Informal Worker Access to Formal Social Protection in Sub-Saharan Africa: Evidence from Kenya and Tanzania

Abstract Based on a combination of detailed survey data and in-depth interviews with informal workers in the construction, micro-trade, and transport sectors in urban areas of Kenya and Tanzania, this article explores the extent to which informal worker associations facilitate member access to formal social insurance schemes. The article provides unique insight into what characterises informal workers, and the circumstances under which associations may ensure social protection for their members. The analysis shows that members of informal worker associations are significantly more likely to participate in formal insurance schemes compared with non-members, albeit with some variation across location, sector, and worker types. These divergences relate partly to sector-specific and institutional constellations, as well as the ways in which informal worker associations function. Moreover, the article shows that informal worker associations often play a dual role by providing both direct short-term social cushioning and enabling enrolment in formal social insurance schemes. Their part in ensuring the achievement of universal social protection must therefore not be underestimated.


Introduction
On a global scale, the informal economy provides a living for more than 61 pct. of the world's employed population (United Nations Development Programme, 2021). In sub-Saharan Africa (SSA), informal employment represents 86 pct. and 92 pct. of men and women's total employment, respectively (International Labour Office [ILO], 2018). Despite their contribution to the economy, informal workers often face discrimination, exploitation, and uncertain livelihoods; this is combined with an absence of, or at best inadequate, social protection. Since the turn of the century, however, there has been a growing recognition of the broader development potential of social protection, and it is increasingly considered to be at the 'forefront of the development agenda' (Hickey, Lavers, Niño-Zaraz ua, & Seekings, 2020). Since around 2010 several countries in SSA have witnessed an expansion of contributory health insurance and pension The remainder of the article is organised as follows: section 2 provides a brief overview of some of the challenges in extending social protection to informal workers in Kenya and Tanzania and the role played by associations in this endeavour. Section 3 introduces the theoretical framework on power resources and the alternative infrastructures of inclusion approach. Section 4 introduces the data and methods used, followed by the analysis and discussion in sections 5 and 6, respectively. Finally, section 7 concludes.

Extending formal social insurance to informal workers in Kenya and Tanzania
In both Kenya and Tanzania, the provision of social protection has traditionally taken many different forms characterised by fragmented interventions, due in part to the various actors involved. More integrated approaches are nevertheless emerging, including universal non-contributory pension schemes in both Kenya and Zanzibar (but not mainland Tanzania). 3 However, there is still little recognition of the broader transformative role of social protection (Adesina, 2011), and the limited progress made to extend health insurance to informal workers was further set back by . However, as shown by the data in this article, in both countries, the National Health Insurance Fund (NHIF) has been more successful in engaging informal workers than the National Social Security Fund (NSSF), in part because the former covers the household whilst the latter only covers the individual. Overall, uptake of informal workers in the schemes remains limited due to the perceived mismatch between the contributions and benefits received, long eligibility periods, lack of information, and registration problems, whilst inconsistency in contributions (once a member) relates to factors like premium cost/irregularity of earnings a result of which members become inactive or discontinue membership altogether (Indimuli et al. 2023). Thus, making the voluntary NHIF contributions mandatory as is being discussed in Kenya, would likely not succeed in retaining members.
Given that the attainment of social insurance is traditionally linked to institutions and workers in formal sector employment, it is largely inadequate to the needs of informal economy workers. Recent studies challenge the idea that social protection is associated only with formalised employment including Alfers et al. (2022), Awortwi and Walter-Drop (2018), and du Toit and Neves (2014). Focussing on the case of Kenya and Tanzania, Riisgaard et al. (2021) add to this literature tracing and comparing the organising histories of informal workers and providing in-depth discussions of the plurality of alternative forms of social protection provided by associations, including how these compare with the more 'top-down' formal social protection schemes. The authors find that although associations have different frameworks for offering informal social protection, with variation across sectors, in general, associations remain the bedrock of informal workers' social protection in both countries. Moreover, much of the welfare cushioning falls on preventive social insurance ranging from one-off responses to specific problems to broader concerns. Almost all associations support members during sickness and death, with a few others intermittently supporting maternity and unemployment. Thus, although such support is not comprehensive, Riisgaard et al. (2021) demonstrate the immense importance of informal workers' collective agency in meeting (even if inadequately) the various social protection needs of association members. Recognising this could help inform efforts to reframe the more 'top-down' national social protection policies and systems, which often fail to correspond to the reality of informal workers.
In both Kenya and Tanzania, there are examples of formal schemes engaging informal workers through their associations to better reach informal workers and reduce defaulting. For instance, in the context of extending health insurance to informal micro-traders in Kenya, the National Alliance of Street Vendors and Informal Traders (KENASVIT) has undergone repeated interactions with the NHIF and NSSF. Notwithstanding the representative power of KENASVIT such engagements could also be interpreted as reflecting elements of capture by the authorities to discipline rather than support the needs of informal actors. This will be discussed in subsequent sections. In general, Kenya has a longer track-record in handling informality, and thus formal institutional linkages between informal workers and the government (for example, the Micro Small Enterprise Authority (MSEA) and transport Savings and Credit Cooperative Societies (SACCOs)) are more established, compared to Tanzania where such experience is limited to ad hoc coordination and consultation. In 2015, the Tanzanian NHIF introduced the group-based insurance scheme KIKOA, extending informal worker access by allowing members (of registered associations with a minimum of 20 members), to sign up their spouse and four legal dependents under 21 years, at a reasonable cost. Although the initiative seemed well-suited to informal workers, it presented some challenges as we shall see in section 5. Subsequently KIKOA was phased out ultimo 2019 and revamped into the improved community health fund (CHF). 4 Currently, the government is planning to establish a single national health insurance fund to replace the NHIF and CHF.

Theoretical considerations: power resources and infrastructures of inclusion
This article draws on the power resources approach (PRA) to understand the circumstances under which associations are able to link workers to social insurance schemes, whilst the infrastructures of inclusion framing allows for an alternative interrogation of whether the extension of social insurance to informal workers harnesses social networks to engage with informal workers on adverse terms. This section explores these two different theoretical frameworks.

The PRA application to informal workers
The PRA concept created by Wright (2000) and Silver (2003) was founded on the basic premise that organised labour can successfully defend its interests by collective mobilisation of power: structural power as the power stemming from labour's position in the economic system (which tends to be limited in the case of informal workers), and associational power arising from collective political or trade union worker associations. Different scholars have added conceptual innovations including symbolic power, logistical power, institutional power, and societal power, leading to a vast body of divergent varieties of the PRA that explore how these sources of power are interconnected. 5 Whilst elements of different power resources may be present amongst informal workers, structural power tends to be limited. In relation to informal workers, associational power derives from different types of collective organisations, institutions, and alliances created or engaged in by workers to advance their own interests and conditions (Agarwala, 2013;Mosoetsa, Stillerman, & Tilly, 2016).
For instance, in the case of informal transport workers in Dar es Salaam, Rizzo (2017) shows that associations form as collective attempts to address the everyday challenges of dala-dala workers, including social cushioning and pooling resources to buy vehicles. Despite the persistence of the necessity for 'agency' and 'voice' to bring workers together, Rizzo also shows the limitations and vulnerability of worker organisations to assert robust collective action in lobbying for rights and representation.
In Dar es Salaam's informal construction sector, Wells and Jason (2010) find that the driving forces and motivations behind the formation of the umbrella organisation Tanzanian Association of Informal Construction Workers included the urgent need to increase opportunities for employment and income, as well as attaining legal recognition, empowerment, and facilitating access to markets.
When it comes to informal micro-traders in Kenya, KENASVIT, has been successful in organising and empowering street vendors to improve their business through training, access to credit, and dialogue with local authorities through representation on the MSEA board. Tanzania does not have such formal institutional linkages between informal workers and the government, although intermittent support for the informal sector does exist.
Informal worker access to formal social protection in Kenya and Tanzania 1573 Such examples provide evidence of informal workers' collective associational power resulting in the attainment of especially promotive social protection measures to manage the vulnerability of their employment, yet less so the emergence of transformative social protection that seeks to challenge or reform this precariousness. This highlights the limitations of informal collective action.

Precarious economic inclusion
An analytical concept stemming from several disciplines within social science, infrastructures of inclusion may be defined as 'structured arrangements of actors, institutions, technologies and power relations that link precarious populations into formal economic systems and circuits of capital' (Meagher, 2021, p. 731). The mechanisms through which this 'economic inclusion' is realised include global value chains, micro-finance, and mobile money for the 'unbanked', as well as digital connections to gig work. Seen through this lens, engaging with informal workers constitutes an extractive project to 'mine the "fortune at the bottom of the pyramid"' (Roy, 2010, p. 5). Likewise, scholars have shown that progressive initiatives to advance social protection in informal settings promote extractive forms of governance, and that 'the social networks of the poor are not only harnessed as an instrument for development; they are also the infrastructure that makes informal economies usable for capital' (Dolan & Roll, 2013, p. 139).
Such deliberation about the means of incorporating informal workers into development processes, on terms that are not necessarily beneficial to them, has sparked intense debates about the limitations of inclusive infrastructures and the economic and ideological interests around which they are configured (for example, Alfers, 2020; Ouma & Ad es ın a, 2019). It has been posited that 'it is not clear if inclusive calls for a social protection floor, corporate engagement with informal workers, and a new social contract represent a means of transforming informality, or making it easier to exploit' (Meagher, 2021, p. 738).
The potential instrumentalisation of associations by governments and other actors calls for a reflection of the potential downsides of powerful social insurance providers in terms of prioritising market share and resource flows over effective benefits. Being in a group reduces defaulting due to the social pressure to contribute; yet this is also an involuntary aspect of 'inclusive' but rather top-down infrastructures. Studies looking at the 'dark side of social capital' find that the 'pressure to share' often has negative implications (Di Falco, Lokina, Martinsson, & Pin, 2019) and may even lead to individuals evading their 'sharing obligations' (Di Falco & Bulte, 2011).
Evidence of disadvantageous or even disempowering aspects of inclusion does emerge in this article, for instance the case of the KIKOA scheme. However further in-depth investigation of the relation between the relevant formal actors (public and private) and informal associations is necessary to draw more specific conclusions. Since the focus of this article is to examine the link between association membership and formal social insurance the primary object of empirical investigation is the informal worker to which we turn next.

Data and method
The article is based on survey and interview data with informal workers carried out during 2018-2019 in Kenya and Tanzania as part of a larger collaborative project. 6 The data were collected for workers engaged in construction, micro-trade, and transportsectors that are highly prone to informality and have a high concentration of workers in urban areas. As such, data were sampled from two urban centres in each country: Nairobi and Kisumu (Kenya), and Dar es Salaam and Dodoma (Tanzania). Comparing a larger urban area (main city) with a smaller city in each country, allowed for observing differences which relate to the degree of urbanisation. In each location three zones were identified through transport hubs, which were used to sample workers across the three sectors. The survey covered wageworkers, own-account workers, and micro-businesses with a maximum of two employees. Whilst 75 pct. of workers were sampled randomly, 25 pct. of workers were sampled through associations identified purposively for the project to ensure a broad representation of different types of associations. The total sample (random and purpose-based) consisted of 1462 workers, which after data cleaning led to a final sample of 1385 observations (644 in Kenya and 741 in Tanzania). 7 The interviews include key informant interviews (KII) and focus group discussions (FGD) with different association members. Table 1 provides selected summary statistics showing that 41 pct. of workers belong to an association; the share is higher in Kenya at 48 pct. compared with 34 pct. in Tanzania. The incidence of association membership also varies substantially across the sectors and countries. 8 Table 1 indicates that 26 pct. of workers contribute to a social insurance scheme and the share is almost double in Kenya (34 pct.) compared with Tanzania (18 pct.). Across both countries, most workers referred to health insurance specifically: out of the 358 workers that answered positively to contributing to some type of social insurance scheme, 78 pct. reported health insurance, 16 pct. pension fund, and 7 pct. other insurance. The low share for pensions is expected given the aforementioned limitations, with many workers being inactive NSSF members. Since health insurance extends to immediate family the coverage figure is slightly higher at Source: Author's elaboration based on the survey data. a Association member figures are from the random sample due to the association membership bias in the full sample. b Median earnings are USD 7.6 for the full sample, USD 6.5 for Tanzania, and USD 8.7 for Kenya. Tanzania) and sector. In Kenya, for those that do contribute to health insurance, almost 80 pct. of workers contribute directly to NHIF and 20 pct. through an association, compared with 53 pct. and 47 pct. in Tanzania, due to the KIKOA scheme. Dodoma represents a slightly larger share of the sample followed by Nairobi, Dar es Salaam, and Kisumu. The variable 'local' indicates that in Kenya 26 pct. of workers were born in the city where they work, whilst for Tanzania the share is 46 pct., suggesting a lower incidence of internal migration in the latter. As for gender, 77 pct. of the sample are men, and the mean age is 35 years slightly higher in Kenya. Regarding education, 53 pct. have completed primary education and 37 pct. have secondary schooling and above, with the latter share being substantially higher in Kenya. On average, 25 pct. of workers have had on-the-job training, and 22 pct. have attended a professional training school, both shares higher in Kenya. The share of workers that are selftaught is higher in Tanzania at 64 pct., compared with 41 pct. in Kenya. In terms of worker types, own-account workers represent the vast majority with 52 pct. on average and as high as 64 pct. in Tanzania, whereas in Kenya wageworkers make up the largest share with 55 pct., whilst 38 pct. are own-account workers. Micro-businesses constitute only around 11 pct. of the sample. The sector distribution is relatively even, although with a higher share of trade and transport workers in Tanzania and of construction workers in Kenya. Finally, mean earnings are similar across the two countries, yet median earnings are substantially higher in Kenya. 9 Among the 800 workers that are association members there is variation in the types of associations, with more than 50 pct. report belonging to a SACCO/vicoba/chama (SVC). This is a rather broad category encapsulating a variety of associations that workers in the same sector may belong to, ranging from the more formally established SACCOs to informal welfare groups known as chamas. As with social insurance enrolment, there is also substantial variation by country, with the incidence of SVC being significantly higher in Kenya at 64 pct. compared with 39 pct. in Tanzania. The second categorywork-related associations (WRA)includes trade-specific and business or work colleague/gang type associations and has an overall incidence of 44 pct., with a higher share in Tanzania at 58 pct. compared with 29 pct. in Kenya. Around 5 pct. of workers belong to either a woman, youth, or other type of association. 10 Through the power of collectiveness and enhanced information associations may facilitate informal worker enrolment in formal schemes or alternatively act as vehicles to promote participation in the name of economic inclusion. However, associations themselves may also provide the kind of social protection that is more suited to the needs of their members, who in turn have less to gain from joining formal insurance schemes. We turn to this in the following sections.

Linking informal workers to formal social insurance: fostering worker power or adverse inclusion?
In order to examine the relation between association membership and social insurance enrolment a standard probit model is used where social insurance is the outcome (dependent) variable taking a binary form and the same for association membership (the main independent variable of interest). Table 2 column 1 shows that, as expected, members are 26 pct. more likely to contribute to social insurance compared with non-members, when key worker characteristics, location and worker types are controlled for. In column 2, sectors are added, which reduces the association membership variable to 20 pct., as some explanatory power is captured by the transport workers category. In column 3, the membership variable is split into the different types of associations (SVCs, WRA, and others) and the SVCs grouping is driving the result with members of these types of associations being 29 pct. more likely to have access to social insurance compared with non-members. Since one of the key functions of SACCOs is loan provision, either directly or indirectly via establishing agreements with banks, one interpretation could be that members use the loans to pay for social insurance contributions. Yet, that would suggest social insurance participation also being determined by earnings or assets, which Table 2 shows is not the case as neither is significant when other factors are considered. 11 The relation between SVC and social insurance access therefore seems to be a result of the associational setup or purpose, suggesting the presence of associational power. Among those that are members of SVCs, a substantial share (65 pct.) report receiving loans as the main benefit, thus these kinds of associations are likely to serve several purposes including financial cushioning for their members (promotive social protection) and facilitating access to social insurance (preventative social protection).
In terms of the remaining control variables, married and more educated workers are more likely to contribute to social insurance. This is likely due to married workers having dependents Informal worker access to formal social protection in Kenya and Tanzania 1577 whilst workers with higher education tend to be better informed and more aware of the importance and existence of formal schemes. In terms of location, workers in Nairobi, Kisumu, and Dodoma are generally more likely to be part of a social insurance scheme, compared to those based in Dar es Salaam. Participation in social insurance is generally more widespread in Kenya compared to Tanzania, yet Dodoma workers have access to the more affordable CHF option, which explains why enrolment there is higher than in Dar es Salaam. Moreover, across locations, individuals that are working in the city in which they were born (local) have a higher chance of social insurance enrolment, which could be related to them having more knowledge of available schemes. As mentioned, transport sector workers are significantly more likely to be social insurance up-takers compared with both construction and micro-traders; this holds regardless of membership status revealing substantial sector variation.
Finally, both micro-businesses and wageworkers are more likely to have social insurance compared with own-account workers, independent of membership. However, most micro-businesses do have association membership and that group is more likely to have formal social insurance compared to workers without associational belonging. 12 Indeed, associations play an important role in terms of registering members and in some cases also preventing default, as explained by a member being asked about the relevance of groups in social insurance schemes: All members of my group are registered for NHIF and are contributing at least KES 200 every month through the group. If a member is unable to pay, the group takes the money from the members' savings and pays for him or her. The member who is being paid for is considered to have a debt which he or she must pay. (FGD micro-traders, Kisumu, 2018) In this case, the group draws on worker savings to pay the remaining part of the KES 500 individual NHIF contribution. Among Tanzanian micro-traders, KIKOA membership was also organised through associations: The management of the association already introduced all their members to NHIF but every member should pay TSZ 78,000 to get the member card for the four member of their family that's what they did in their association. (FGD micro-traders, Dar es Salaam, 2019) In other cases, association members found it challenging to enrol as a group: It was a challenge to contribute consistently. Now we don't contribute to NHIF as a group. Each member contributes alone. Members who have discontinued depend on the group for assistance. (FGD micro-traders, Kisumu, 2018) Similarly, the Nairobi Informal Sector Confederation (NISCOF), an umbrella association for Kenyan micro-traders also made efforts to register their members in the NHIF scheme but experienced a challenge paying premium as a group and opted for each member to make their payment individually, which seems to be the most common practice overall.
Other associations share information on NHIF schemes with their members and encourage them to join in smaller groups which reduces the quarantine period from 60 to 30 days. These examples of NHIF engagement attest to the collective capacity among informal micro-traders, ensuring that member contributions are being channelled into premiums payments with certain group advantages like reduced quarantine periods. However, evidence of individual savings covering default (implying a personal debt for the worker) and/or cases where group contributions failed (like KIKOA) suggest an alternative interpretation of associations serving as instruments to extract value from informal workers, disguised as support or empowerment. Moreover, workers express feeling excluded when schemes did not provide adequate services: The difference between KIKOA insurance and NHIF insurance is that there are some services during the first year that as a member of KIKOA insurance you won't get them that thing make us feel like we have been excluded, the second year some services try to be inclusive. On the third year there are some members who have been directly affected they were told that there's a bill been sent to the parliament which means KIKOA has been removed from providing insurance services. (FGD micro-traders, Dar es Salaam, 2019) In fact, several workers told stories of KIKOA not providing services nor refunding money, despite the packages being tailored to groups of informal workers.
Turning to wageworkers, the link between associational belonging and access to social insurance is particularly strong in Tanzania, and mostly among transport workers 13more specifically dala-dala operators. Interesting although initially being considered as employees by the NHIF, the dala-dala were granted special permission by the government to enrol in the KIKOA scheme, intended for own-account workers. This could also be interpreted as NHIF attempting to extract capital from as many workers as possible.
Ironically, however, boda-boda 14 who are own-account workers seem to rely primarily on associations to provide social cushioning and support in case of injuries: For example, if we don't help each other, then we cannot work and thus fail to get income. Even if we contribute, we cannot contribute all the amount needed for the person, but the amount helps. It also depends on the extent of the injury. If its very serious, then we ask other boda-boda to assist from other stands. (FGD transport, Dar es Salaam, 2019) The transport sector in Kenya is different from in Tanzania, with Kenyan wageworkers in that sector being significantly (54 pct.) more likely to have social insurance compared with other sectors, regardless of them being association members ( Table 9 in Supplementary Material). This is because, since 2012, matatu (mini-bus) operators are required by law to join SACCOs. 15 These SACCOs operate as quasi-legal entities registered by the State Department for Cooperatives and are expected to manage registered vehicles and clear workers for licencing by the National Transport and Safety Authority (NTSA), issuing them with a Certificate of Good Conduct by the Criminal Investigation Department. Furthermore, the NTSA Act (2014) requires SACCOs to employ workers on formal work contracts (although many workers do not have such), provide them with leave days, and make statutory contributions to health and pension schemes. Transport SACCOs may therefore be seen more as employers who largely favour the interest of vehicle owners over worker concerns, often leading to tensions, as elaborated in section 6. In other words, even though the relation between owners and drivers (workers) resembles an established employer-employee relation, it is often void of the social protection provisions found in a legally binding contract. Yet, in some cases, SACCOs make daily deductions from workers' wages and remit monthly to NHIF on behalf of workers, whilst others encourage workers to enrol voluntarily: We are trying a new strategy of ensuring that our employees have medical cover. We have invited Britam Insurance to speak to them on the importance of these covers including the NHIF. (KII, MOWUOK Transport Company Manager, Kisumu, 2018) However, many matatu workers are reluctant to join employer-linked social insurance due to fear that their job mobility might be restricted. Moreover, most transport workers belong to chamas that cater for their welfare needs, including bereavement, sickness, and school fees support. Such support is often limited and may not cover full costs, yet the informal social cushioning provided by the association is more flexible and tailored.
In contrast to matatu workers, own-account boda-boda drivers are not required to belong to associations to be licenced, yet many of them voluntarily belong to informal groups and selfgoverned motorcycle associations. These associations, such as the Boda-boda Safety Association of Kenya (BAK), are characterised by strong collective capacity operating with the specific purpose of furthering workers' interests (and not simply those of owners/employers), including the provision of both formal and informal social protection: In this case NHIF premiums are paid individually, providing little indication that BAK has been 'harnessed by the formal system' to extract assets from its members. Rather by having its own voluntary contributory scheme providing ad hoc informal benefits, the association seems to rely on and foster worker power. Regardless of the interpretation, the data shows that in Kenya there is a strong positive relation between membership and social insurance for boda-boda drivers, and that represents the significance of the membership variable in Table 2. 16 The preceding analysis reveals that the relationship between association membership and social insurance participation is more nuanced when looking at different worker types across sectors and country-specific contexts. Although not reflecting the traditional understanding of associational power as conceptualised in terms of representing workers vis-a-vis the authorities or pressuring employers into providing social protection, the examples presented demonstrate how associational belonging can further worker access to preventive social protection. Ironically, it seems that associations are most successful in linking workers to social insurance, when contributions are individual as opposed to group-based. However, whether the nature of this link to the formal sphere is dominated more by associations encouraging the protection of worker interests or enforcement from above through infrastructures of inclusion likely varies by case. In Tanzania, the KIKOA scheme seemed to represent an example of strong collective power, at least in the negotiation phase of the scheme. However, the motivations behind the scheme have been debated and like other progressive initiatives to advance social protection and sustainable services for the poor, the scheme could also be interpreted as constituting a 'bridge for more extractive forms of governance driven by the interests of digital, financial and venture capital' (Meagher, 2021, p. 747). This would also align with the fact that adverse selection led to its demise and thus KIKOA may present an example of a failed social protection infrastructure reinforcing rather than addressing inequality.

Discussion: top-down versus bottom-up social protection
Kenya has a higher incidence of social insurance coverage for informal worker compared to Tanzania, associational attachment is also deemed stronger, and a substantial number of workers have enrolled in social insurance through their associations. Yet, in both countries and across locations, a significant coverage gap remains, despite the need for especially health insurance, and this deficit is linked to several factors including cost and lacking information: We cannot afford to pay KES 500; the NHIF failed to discuss with the public when they increased the amount to KES 500. (FGD micro-traders, Nairobi, 2018) I was a member when NHIF payment used to be 160. I stopped paying. (FGD microtraders, Kisumu, 2018) Because my income is not enough to cover up for health insurance [TZS30,000 per month], I also have never been sensitized about health insurance and its importance. I just hear that it really is helpful when a person is sick but I do not know anything beyond that. (FGD micro-traders, Dodoma, 2019) We would really like to join health insurance funds but we do not know where to start from, we do not know what the procedures are, personally I would really like to join because most of the people say it really is helpful to them. (FGD micro-traders, Dodoma, 2019) There are a lot of challenges we face in our businesses but I think the social security funds and health insurance are the major issues. So I would like to urge organizations such as NSSF and NHIF to accept small entrepreneurs as members with minimal obstacles and criteria. (FGD micro-traders, Dodoma, 2019) The above quotes come from micro-traders yet among construction workers the incidence of social insurance participation is even more limited despite most construction workers considering themselves to be wageworkers, which in theory means that they should be in a contractbased relationship bound by legal provisions. However, construction employment relations are often characterised by being non-transparent and casual in nature and, unlike the formal sector, employer deducted social insurance contributions are rarely observed in practice in the informal sector. In general, construction workers do not perceive social insurance as a right and, given the lack of labour inspections on construction sites, workers are easily exploited. Some associations encourage members to register for the NHIF, yet the difficulty of having group insurance is often mentioned, again with cost being a constraint, as indicated by members of different construction groups: People are different. Not all members of the group have NHIF. There are those that are being pushed to have insurance. We live from hand-to-mouth and it also makes it hard for us to pay the NHIF. (FGD construction, Nairobi, 2019) I can go for a week without a job. It becomes hard for me to accumulate KES 500 at the end of the month. It is easier for the formal workers because theirs is deducted from monthly salary. (FGD construction, Nairobi, 2019) However, cost is not the only deterrent since some workers pay more in association fees and ad hoc contributions than they would pay for subscription to social insurance, 17 indicating that it is more about willingness than ability to pay. In other words when weighing up the pros and cons workers generally find that associations bring more benefits than formal social insurance schemes. Fluctuating earnings seem to matter more than the level, which is in line with Table 2 and Indimuli et al. (2023). Moreover, cumbersome processes and inadequate services are also frequently mentioned as deterrents: If you are sick and you want approval from NHIF, the process is hard. They take you to and from. For MRI clearance you are sent to headquarters (at Upperhill) to get approval. (FGD micro-traders, Nairobi, 2018) There is no client service in public hospitals. They mistreat us. Nurses are very bad. The services are poor. The machines are sometimes not working. However, the private hospitals (where we pay cash) treat us very well. (FGD micro-traders, Nairobi, 2018) Finally, penalty fees upon default also dissuade workers from joining the formal schemes, as expressed by a construction group: We formed a group to take care of health insurance especially sickness. I have seen my friends being fined for default and I have lost hope in the cover. I now go to mission hospital. As a group we have emergency fee for accidents occurring on-site. Our Constitution clearly states that the nuclear family will be covered in case of sickness and death, we are building our own insurance … we assist members until they get well. (FGD construction, Nairobi 2019) The above testifies to the importance of informal cushioning measures which most associations provide suited to worker needs, as also indicated by a group of women micro-traders: We contribute towards welfare. If you are sick or have any problem such as funeral, lacking school fees. The group discusses and agrees to give you some money. (FGD micro-traders, Kisumu, 2019) Thus, although social insurance schemes have more comprehensive covers, there is often a preference for the more flexible informal social protection that associations provide, including Informal worker access to formal social protection in Kenya and Tanzania 1581 on-demand welfare cushioning as well as revolving savings and credit schemes enabling easy borrowing for members: The group has a credit facility for its membership but based on shared contribution. 'We are giving shares, if any worker wants to buy. The association can give one money and pay back'. (KII construction, Kisumu, 2018) Such examples of collective resource pooling for purposes of mutual support arguably qualifies as associational power. When understood in this broader sense, associational power may also be said to include aspects like providing information on job opportunities, negotiating for terms of employment, and mediating disagreements between workers and employers. These functions resemble what trade unions do in a formal employment setting and, along with welfare issues, savings, and credit (promotive social protection), constitute some of the key association benefits amongst informal workers in Kenya.
In both countries, associational weakness is related to members defaulting on association payments and often not being able to participate in scheduled meetings due to long hours of work and movement to different sites. Collective capacity is related to membership density, yet the degree of associational power is not solely related to numbers since among the matatu workers there is a high membership rate, yet associational belonging is considered weak. This is likely a result of workers not coming together voluntarily but rather being legally required to belong to SACCOs to be licenced. These associations are regulated by the government and mandated to negotiate with authorities on various matters affecting the sector, including worker registration, access to jobs, and social insurance contributions, even if this is not always the case as indicated by a group of both matatu drivers and boda-boda riders: The SACCO was supposed to get for us the TLB, NHIF, and NSSF. When the NHIF was KSH 160 they used to pay but when it was increased to Ksh 500 they stopped paying. According to the BIG 4 Agenda 18 the government should generate that money through increased tax and this will guarantee free treatment. (FGD transport, Nairobi, 2018) The above is a good example of formal economy actors, including the state, rendering informal networks and organisational systems legible to facilitate access to informal economic spaces (Dolan & Roll, 2013;Elyachar, 2010;Roy, 2010). Moreover, most of the issues addressed by SACCOs are often not related to working conditions or rights to social protection, but rather to factors impeding a smooth operation of the fleet. Since these associations mostly favour the vehicle owners' interests rather than guaranteeing workers' rights, sub-groups often form with the purpose of informal social protection provision (including savings and credit). Thus, in general, these associations are not considered a means for workers' associational power in terms of voicing their concerns or demanding better working conditions, yet some of the well-functioning SACCOs do invest in workers' capacity building by organising training as expressed by the chairman of a boda-boda association: We want the members to be fully qualified riders by undergoing training and getting insurance. (KII transport, Kisumu, 2018) and facilitating access to loans as stated by a group of both matatu drivers and conductors: Some of the SACCOs allow the employees to save with them and receive loans three times their savings. The driver and the conductor save half their salary daily. At the end of the month they pay for NHIF as required by NTSA while some pay even for NSSF. (FGD transport, Nairobi, 2018) Their role in furthering promotive social protection measures combined with the fact that the incidence of SVCs as a group is substantially higher in Kenya also helps explain why there is a significant difference between the earnings of association members versus non-members in Kenya, whilst in Tanzania earnings are not related to membership. 19 In Tanzania, however, WRA and welfare-focused chama associations are more common than loan/training-providing SACCOs, pointing to differences between the two countries in terms of the reasons for workers forming into associations and the benefits received. Overall, the Kenyan informal transport sector is more organised than the equivalent sector in Tanzania which is characterised by more informal cooperation. In general, Tanzanian associations, especially in construction, are generally less established, smaller, and have not been as active in terms of facilitating members to join social insurance schemes.
Common across countries and sectors is that associations have yet to venture beyond measures that fall under the preventive and promotive types of social protection, as a relatively small share of workers report voice and representation (transformative social protection) as a key association benefit. 20 However, associations may still act as potential mechanisms for workers to voice their concern. For instance, across sectors in Tanzania, 43 pct. of association members report having contacted their associations on issues related to representation, of which 40 pct. claimed it helped resolve the issue. Similarly, in Kenya, 30 pct. of workers across sectors made contact on issues related to representation, of which 30 pct. stated it helped resolve the issue. Thus, even if not cited as a main benefit, in some cases associations play an important representative role. This seems to be the case especially for workers in the capital cities as workers in the second cities of Kisumu and Dodoma list voice as being a less important associational benefit. It may also be related to larger umbrella associations like KENASVITthe key purpose of which is voice and representationbeing more present in Nairobi: My role in the board is to advise the government on issues pertaining to MSE and to 'encourage Kenyans to be entrepreneurs'. (KII, KENASVIT Chairman, Nairobi, 2018) Here, the interviewee is referring to the board of MSEA, which includes an informal trader representative. In general, being more organised with a solid leadership structure and a larger membership base, umbrella associations like KENASVIT command greater associational power, not least on the representation (transformative) dimension, whereas the smaller standalone (grassroot) associations, which constitute the majority, focus more on inward-looking welfare and work-related issues (preventive and promotive types of SP). In both countries the latter constitute most associations outside the main city. In Tanzania, VIBINDOthe umbrella association for micro-tradershas also demonstrated its capacity to represent member interests by facilitating access to the political and administrative system: Sometimes if you have a problem or any kind of challenge with the municipal the leaders they [VIBINDO] can go there to present us … If you face the big problem you can call the VIBINDO, or come to the office to tell them this is too big for us please support us. (KII micro-traders, Dar es Salaam, 2018) VIBINDO also has contact with the national authorities and was the only informal economy counterpart to NHIF when the KIKOA scheme was negotiated, testifying to their ability to leverage institutional power in addition to their associational strength.

Conclusion
This article has examined the enabling or disabling circumstances under which informal worker associations facilitate informal worker access to formal social insurance. Overall, the results show that association members are significantly more likely to be enrolled in social insurance and this relation is particularly strong among certain sectors and worker types. Whether this informal-formal link is an indication of 'bottom-up' associational power or 'top-down economic inclusion' efforts to discipline and exploit the assets of informal workers in the name of empowerment appears to depend on the specific country/sector case.
Regardless of the interpretation, there are numerous challenges, both conceptually and practically, when it comes to engaging informal workers in social insurance schemes. Creating more 'inclusive' infrastructures is unlikely to be an easy panacea in contexts where informal workers Informal worker access to formal social protection in Kenya and Tanzania 1583 traditionally operate in highly organised and self-governed structures. In both Kenya and Tanzania, the concept and basis of social protection can be traced back to before the colonial era where informal protection measures were practiced at the family and community levels. Thus, from a social protection perspective, associations are thus likely to remain important, as their benefits are tailored to the specific needs of informal workers, and their flexible set-up allows ad hoc access to contributions in the form of savings. Moreover, the shortcomings of the formal schemes were also revealed during COVID-19. For instance, in Kenya, where the pandemic led to a partial lockdown, substantial job losses in the formal sector meant that workers defaulted on their regular contributions to social insurance schemes, in turn hindering their ability to cushion members. In the informal sector reduced earnings meant disruptions in association contributions in turn affecting welfare service provisions, yet many associations continued to issue loans to members, even at reduced interest rates.
In Tanzania, the existence of COVID-19 was largely denied, and the formal segment of the population continued working, hence mandated social insurance contributions were unaffected. However, temporary curfews and seat restrictions adversely affected transport workers and micro-traders, meaning association contributions dropped, and welfare services stopped to cater for an increasing number of loan requests. By contrast, construction associations suspended the provision of loans to continue providing welfare support to their members. Overall, such coping mechanisms demonstrate the flexibility of associations and thereby their importance in reducing the exposure of informal workers, especially in times of crisis.