostly the majority in the districts is resulting in surging apathy of the excluded people toward the assemblies. The people are beginning to lose trust in the district assembly as a potent tool for grassroots participation 1 3 and development. Hence, they are turning to their personal agency to improve their lives. A participation process that is hijacked and is elite-driven is highly unlikely to propel development that truly serves the needs of the poor at the grassroots (Blair 2000;McNulty 2015;Crook 2003;Devas and Grant 2003;Khan 2013;Crook and Sverrisson 2001).

Conclusion
Based on theoretical conceptualizations, donors and other decentralization enthusiasts strongly advocated decentralization as the silver bullet for increasing citizens' participation in governance, with the concomitant effect of promoting pro-poor development (World Bank 2001;Crook 2003;Crawford and Hartmann 2008). This paper questions this popular claim by using Ghana's decentralization reform to assess how pro-poor citizen participation in local development planning has been. The assessment reveals that, Ghana's Constitution and the Local Government Act have treated citizens as a homogenous group, hence these laws require district assemblies to ensure generic participation of communities in district development planning. Although homogenous community participation disadvantages the poor, in practice, community-level participation of citizens is extremely limited. Opinion leaders and ordinary citizens in both Nadowli-Kaleo and Atwima Kwanwoma districts have roundly rejected local bureaucrats' assertions of broad communities' involvement in planning, firmly proclaiming that citizens are rarely involved in the district assemblies' local planning.
The limited evidence of participation in the district assemblies' planning is found at the area councils and the district level. However, these participatory channels are not opened to the public. Local elites and representatives are rather handpicked to participate at those levels. The poor are excluded in such deliberations. This has led to the marginalization of the poor in key planning decisions. The non-existence of community-level planning and the use of local elites in the area councils and the district level has resulted in a planning structure that has been captured and driven by local elites (Crawford 2004). Therefore, the development needs and aspirations of the poor are unlikely to be captured by a plan whose preparation process has significantly obliterated and silenced their voices. The unfavourable planning and decision-making process has left the poor behind in a state of voicelessness and powerlessness despite decentralization.
Since the districts' plans are failing to serve the poor as initially anticipated, there is surging apathy regarding these excluded segments of the society towards the assemblies and their activities. The current structures of participation used by the district assemblies are designed to encourage representative decision-making. This works against the articulated intentions of the country's decentralization. This does not empower the people at the grassroots as touted by Ghana's decentralization which explicitly promised to give "power to the people" (Debrah 2014, 56;Alam 2011, 147). After practising decentralization for over three decades in Ghana, citizens' participation in governance, planning, and decision-making remains extremely weak and highly elitist. This finding does not only call into question the one-size-fits-all proposition by donors and other decentralization advocates that decentralization enhances citizens' participation and increases government responsiveness to the poor; but it also challenges such claims. From this assessment, decentralization can only improve grassroots participation and amplify the voices of the poor in subnational governance, development, and decision-making processes if pro-poor participation structures are solidly ingrained in decentralization designs and praxis.