For Mwanza:

In Mwanza, there are two districts and each has urban and rural wards. Urban wards (14 in number) were purposively selected because they receive solid waste management (SWM) services. Every urban ward receives services either from a Community Based Organization (CBO) or a private company working under a contract so all urban areas irrespective of in‐ come levels receive SWM services [8]. For the survey, stratified sampling was used. It was a disproportionate stratified sampling because for all the 14 urban wards (which were the strata), on average 13-14 households were interviewed irrespective of population figures in each ward. The goal was to have each ward represented by a minimum number of 10 house‐ holds and to arrive at a maximum total sample of 200 households. Using the population cen‐ sus data and with the help of the ward leaders and the public health officers posted to each ward, households were randomly picked from a list and questionnaires administered.

#### For Kisumu:

In Kisumu, SWM is patterned much more along income levels. Previous empirical work, see [9], indicated that the council had not officially permitted non-state actors to operate but all SWM activities by these actors went on unofficially. SWM service providers (the non-state actors) defined their clients on the basis of their income. Community self-help groups were common in low income areas while private companies dominated high and middle income areas. Therefore a list of low, middle and high income estates was made, and then a few es‐ tates from each of the three categories were randomly picked

NB: Just like in Mwanza, Kisumu has civic wards, some of these wards are estates in them‐ selves for instance, Nyalenda is a ward and an estate at the same time, while in other wards, there is more than one estate. There are in total 17 civic wards covering 41 estates.

There are about 41 estates recognized by the council (11 high income, 17 middle income and 13 low income). The council provides waste management services in only 12 of these estates. The study aimed to administer 200 questionnaires just like in Mwanza. About half the num‐ ber of estates in each income category mentioned earlier was randomly selected. The study ended up administering questionnaires in 6 high income, 9 middle income and 7 low in‐ come estates. About 10 questionnaires were administered in each of the estates selected. The number of households per estate varies from about 3,200 to about 12,000. Selecting house‐ holds within the selected estates was done differently depending on the kind of estate ar‐ rangement. In planned estates like the Railway estates, the houses are numbered and organized in a certain pattern so it was easy to do systematic sampling, selecting every fourth household. In the informal estates like Nyalenda, the houses are not numbered or ar‐ ranged in any particular order,

the researchers were guided by the village names within Nyalenda which were listed and then one household from each village was randomly selected. To avoid covering a village more than once, the study used one research assistant per estate.

The income categorization used to obtain the sample reflects the general pattern of service provision in the town. The estates were classified according to one of the three distinct in‐ come categories. From each group of estates about half was randomly selected. Within each selected estate about 10 households were randomly selected. Therefore, based on the argu‐ ment of inferential statistics, the data can be considered to reflect the situation of service pro‐ vision among the different income categories in the town as a whole

For Jinja:

For Mwanza:

218 Environmental Change and Sustainability

For Kisumu:

ranged in any particular order,

In Mwanza, there are two districts and each has urban and rural wards. Urban wards (14 in number) were purposively selected because they receive solid waste management (SWM) services. Every urban ward receives services either from a Community Based Organization (CBO) or a private company working under a contract so all urban areas irrespective of in‐ come levels receive SWM services [8]. For the survey, stratified sampling was used. It was a disproportionate stratified sampling because for all the 14 urban wards (which were the strata), on average 13-14 households were interviewed irrespective of population figures in each ward. The goal was to have each ward represented by a minimum number of 10 house‐ holds and to arrive at a maximum total sample of 200 households. Using the population cen‐ sus data and with the help of the ward leaders and the public health officers posted to each ward, households were randomly picked from a list and questionnaires administered.

In Kisumu, SWM is patterned much more along income levels. Previous empirical work, see [9], indicated that the council had not officially permitted non-state actors to operate but all SWM activities by these actors went on unofficially. SWM service providers (the non-state actors) defined their clients on the basis of their income. Community self-help groups were common in low income areas while private companies dominated high and middle income areas. Therefore a list of low, middle and high income estates was made, and then a few es‐

NB: Just like in Mwanza, Kisumu has civic wards, some of these wards are estates in them‐ selves for instance, Nyalenda is a ward and an estate at the same time, while in other wards,

There are about 41 estates recognized by the council (11 high income, 17 middle income and 13 low income). The council provides waste management services in only 12 of these estates. The study aimed to administer 200 questionnaires just like in Mwanza. About half the num‐ ber of estates in each income category mentioned earlier was randomly selected. The study ended up administering questionnaires in 6 high income, 9 middle income and 7 low in‐ come estates. About 10 questionnaires were administered in each of the estates selected. The number of households per estate varies from about 3,200 to about 12,000. Selecting house‐ holds within the selected estates was done differently depending on the kind of estate ar‐ rangement. In planned estates like the Railway estates, the houses are numbered and organized in a certain pattern so it was easy to do systematic sampling, selecting every fourth household. In the informal estates like Nyalenda, the houses are not numbered or ar‐

the researchers were guided by the village names within Nyalenda which were listed and then one household from each village was randomly selected. To avoid covering a village

The income categorization used to obtain the sample reflects the general pattern of service provision in the town. The estates were classified according to one of the three distinct in‐

there is more than one estate. There are in total 17 civic wards covering 41 estates.

tates from each of the three categories were randomly picked

more than once, the study used one research assistant per estate.

In Jinja, there are three divisions [6]. Within each division there are parishes but solid waste management services have been contracted out per division and the work is given to two contractors. The divisions are:


One contractor serves both Central and Walukuba divisions while the other contractor serves Mpumudde division. Service is therefore not structured along income levels as in Ki‐ sumu neither per ward as is the case in Mwanza. Waste is collected from skips (collection points) and not directly from the households. Contractors are paid per emptied skip. There are 119 collection points (skips) in Central, 10 in Walukuba and 20 in Mpumudde. Central and Mpumudde Divisions were picked for the study in order to show the differences (if any) in service provided by the two contractors. Frequency of questionnaires administered was higher in Central with about 180 questionnaires randomly distributed and 38 randomly distributed in Mpumudde. These frequencies were more or else in line with the distribution of collection points which are many in Central division. In the end, in total 218 household questionnaires were administered, so the aim to get a total minimum of 200 questionnaires, as for the other two towns, was reached.

With a list of street names (also referred to as roads/avenues/zones which are equivalent of the villages in Kisumu), households (numbering up to 10 in certain streets) were randomly selected from each street.

#### **2.3. Theoretical framework**

In East Africa, the national state in each country has been and is the predominant actor when it comes to service provision. Recently though, the influence of market and civil soci‐ ety stakeholders is evident particularly in service provision at the local level. In this paper these developments are analysed using the policy arrangement approach developed by Arts and others [5, 10, 11]. As an approach, it has been mainly used in studies conducted in West‐ ern countries but can also provide a framework against which solid waste management is analysed in East Africa given the entry of non-state actors. Policy arrangement is defined as the temporary stabilization of the content and organization of a policy domain at a specific level of policy making. It is temporary because arrangements are under pressure of constant change either by policy innovations on the ground or by processes of political moderniza‐ tion. It has four main dimensions namely:

*For Jinja:* 

 Central Division Walukuba Division Mpumudde Division


In Jinja, there are three divisions [6]. Within each division there are parishes but solid waste management services

One contractor serves both Central and Walukuba divisions while the other contractor serves Mpumudde division. Service is therefore not structured along income levels as in Kisumu neither per ward as is the case in Mwanza. Waste is collected from skips (collection points) and not directly from the households. Contractors are paid per

contractors. Frequency of questionnaires administered was higher in Central with about 180 questionnaires randomly distributed and 38 randomly distributed in Mpumudde. These frequencies were more or else in line with the distribution of collection points which are many in Central division. In the end, in total 218 household questionnaires

have been contracted out per division and the work is given to two contractors. The divisions are:


These four dimensions of a policy arrangement are inextricably interwoven. This means that any change on one of the dimensions induces change on other dimensions. This relationship is sym‐ bolized by the tetrahedron, in which each of the corners represents one dimension (Figure. 2). decision-making; and The current policy discourses where the concept of discourse refers to the views and narratives of the actors involved—in terms of norms and values, definitions of problems and approaches to solutions—in this case, the study captures network governance These four dimensions of a policy arrangement are inextricably interwoven. This means that any change on one of

the dimensions induces change on other dimensions. This relationship is symbolized by the tetrahedron, in which

5

**Figure 2.** Tetrahedron showing Relation between Dimensions of Policy Arrangement

each of the corners represents one dimension (Figure. 1).

sg el o

An analysis of an existing policy arrangement, including its problems or sticking points, concerns all four dimensions of the concept. The methods for mapping out the relevant ac‐ tors, their coalitions and oppositions are familiar from network analysis. Methods are also available for assessing power relationships. Then existing rules of the game in the arrange‐ ment have to be reconstructed: Who decides on the agenda? Who participates in the policy game? Who is excluded? Who takes the decisions? Discourse analysis provides systematic instructions for analyzing the fourth dimension: What are the main concepts in policy dis‐ course and the policy programme? What are the basic assumptions of the policy? What do relevant policy documents contain? How do the various players in the field interpret the policy concepts and basic assumptions?[12].

According to [11], policy innovations can be initiated from each of the dimensions. Policy agents may decide: (1) to allow more or new actors to participate in policy making or in coali‐ tion formation; (2) to reshape power relations, for example by adding to or withdrawing re‐ sources from a policy arrangement; (3) to reformulate the rules of the game on the basis of which policies are made; and (4) to reformulate the policy discourse concerned, for example by redefining its core concepts. However, innovations in one dimension tend to have consequen‐ ces for other dimensions, and even for the arrangement as a whole. In other words, in some cas‐ es changes have been initiated by new coalitions (e.g. the participation of citizen groups), whereas in other cases they are provoked by innovative discourses, or reinforced by rules and resources, setting off a chain reaction of changes in all aspects. Finally, this chain may lead to the change of *entire* policy arrangements. The approach of policy arrangements helps to ana‐ lyse such changes.

#### **2.4. Changing political dynamics in service provision**

**•** The actors and their coalitions involved in the policy domain; and

outcomes and how;

220 Environmental Change and Sustainability

suit of policy and decision-making; and

sg el o

reached.

*For Jinja:* 

 Central Division Walukuba Division Mpumudde Division

 **2.4 Theoretical framework:** 

has four main dimensions namely:

decision-making; and

Rules of the game

policy concepts and basic assumptions?[12].

**Figure 2.** Tetrahedron showing Relation between Dimensions of Policy Arrangement

the study captures network governance

each of the corners represents one dimension (Figure. 1).

**•** The division of power and influence between these actors, where power refers to the mo‐ bilization, division and deployment of resources, and influence to who determines policy

households (numbering up to 10 in certain streets) were randomly selected from each street.

In Jinja, there are three divisions [6]. Within each division there are parishes but solid waste management services

One contractor serves both Central and Walukuba divisions while the other contractor serves Mpumudde division. Service is therefore not structured along income levels as in Kisumu neither per ward as is the case in Mwanza. Waste is collected from skips (collection points) and not directly from the households. Contractors are paid per emptied skip. There are 119 collection points (skips) in Central, 10 in Walukuba and 20 in Mpumudde. Central and Mpumudde Divisions were picked for the study in order to show the differences (if any) in service provided by the two contractors. Frequency of questionnaires administered was higher in Central with about 180 questionnaires randomly distributed and 38 randomly distributed in Mpumudde. These frequencies were more or else in line with the distribution of collection points which are many in Central division. In the end, in total 218 household questionnaires were administered , so the aim to get a total minimum of 200 questionnaires, as for the other two towns, was

With a list of street names (also referred to as roads/avenues/zones which are equivalent of the villages in Kisumu),

In East Africa, the national state in each country has been and is the predominant actor when it comes to service provision. Recently though, the influence of market and civil society stakeholders is evident particularly in service provision at the local level. In this paper these developments are analysed using the policy arrangement approach developed by Arts and others [5, 10, 11]. As an approach, it has been mainly used in studies conducted in Western countries but can also provide a framework against which solid waste management is analysed in East Africa given the entry of non-state actors. Policy arrangement is defined as the temporary stabilization of the content and organization of a policy domain at a specific level of policy making . It is temporary because arrangements are under pressure of constant change either by policy innovations on the ground or by processes of political modernization. It

 The division of power and influence between these actors, where power refers to the mobilization, division and deployment of resources, and influence to who determines policy outcomes and how; The rules of the game currently in operation, specifically the formal procedures for pursuit of policy and

 The current policy discourses where the concept of discourse refers to the views and narratives of the actors involved—in terms of norms and values, definitions of problems and approaches to solutions—in this case,

These four dimensions of a policy arrangement are inextricably interwoven. This means that any change on one of the dimensions induces change on other dimensions. This relationship is symbolized by the tetrahedron, in which

Resources

have been contracted out per division and the work is given to two contractors. The divisions are:

**•** The rules of the game currently in operation, specifically the formal procedures for pur‐

**•** The current policy discourses where the concept of discourse refers to the views and nar‐ ratives of the actors involved—in terms of norms and values, definitions of problems and

These four dimensions of a policy arrangement are inextricably interwoven. This means that any change on one of the dimensions induces change on other dimensions. This relationship is sym‐ bolized by the tetrahedron, in which each of the corners represents one dimension (Figure. 2).

5

An analysis of an existing policy arrangement, including its problems or sticking points, concerns all four dimensions of the concept. The methods for mapping out the relevant ac‐ tors, their coalitions and oppositions are familiar from network analysis. Methods are also available for assessing power relationships. Then existing rules of the game in the arrange‐ ment have to be reconstructed: Who decides on the agenda? Who participates in the policy game? Who is excluded? Who takes the decisions? Discourse analysis provides systematic instructions for analyzing the fourth dimension: What are the main concepts in policy dis‐ course and the policy programme? What are the basic assumptions of the policy? What do relevant policy documents contain? How do the various players in the field interpret the

According to [11], policy innovations can be initiated from each of the dimensions. Policy agents may decide: (1) to allow more or new actors to participate in policy making or in coali‐ tion formation; (2) to reshape power relations, for example by adding to or withdrawing re‐ sources from a policy arrangement; (3) to reformulate the rules of the game on the basis of which policies are made; and (4) to reformulate the policy discourse concerned, for example by

Actors

Discourse

approaches to solutions—in this case, the study captures network governance

The actors and their coalitions involved in the policy domain; and

The existence and performance of local authorities in East Africa in service provision, has a his‐ torical component depicting changes that have occurred dating back to the 1960's when the three East African countries attained independence. Olowu [13] explains that when these coun‐ tries attained political independence with formal structures of democratic, representative gov‐ ernment, political leaders in their bid to consolidate political power then opted for highly centralized modes of governance. This centralized mode of governance was reinforced by a culture of politics of patrimony in which all powers and resources flow from one source of power ('the father of the nation') to clients to shore up the regime. This pattern of power and re‐ source distribution was strongly supported by both domestic and external actors until the late 1980s. The reasons adduced for adopting this approach included –rapid economic and social development actualized through centralized planning, unity and national integration, contain‐ ment of corruption and political stability. In fact the argument was that if decentralization would be necessary at all it must be in the form of administrative decentralization or deconcen‐ tration—the sharing of responsibilities between central and local administrations which do not exercise any discretionary authority nor dispose of resources. Yet the 1990s marked an era of political and democratic approaches wherein decentralization was progressively being seen (by governments, external actors and the increasingly influential civil society lobbies) as a means of enhancing democracy and citizen participation and (by governments and external ac‐ tors) as a way of reducing the role, and in particular the expenditures, of the central govern‐ ment [14]. Over time, these changes have necessitated governmental reconfigurations, many of which have a powerful 'local' governance orientation. They include resurgent regional organi‐ zations, public private partnerships in infrastructure creation and maintenance and service de‐ livery, decentralisation, devolution and deconcentration of expertise and accountabilities within government departments, and contractual relationships between government and community providers, among others [15]. Non-state actors are increasingly getting involved in service provision. To date though, none of the three East African countries has any municipal service that is completely privatized as yet neither are there distinct policies on privatization of service provision within local authorities. The services most experimented with so far are solid waste management and water supply, but the former more than the latter.

#### **2.5. Existing solid waste management arrangements**

The arrangements of SWM in the three urban centres differ. This is captured within the four dimensions of policy arrangement:

#### *2.5.1. Rules*

#### *2.5.1.1. Jinja*

Jinja has three administrative divisions: Central, Walukuba and Mpumudde. Due to the efforts to move towards privatization in the country, the council contracted out solid waste manage‐ ment through open bidding, specifically collection and transfer of waste to the disposal site. The arrangement is an annual contract between the municipal council and the private entrepre‐ neurs and in the financial year 2008/2009 two contractors won the tenders [16]. One serves two divisions: Central and Walukuba, while the other contractor provides service to Mpumudde. The legitimacy of the contractors is realized in a number of ways. First the Jinja Solid Waste Management By-Laws 2005, in their objectives recognize the role of private companies in the collection and disposal of waste when this is practiced in a sustainable manner and at a fee. The two contractors serving then were thus officially recognized by the council and in turn by a number of the households they served. Secondly, the tenders for contracts in SWM are adver‐ tised through the media and as earlier mentioned, there is open bidding. Contracts for compa‐ nies that were involved previously can only be renewed on the basis of their performance. As far as decision making is concerned, the municipal authority is still at the helm of SWM, mak‐ ing policies and seeing to their implementation. The contractors do not attend council meetings and are therefore unable to make or influence decision-making. From an interview with one of the contractors however, it became clear that they are free to voice their opinions directly to the town clerk, which may or may not be taken into account when formulating policies.

#### *2.5.1.2. Mwanza*

In Mwanza, there are 21 wards, out of which only 14 wards receive solid waste management services. These are the wards in the urban sections of Mwanza city. Privatization of solid waste management resulted in the council awarding contracts to groups and in the financial year 2008/2009, contracts were awarded to Community Based Organizations and two private com‐ panies that serve the wards in the Central Business District [7, 16]. Every other ward is served by one or two Community Based Organizations (CBOs). In terms of legitimacy, it should be noted that first, the CBOs and the two private companies are legitimate organizations, official‐ ly recognized by the council and the people they serve as revealed by the interviews and house‐ hold survey. They are awarded formal contracts after having won through a democratic process. The groups undergo registration as solid waste management service providers and pay a registration fee of Tshs25,000 (USD17.85). The private companies pay taxes to the Tanza‐ nia Revenue Authority. Secondly these contractors are well known to the people they serve be‐ cause the members of these CBOs are local and belong to/are residents in the wards they serve.

#### *2.5.1.3. Kisumu*

In Kisumu, even with the municipality as the central locus of authority, see [17], legitimacy re‐ mains a key concern. This is so because there are questions regarding the legal mandate accord‐ ed to groups providing SWM informally. Most of these informal groups are registered by the ministry in charge of community development, they are however not formally recognized by law as actors in the domain of solid waste management. The presence and activities of these groups are nevertheless known by the council and some of these groups (see table 1 below) even responded to be operating through some form of 'franchise'1 in areas allocated to them by the municipal authority. With no legal papers to show the arrangements they are part of, most if not all of these groups are not legitimate in the SWM arena. This impinges on a number of is‐ sues, for instance seeking legal redress in case of payment defaults becomes a problem. Getting donor assistance also becomes a problem because questions will arise as concerning ties to the public, transparency and adherence to the mission of a group, representative status and the re‐ lationship between the group and the community served. On the other hand, in terms of com‐ munity support, openness of information, democratic decision-making, these groups can be considered more legitimate than some official actors are.


**Table 1.** Legitimacy of groups involved in SWM services Source: Author based on field work

#### *2.5.2. Actors and coalitions*

*2.5.1. Rules*

222 Environmental Change and Sustainability

*2.5.1.1. Jinja*

*2.5.1.2. Mwanza*

*2.5.1.3. Kisumu*

Jinja has three administrative divisions: Central, Walukuba and Mpumudde. Due to the efforts to move towards privatization in the country, the council contracted out solid waste manage‐ ment through open bidding, specifically collection and transfer of waste to the disposal site. The arrangement is an annual contract between the municipal council and the private entrepre‐ neurs and in the financial year 2008/2009 two contractors won the tenders [16]. One serves two divisions: Central and Walukuba, while the other contractor provides service to Mpumudde. The legitimacy of the contractors is realized in a number of ways. First the Jinja Solid Waste Management By-Laws 2005, in their objectives recognize the role of private companies in the collection and disposal of waste when this is practiced in a sustainable manner and at a fee. The two contractors serving then were thus officially recognized by the council and in turn by a number of the households they served. Secondly, the tenders for contracts in SWM are adver‐ tised through the media and as earlier mentioned, there is open bidding. Contracts for compa‐ nies that were involved previously can only be renewed on the basis of their performance. As far as decision making is concerned, the municipal authority is still at the helm of SWM, mak‐ ing policies and seeing to their implementation. The contractors do not attend council meetings and are therefore unable to make or influence decision-making. From an interview with one of the contractors however, it became clear that they are free to voice their opinions directly to the

town clerk, which may or may not be taken into account when formulating policies.

In Mwanza, there are 21 wards, out of which only 14 wards receive solid waste management services. These are the wards in the urban sections of Mwanza city. Privatization of solid waste management resulted in the council awarding contracts to groups and in the financial year 2008/2009, contracts were awarded to Community Based Organizations and two private com‐ panies that serve the wards in the Central Business District [7, 16]. Every other ward is served by one or two Community Based Organizations (CBOs). In terms of legitimacy, it should be noted that first, the CBOs and the two private companies are legitimate organizations, official‐ ly recognized by the council and the people they serve as revealed by the interviews and house‐ hold survey. They are awarded formal contracts after having won through a democratic process. The groups undergo registration as solid waste management service providers and pay a registration fee of Tshs25,000 (USD17.85). The private companies pay taxes to the Tanza‐ nia Revenue Authority. Secondly these contractors are well known to the people they serve be‐ cause the members of these CBOs are local and belong to/are residents in the wards they serve.

In Kisumu, even with the municipality as the central locus of authority, see [17], legitimacy re‐ mains a key concern. This is so because there are questions regarding the legal mandate accord‐ ed to groups providing SWM informally. Most of these informal groups are registered by the ministry in charge of community development, they are however not formally recognized by


**Table 2.** Summation of SWM Indicators in the three urban centres Source: Author based on field work

<sup>1</sup> The word franchise is in quotation marks because there are no legal papers to show for it and the arrangement is only fran‐ chise by name but not in actual sense.
