Economic and health indicators in Nigeria

The year 2020 was challenging for peoples and governments all over the world, but in few places has the pandemic put such a strain on the domestic social contract as in Nigeria. In 2020, GDP contracted by 3–4 percent, recovering only slightly to a GDP growth of 1.5 percent in 2021 (IMF 2020). At the same time, the COVID-related oil price crash drastically diminished Nigeria’s export earnings and, as a result, its available public revenue at both the central and state levels (Chinery, Sayne, and Gillies Reference Chinery, Sayne and Gillies2020).

Because of this economic and financial crisis, the government requested, and received, emergency financial assistance of USD 3.4 billion from the IMF in April 2020 (IMF 2020). This marked the beginning of several loans that the government considered as needed to contain the fall out of the crisis as COVID had plunged Nigeria in the worst recession recession since the 1980s (Lain et al. Reference Lain, Vishwanath, Alik-Lagrange, Amankwah, Contreras-Gonzalez, Jenq, Mcgee, Oseni, Palacios-Lopez and Sagesaka2021). It was estimated that a total of USD 11 billion in loans could increase Nigeria’s debt stock by 50% (Chinery, Sayne, and Gillies Reference Chinery, Sayne and Gillies2020). This in turn emphasizes the importance of innovating the social contract with informal workers, especially given the uncertainties in developing country oil exports in the coming decades. While the Nigerian economy did recover due to an increase in the international oil price in 2021 and beyond, the 2020 crisis laid bare the immense consequences Nigeria will be confronted with when its competitiveness in the oil market eventually reduces.

In parallel to economic turmoil, Nigeria is falling behind the United Nations’s Sustainable Development Goal 3.8 (SDGs, a collection of interlinked objectives adopted as a universal call to action to end poverty) to universal health coverage, including financial risk protection, access to quality essential healthcare series and access to safe, effective, quality and affordable essential medicines and vaccines for all. This comes at a time that the WHO has called for low-income countries to spend 1% of GDP more on primary health care (United Nations 2019). This however is difficult for developing countries with an uncertain revenue, which is the case of low-income countries that are disproportionally reliant on external oil revenues. In 2005, Nigeria started the National Health Insurance Scheme (NHIS) with the goal of improving access and limiting financial exposure to unexpected healthcare needs. Despite the establishment of the NHIS, Nigeria is 142 out of 195 in terms of healthcare access and quality (Fullman et al., Reference Fullman, Yearwood, Abay, Abbafati, Abd-Allah, Abdela, Abdelalim, Abebe, Abebo and Aboyans2018), and the country spent USD 58 per person on primary healthcare in 2019 (WHO 2021). Less than 5% of Nigerians are insured with the NHIS and less than 3% with private health insurance (Aregbeshola and Khan Reference Aregbeshola and Khan2018). In the last available study on catastrophic health expenditure in Nigeria, it was found that 16.4% of households incurred catastrophic health expenditure at the 1% threshold of total consumption.

Like in other locations, the health insurance goal in Nigeria has been to protect insured against financial ruin because of unexpected out-of-pocket (OOP) expenditures. Nigeria has generally relied on voluntary premiums from informal households to expand health coverage, even if that has proven difficult (Lagomarsino et al. Reference Lagomarsino, Garabrant, Adyas, Muga and Otoo2012) and most Nigerian households by and large remain reliant on OOP expenditures to finance healthcare needs. This level of OOP spending is in fact larger than total government spending on health (Aregbeshola and Khan Reference Aregbeshola and Khan2018). OOP health expenditure was responsible for a 0.8% rise in poverty, which represents an additional 1.3 million Nigerians being pushed below the poverty line (Aregbeshola and Khan Reference Aregbeshola and Khan2018). The WHO found that 15% of the population has health expenditure higher than 10% of total household spending (WHO 2021). In a developmental context in general, and with the threat of COVID-19 specifically, healthcare is a priority for Nigeria.

Governance and healthcare spending are positively correlated with health coverage (Ranabhat et al. Reference Ranabhat, Jakovljevic, Dhimal and Kim2020). Both of those, however, are facing additional stress in developing countries as a result of the COVID-19 pandemic. This has been even more so the case in oil producing developing countries like Nigeria that are confronted with future uncertainties related to international oil markets and revenues. Besides price, COVID-19 related stimulus packages around the world appeared to be accelerating peak oil demand, which will only foster competition between oil producers in the future. It is expected that developing country producers in sub-Saharan Africa will be unable to compete with producers in the Persian Gulf and Russia (Moerenhout and Bellman Reference Moerenhout and Bellman2021).

Taxation of informal workers in commodity-dependent developing countries

In many commodity-dependent countries, including oil-exporting developing countries like Nigeria, broadening the tax net is challenging because a large percentage of the population works in the informal sector. Recent years have seen an increased interest in taxing the informal economy, especially in developing countries were the informal sector forms a large share of GDP and cash-strapped governments could benefit from taxing this potential large source of revenue (Schneider and Klinglmair Reference Schneider and Klinglmair2004). Besides Nigeria, many other countries in Sub-Saharan Africa that are commodity-dependent also reach 90% of employment in the informal sector (ILO 2018).

Many efforts so far have focused on encouraging informal businesses to register. Unfortunately, these efforts have by and large not yielded the hoped for results. This is mostly because of two factors. On the one hand, since informal workers and firms are so omnipresent, campaigns to formalize them are very expensive (Benhassine, McKenzie, Pouliquen and Santini Reference Benhassine, McKenzie, Pouliquen and Santini2018). On the other hand, informal workers are expecting certain benefits in exchange for formalization, which brings with it tax and other administrative costs (de Mel, McKenzie and Woodruff Reference de Mel, McKenzie and Woodruff2012; Fajnzylber, Maloney and Montes Rojas Reference Fajnzylber, Maloney and Montes Rojas2009a, Reference Fajnzylber, Maloney and Montes Rojas2009b; Loeprick Reference Loeprick2009; McCulloch, Schulze and Voss Reference McCulloch, Schulze and Voss2010; Perry, Maloney, Arias, Fajnzylber, Mason and Saavedra Reference Perry, Maloney, Arias, Fajnzylber, Mason and Saavedra2007).

This brings us to the core problem of broadening the tax net in commodity-dependent countries. When prices are high and revenues solid, there is no immediate incentive to broaden the tax net. In fact, it can be considered as a political risk because social contracts in most commodity-dependent countries are based on the exchange of political acquiescence and acceptance of the elite as manager of resources, for the distribution of benefits based on those resource rents. When prices are low, however, there is not enough revenue to run successful formalization campaigns and informal workers and businesses are less keen on joining because there is less consumption of their services as a result of fiscal deficits. The reduction of consumption during low price periods, and the challenge of translating resource rent in investment rather than consumption during high price periods is indeed at the core of the Dutch Disease. This is exactly what provides the rationale for tax-for-services programs in commodity-dependent countries: increase the appeal so informal workers themselves sign up.

Besides widening the tax net, and improving healthcare access, the goal of tax-for-services programs is to develop efficient taxation policies that can then also lead to greater accountability from the state, improved governance and state building (Brautigam, Fjeldstad, and Moore Reference Brautigam, Fjeldstad and Moore2008). In Ghana (dependent on revenues from oil, gold, and cocoa exports), for example, taxation has led to demands for greater accountability and has prompted political negotiation to sustain revenue raising (Prichard Reference Prichard2009; Prichard Reference Prichard2015).

To achieve universal health coverage by 2030,several countries have been discussing and experimenting with health financing schemes that can expand coverage from formal to informal households (Friebel et al. Reference Friebel, Josephson, Forman and Calza2020). Most healthcare financing systems now follow a hybrid approach, relying on income taxes, VAT and premiums (Friebel et al. Reference Friebel, Josephson, Forman and Calza2020). Paying income taxes in exchange for health coverage is at the heart of the suggested tax-for-health-services programs discussed in this article.

As Dube (Reference Dube2014) noted, successfully taxing the informal sector requires clearly outlining its objective, that is to either fund specific development goals, to encourage formalization or to simply raise revenue. And given the well-established resistance of informal workers to pay taxes given its perceived limited returns (Fjeldstad and Heggstad Reference Fjeldstad and Heggstad2012), it seems clear that broadening the tax base to informal workers would require programs to clearly link tax payment to service delivery, and to gain political and citizen support by understanding what citizens value the most (Fjeldstad Reference Fjeldstad2014). This consultative approach is highly encouraged to understand taxpayer needs and to build said social contract (Dube Reference Dube2014). Such trust-building initiatives have shown to be positively correlated with tax morale (Torgler Reference Torgler2011), itself reducing engagement in the informal economy (Williams, Horodnic, and Burkinshaw Reference Williams, Horodnic and Burkinshaw2016). However, in countries where tax morale and trust in the government is low, the choice of service to deliver is critical – it needs to be perceived as essential by the majority of citizens, to be relatively low-cost for the taxpayer and to be quickly delivered (i.e., infrastructures may take some time and would require significant funds). This in turn emphasizes the importance of innovating the social contract with informal workers, especially given the uncertainties in developing country oil exports in the coming decades. The goal is to thus considerably alter informal workers’ cost-benefit analysis of paying taxes. Access to healthcare would tick all the boxes, particularly in the Nigerian context.

Tax-for-Health-Services programs in a Nigerian context

Nigeria is a good case study for the potential of tax-for-health-services programs for several reasons. First, more than 90% of working-age citizens work in the informal sector. Excluding agriculture, this is still between 75% and 89% (ILO 2018). Second, Nigeria has struggled to raise funds for their healthcare system in a reliable and sustainable way. Third, currently a majority of Nigerians are dissatisfied with the provision of essential services, and this strongly reduces their tax morale. A 2018 tax perception survey conducted by the Nigerian Economic Summit Group (NESG) covering 10,000 households and over 5,000 small enterprises in Nigeria showed that a fifth of taxpayers believe it is not wrong to not pay taxes, while more than half believe it is wrong but understandable not to pay taxes. However, almost 65% of all Nigerians did agree that authorities have the right to make people pay taxes (McCulloch and Moerenhout Reference McCulloch and Moerenhout2019). Fourth, the idea is not pulled out of thin air. To improve healthcare access and expand the tax net, practitioners and politicians in Nigeria have considered a tax-for-health-services program, which aims to enhance the relationship between informal sector taxpayers and the tax authority by rewarding compliant taxpayers in the informal sector with health coverage. In particular, the Nigeria Governors’ Forum (NGF) has been spearheading this initiative and supporting governors in the design and implementation of the initial stage of the programs.

Tax earmarking to move beyond rentier social contracts

The evolution of the social contract has received a lot of interest in the literature. Loewe, Zintl, and Houdret (Reference Loewe, Zintl and Houdret2021) define the social contract as the entirety of explicit or implicit agreements between all relevant societal groups and the sovereign (i.e., the government or any other actor in power), defining their rights and obligations toward each other. Social pacts are considered as building blocks of the social contract which are more defined in scope and time (Loewe, Zintl and Houdret Reference Loewe, Zintl and Houdret2021).

In oil-exporting developing economies, many social pacts are under threat, effectively destabilizing the domestic political system in which redistribution paid for political acquiescence (Rougier Reference Rougier2016). Some of the most important social pacts such as the provision of public employment and food, energy and housing subsidies are unsustainable in low oil price periods, when the government has lower revenues from the sale of hydrocarbons. As a result, evolving the social contract beyond a rentier system by introducing taxation is very controversial, because it increases obligations on the side of the people all while unilaterally taking away some of the redistributive role of the state.

In comes tax earmarking, which consists of dedicating specific tax revenue for a particular expenditure purpose. It can guarantee funding for government priorities by ring-fencing it from competing political interests and bypassing budgetary processes (WHO 2019). Tax earmarking is not a panacea. Earmarked revenues are procyclical, increasing during times of economic expansion and decreasing during economic downturns (Mossialos et al. Reference Mossialos, Dixon, Figueras and Kutzin2002). Moreover, tax earmarking can create rigidities in the budget, potentially leading to inefficient or event ineffective allocation of resources as funded programs must spend all mobilized revenues without going into reserve, or may lead to underfunded program (Wilkinson Reference Wilkinson1994).

However, despite certain shortcomings, tax earmarking has been shown to build stronger trust among citizens by explicitly defining the benefits of taxation and establishing transparency between taxation and its public spending. It also provides an opportunity for taxpayers to engage with the state to communicate their preferences regarding public spending or to request accountability if funds were not adequately allocated, ultimately leading to better governance (Prichard Reference Prichard2010). A study in Bo, Sierra Leone (Jibao and Prichard Reference Jibao and Prichard2013), showed how communicating revenue and expenditures to their citizens has led to increase support for local taxation. Similar results were found in Guinea (Korsun and Meagher Reference Korsun and Meagher2018) where tax revenue doubled after tax was earmarked to the construction of new market stalls.

Explicitly allocating funds to tangible benefits such as the construction of a new road or access to specific services can lead to greater revenue mobilization, build trust in the government and their ability to provide public services and, ultimately, increase tax morale. While revenue mobilization remains the most appealing argument to introduce a new tax, citizens may benefit in the long-term from improved governance and accountability also (see the example of Ghana in Prichard (Reference Prichard2009) and Prichard (Reference Prichard2015), and the example of Mexico in Mossialos et al. (Reference Mossialos, Dixon, Figueras and Kutzin2002)). This eventually strengthens the social contract by enhancing both the rights and obligations on the side of the citizens (Ulriksen and Plagerson Reference Ulriksen and Plagerson2014).

Support for tax earmarking

This article’s first question is to what extent there would be support for tax earmarking programs among the informal sector in Nigeria. In most societies, members cluster in vertical or horizontal subgroups (Loewe, Zintl and Houdret Reference Loewe, Zintl and Houdret2021). Informal workers constitute such a horizontal group. We argue that informal workers are a key party to the social contract at large but often remain overlooked. This is mostly because in oil rentier states, as the theory suggests, the level of revenue collection that can come from a contract with informal workers may appear rather low, especially when oil prices are high, as they were during the recent two decades, apart from two strong price drops in 2014–2016 and 2020–2021.

While in parts the social contract is determined by implicit, tacit agreements, we doubt this would be possible for raising taxes on informal workers. Tax morale is low and the valuation of services even lower (McCulloch, Moerenhout and Yang Reference McCulloch, Moerenhout and Yang2011). Having government raise income taxes without an explicit agreement of what there will be in return could provoke political instability, exactly because there is no pact, and the overall social contract is weak. Moreover, both revenue administrators and politicians lack incentives to successfully implement these policies. While revenue authorities perceive taxing the informal sector as additional low-rewarding work with respect to meeting their annual targets (Joshi and Ayee Reference Joshi, Ayee, Brautigam, Fjeldstad and Moore2008), politicians have been reluctant to pursue unpopular fiscal policies which could damage their re-election chances, especially in settings where there has been some evidence of informal agreements between the state and informal workers to restrict the tax burden of the informal economy (Holland Reference Holland2014; Joshi, Prichard and Heady Reference Joshi, Prichard and Heady2014; Kenyon and Kapaz Reference Kenyon and Kapaz2010).

As mentioned earlier, earmarking offers citizens a clear and transparent link between the funding source (taxes) and its expenditure (public services) which, if well implemented, could build the foundations of a social contract between the government and its citizens. And while politicians have shown a reluctance in expanding taxation to large electoral bases, earmarking revenue collected from taxation could allow governments to increase acceptance in tax reforms aimed at funding popular programs and expand their coverage to the poorest (Cashin, Sparkes and Bloom Reference Cashin, Sparkes and Bloom2017; Ozer et al. Reference Ozer, Bloom, Valle, Banzon, Mandeville, Paul, Blecher, Sparkes and Chhabra2020).

Cashing, Sparkes and Bloom (Reference Cashin, Sparkes and Bloom2017) identified at least 80 countries which have earmarked revenue or expenditures to fund healthcare services, showing a political willingness from a number of administrations to prioritize access to healthcare. Given that the vast majority of workers in the informal sectors belong to the lowest wealth quintiles and are therefore more likely to postpone access to medical care,Footnote ² we would expect informal workers to more likely support expanded taxation if earmarked to increased healthcare provision. In developing countries where trust in the government remains low, we would therefore expect support for earmarking programs to be correlated to individuals’ beliefs in the state’s ability to provide the promised benefits of registration and tax compliance.

Preferences for a health for tax earmarking program

This article’s second question is what tax earmarking programs could generate more buy-in from informal worker in Nigeria. For this, we run a conjoint experiment (see below) where we verify different programs across six attributes. Besides the literature, these attributes were decided on through in-depth interviews with the Nigerian Governors Forum, tax authorities, commissioners of finance, ministries of health, health insurance agencies, care facilities and healthcare development agencies. We conducted these interviews in each of the 12 states the conjoint experiment was conducted in. We also pre-tested the conjoint experiment with enumerators and then during the piloting of the survey to make sure respondents understood the questions and agreed on their importance. We explained to respondents that we wanted their feedback on a few alternative policies that the government could take to provide free or low-cost healthcare in exchange for paying taxes. We offered them the chance to ask if they did not understand a policy option well. The final six attributes were:

1. 1. Tax level: Individuals determine the maximum amount of tax they are willing to pay (WTP) in exchange for a health insurance by estimating the perceived expected gain in well-being (from current or potential future illness) and loss in income (Cairns and Shackley Reference Cairns and Shackley1993; Pauly Reference Pauly and Sloan1995). WTP measures individuals’ willingness and ability to trade money for health (Bala et al. Reference Bala, Mauskopf and Wood1999). We expect that programs with lower tax levels are more popular.

2. 2. Deductible level: Various studies on health insurance preferences have suggested, in line with neoclassical economic theory, that health insurance programs with higher deductible and co-pays lower the appeal of said program (Barringer and Mitchell Reference Barringer and Mitchell1994; Kerssens and Groenewegen Reference Kerssens and Groenewegen2005).

3. 3. Start of coverage: Studies suggest that the timing of compensation or distributional measures that coincide with reforms is critical, with social programs starting before or together with reforms generating higher trust and support than social programs that would only start after reforms (IMF 2013; Guillaume, Zytek and Farzin Reference Guillaume, Zytek and Farzin2011).

4. 4. Who to pay: We expect that preferences and tax compliance improve if the collecting institution is perceived to be trustworthy (i.e., less corrupt) and accountable (Prichard, Custers, Dom, Davenport and Roscitt Reference Prichard, Custers, Dom, Davenport and Roscitt2019). In Ghana, efforts to formalize the informal sector were implemented collaboratively with associations who collected the taxes themselves and remit revenues to the government (Joshi and Ayee Reference Joshi, Ayee, Brautigam, Fjeldstad and Moore2008)

5. 5. How to pay: Recent reforms aimed at reducing the cost of compliance by simplifying methods for paying taxes, privileging online payments via banks or SMS over interactions with tax collectors (Prichard, Custers, Dom, Davenport and Roscitt Reference Prichard, Custers, Dom, Davenport and Roscitt2019). Not only would online operations prevent delays but reduces the risk of harassments and corruption from tax collectors (Brautigam et al. Reference Brautigam, Fjeldstad and Moore2008), which could undermine trust in the tax system. In Kenya, the introduction of the iTax was mainly motivated by a willingness to reduce the risk of abuses from tax collectors (Franzsen and McCluskey Reference Franzsen and McCluskey2017).

6. 6. Specific perks: Governments can also offer additional perks to convince workers. We included three: (1) the extension of healthcare to children, which is suggested to improve income (Molina-Vera Reference Molina-Vera2021); (2) The maintenance of healthcare when income is lost, which studies suggest can reduce precautionary saving and help productive job search (Gruber and Madrian Reference Gruber and Madrian1997; Chou, Liu and Hammitt Reference Chou, Liu and Hammitt2003); (3) the provision of a state identify card that informal workers could use to access new government services, which has played a key role in several developing countries to connect informal workers with government benefits and services (Winkler, Ruppert Bulmer and Mote Reference Winkler, Ruppert Bulmer and Mote2017; Alfers Reference Alfers2013).

Our second hypothesis is about what programs are more popular. Given the lower-income levels of informal workers, we hypothesize that the neoclassical economic elements are of primary concern and that therefore the tax level (primarily) and deductible level (secondarily) matter more than how and who to pay. This hypothesis is rooted in several theoretical fields. First, rational choice theory has long argued that individuals make decisions based on a weighing of costs and benefits (Becker Reference Becker1976). Similarly in public choice theory, it is widely held that voters (or people in general) prioritize policies that minimize personal costs because they directly impact their financial well-being. Adjacent theories such as those focused on the so-called “basic needs approach” also highlight that for low-income individuals, affordability is paramount and so cost considerations are a primary concern in evaluating alternative options (Streeten et al. Reference Streeten, Burki, Haq, Hicks and Stewart1982). Behavioral economics too, address constraints for TFS programs. Key texts highlight that for poorer individuals, the immediate costs and financial stressors outweigh long-term benefits due to issues such as present bias (i.e., the choice for a smaller present reward over a bigger future reward), in combination with limited financial literacy (National Academies of Sciences and Medicine 2023). This also supports the notion that for TFS programs, poorer informal workers would have a preference for the cheaper options. Several development economic scholars have pointed to such critical short-term costs that can dissuade poorer people, or lower short-term costs that can incentivize them to opt in (Banerjee and Duflo Reference Banerjee and Duflo2011; Deaton Reference Deaton2013).

Drivers of preferences for health for tax earmarking

On the other hand, distance to benefits relates to individual preferences and perceived value of services the government can provide in exchange for tax payments. Here too public choice scholars and public policy scholars (e.g., John Reference John2013) have long argued that those who perceive direct benefits from collective goods are more likely to contribute and cooperate, as are those that have personal experience with the benefits of proposed policies. This is also a notion in behavioral economics on health decisions and wealth have indicated that personal experience and proximity to benefits can influence decision-making towards more cooperation.

While these two measures of distance are interconnected, they are however, distinct. We could think of distance to government as perceived or expected reciprocity from tax payments while distance to benefits as “real” reciprocity (see below how distances to government and to benefits are measured). By building social pacts and therefore reducing the distance to benefits, we believe that governments suffering from low levels of trust can be perceived as more legitimate, allowing to strengthen or renew the social contract.

Our two measures of distance flow from the literature on trust and social contracts. Trust, which is more closely related to our “distance from government” measure, is considered as one of the components of tax morale, a reflection of individual ethics and values and social norms. Given the mounting evidence that drivers of tax morale (and trust in particular) are key determinants of tax compliance (Cummings, Martinez-Vazquez, McKee and Benno Reference Cummings, Martinez-Vazquez, McKee and Benno2009) and that higher levels of tax morale are linked to greater support for tax reforms (Prichard, Custers, Dom, Davenport and Roscitt Reference Prichard, Custers, Dom, Davenport and Roscitt2019), strengthening tax morale through trust-building initiatives can encourage “quasi-voluntary” tax compliance. This can however only be successfully achieved with good governance and political accountability (Bates and Lien Reference Bates and Lien1985). Against this background, taxation could lay the foundations of a social pact between the government and its citizens by directly linking tax payments to the government to public goods and service provision (in this case free or low-cost access to basic healthcare services), connecting revenue mobilization, reciprocity and accountability and fostering trust and strengthening legitimacy. Such connections between revenue mobilization, reciprocity and accountability can be found in Ghana and Ethiopia, where expanded taxation of the informal sector has led to public engagement between the government and business associations (Joshi and Ayee Reference Joshi, Ayee, Brautigam, Fjeldstad and Moore2008; Prichard Reference Prichard2009, Reference Prichard2010).

Social contracts are structured along four parameters: protection, provision, participation and production of hegemony (Sobhy Reference Sobhy2021). Both our measures of distance are closely associated with the concepts of provision, participation but also protection given the government’s role to provide security to its citizens (Furness and Trautner Reference Furness and Trautner2020). In this model, providing security can be thought as supplying both physical (danger, threat) and social protection. We believe that once physical protection is ensured, social protection, through health support, is paramount.

While theoretically appealing, taxing the informal sector, as mentioned earlier, doesn’t come without its challenges, in part due to the limited rewards given the costs, but also because most businesses, perceived by government authorities to be tax evaders, are in fact already registered with and licensed by local government entities (Gallien and Van den Boogard Reference Gallien and Van den Boogard2021). Individuals engaged in low-skilled occupations, such as street vendors, shoe shiners, or waste collectors, may exhibit reluctance towards formalization due to anticipated costs associated with registration and compliance with business regulations, including labor laws (e.g., Maloney Reference Maloney2004; Perry Reference Perry2007). For this large subgroup of the informal sector, we would expect our “distance to benefit” measure to be related to support for tax-for-health-service program.

Data collection

To investigate what determines interest in, and perceptions about tax-for-health-services programs by informal workers, we conduct a large-scale household survey in Nigeria. The data-collection took place in December 2020 and January 2021 with informal workers in two states within each of the six geopolitical zones in Nigeria. The two states in each geopolitical zone were randomly selected. The following states were sampled:

• South South: Edo & Delta

• South East: Anambra & Enugu

• South West: Ogun & Ekiti

• North Central: Kwara & Plateau

• North East: Yobe & Borno

• North West: Kaduna & Kano

The household survey was conducted successfully with 1000 respondents in each state (i.e., 12,000 in all 12 states). The survey utilized a 40/60 rural/urban split to make sure the survey sufficiently captured the urban population that may have more exposure to both taxation and health centers. This split was later adjusted with weights to make sure that the weight of each household reflected its actual probability of selection in the survey. The survey also employed a 50/50 gender split to make sure we have equal representation of men and women in the study. This report summarizes key findings from this exercise, so when referring to “Nigeria,” it refers to these 12 states spread out across 6 geopolitical zones.

The methodology was designed to exclusively sample adult citizens (age 18 and above) that are employed informally. The survey team did this by making sure that employed people within households were not officially registered or those who held jobs lacking basic social or legal protection and employment benefits. It excluded workers who paid income tax, who were employed by the government or any other formal employer, or who had a business or company that is officially registered. This definition of the population relied on the definition of informal sector employees by the Bank of Industry Limited (BOI), which is Nigeria’s oldest and largest development financing institution (Bank of Industry 2018).

The survey used a multi-stage random selection process. First, there was the random selection of dwelling structure using enumeration areas (EA) that were randomly selected by the National Population Commission of Nigeria, followed by randomly generated GPS coordinates within each EA. Second, there was the random selection of households in which the electronic device randomly selected one household after the enumerator entered all households living in the dwelling structure. And third, there was the random selection of respondent within that household. Finally, there was a stringent procedure for quality control in which every two enumerators received a dedicated supervisor, and independent “backcheckers” implemented post-hoc controls.

Conjoint experiments

To explore how informal workers in Nigeria perceive tax-for-health-services programs, we use two analyses. First, we use simple descriptive statistics to identify key characteristics of informal workers in Nigeria, and explain their experiences with healthcare, taxes and government to illustrate why there is openness to a tax-for-health-services program. Second, in order to assess informal workers’ preferences regarding tax-for-health-services programs, we employ a conjoint experiment design. In this design, respondents are presented with two hypothetical tax-for-health-services program profiles, each varying in attributes (i.e., features or characteristics of the program) and the value (i.e., content) within each attribute. The order of attributes is randomized, and the value within each attribute for each profile is randomly chosen. Subsequently, respondents are asked to indicate their preferences for each profile.Footnote ³ To verify the randomness of survey respondent assignment, we conducted simple balance tests which regress the respondents’ basic characteristics on the attributes. The results, presented in Appendix Table A2 show that the coefficients of the attributes are not statistically significant for almost all the basic characteristics, suggesting evidence that the attributes are randomly assigned.

Empirical investigation on public opinion toward tax-for-health-services programs is not an easy task mainly due to the fact that the composition of tax-for-health-services programs is multidimensional: various characteristics of tax-for-health-services programs, such as the level of tax paid on income to program perks, can affect public attitudes. Unlike common vignette experiments, a conjoint design allows researchers to estimate causal effects of numerous factors on respondent’s multidimensional preferences (Hainmueller, Hopkins and Yamamoto Reference Hainmueller, Hopkins and Yamamoto2014).

In recent years, conjoint experiments have been considered one of the primary tools to causally identify multidimensional policy preferences (Horiuchi, Smith and Yamamoto Reference Horiuchi, Smith and Yamamoto2018). Scholars have employed conjoint analysis in a range of policy issues (Christensen and Rapeli Reference Christensen and Rapeli2021) such as environmental and climate change policies (e.g., Bechtel and Scheve Reference Bechtel and Scheve2013), Eurozone bailout policies (Bechtel, Hainmueller, and Margalit Reference Bechtel, Hainmueller and Margalit2017), health policies (e.g., Bridges et al. Reference Bridges, Hauber, Marshall, Lloyd, Prosser, Regier, Johnson and Mauskopf2011), and foreign policies (e.g., Clary and Siddiqui Reference Clary and Siddiqui2021). Moreover, existing studies have suggested that conjoint analysis mitigate social desirability biases since respondents are not directly asked for opinions about socially sensitive attributes in the conjoint set-up (e.g., Carey et al. Reference Carey, Carman, Clayton, Horiuchi, Htun and Ortiz2020; Horiuchi, Markovich and Yamamoto Reference Horiuchi, Markovich and Yamamoto2021). In particular, prior studies have suggested that conjoint analysis is a very effective method to measure economic preferences such as willingness to pay since it is easier for respondents to apply preference ordering and rating when competing alternatives are presented than when one option is given (e.g., Breidert, Hahsler, and Reutterer Reference Breidert, Hahsler and Reutterer2006; Schmidt and Bijmolt Reference Schmidt and Bijmolt2020; Miller et al. Reference Miller, Hofstetter, Krohmer and Zhang2011).

Our experiments begin with a short description mentioning that the profiles show the characteristics of two hypothetical tax-for-health-services programs. Then, respondents judge five comparisons between pairs of tax-for-health-services programs in which we randomize the order and values of multiple attributes in two program profiles. The full list of attributes and their values used in the conjoint experiments are presented in Table 1.

Table 1.

Attributes and values of hypothetical tax-for-health-services programs

Below the two hypothetical tax-for-health-services program profiles, the respondents are asked to show how strongly they support or oppose each of the presented tax-for-health-services programs on a 5-point scale (strongly support, moderately support, neither support nor oppose, moderately oppose and strongly oppose). We also asked respondents to make a forced choice between them by asking which program they support being implemented. We display an example of the pair of the project profiles as well as questions in Appendix Section A2.

Though existing studies using conjoint experiments have frequently relied on the forced choice paired design (e.g., Hainmueller, Hangartner and Yamamoto Reference Hainmueller, Hangartner and Yamamoto2015; Hainmueller and Hopkins Reference Hainmueller and Hopkins2015; Hankinson Reference Hankinson2018), a growing number of studies have raised some serious concerns of using forced choice outcomes as the dependent variable (Abramson, Ko¸cak and Magazinnik Reference Abramson, Ko¸cak and Magazinnik2022; Ganter Reference Ganter2023).Footnote ⁴ In this regard, in the empirical analyses, we use the 5-scale preference as our main dependent variable. Still, we also check the results using forced choice outcomes as the dependent variable and find that the main findings remain substantively same (Appendix Section A3).

In order to evaluate the causal effects of key characteristics of tax-for-health-services programs on public attitude, we estimate the linear regression models with standard errors clustered by the respondent, which is asymptotically equivalent to average marginal component effect (AMCE) estimators (Hainmueller, Hopkins, and Yamamoto Reference Hainmueller, Hopkins and Yamamoto2014). The AMCE quantifies the average change in support (on the 5-point scale) associated with changing a specific attribute from its reference level to a different level, while holding all other attributes constant. We created dummy variables for each level of each attribute and included them in the regression models while excluding those for the reference levels. We created a binary indicators for each level of each attribute and included them in the regression models while excluding those for the reference levels.Footnote ⁵ The AMCE, which does not rely on functional form assumptions about the choice probabilities, captures the causal average effects of each attribute level on the probability that the project will be supported, across all possible combinations of other attributes. Specifically, we estimate the following regression model:

$$\eqalign{ & Suppor{t_{i,j}} = \alpha + {\beta _1}Who:LocalGovernmentAuthorit{y_{i,j}} \cr & + {\beta _2}Who:StateInternalRevenueService{s_{i,j}} + {\beta _3}Level:N6000 - N12000 \cr & + {\beta _4}Level:N24000 - N36000 + {\beta _5}Deduct:LowcostHealthcare \cr & + {\beta _6}How:UUSD + {\beta _7}How:Bank + {\beta _8}HealthCoverage:3MonthsAfter \cr & + {\beta _9}HealthCoverage:3MonthsBefore + {\beta _{10}}Perk:Extended \cr & + {\beta _{11}}Perk:StateIdentityCard + {E_{i,j}}, \cr} $$

where i refers to a respondent, j indexes a round. Those levels that are not included in the models (e.g., Level:N 12,000 − N 24,000) serve as the reference levels used to estimate the marginal effects of each attribute. To check if the estimated β (marginal effect of a given level from reference level) is statistically distinguishable from 0, we conducted two-sided t-tests. In the results section, we report not only estimated coefficients but also standard errors and p-values from these two-sided t-tests.

Operationalizing the measures of “distance”

To test Hypotheses 3 and 4, we consider two types of distance. First, we operationalize measures of “benefit distance” across five variables. First, to measure whether there is a need, we verify whether the respondent has a chronic disease (yes/no) and whether the respondent or his proximate family has accessed healthcare in the last year (yes/no). Second, to measure knowledge about the benefits of insurance, we verify whether the respondent has heard of health insurance (yes/no). Third, to measure trust in providers, we measure confidence in the health clinic (5-scale from very confident to not confident at all), as well as trust in health care institutions generally (do not trust; neutral; trust).

Second, we operationalize measures of “distance to government” across six variables. First, to measure local political involvement, we verify (1) whether the respondent thinks there are opportunities for citizens to voice their opinion on local government area (LGA) business, (2) whether the respondent thinks the LGA is very or somewhat responsive to their needs, and (3) whether the respondent has attempted to contact their MP or local councilor in the last year. Second, to measure their engagement with government through taxation, we check whether respondents have paid at least one tax in the last year. Third, to measure exposure to social programs, we verify (1) the satisfaction of respondents with regards to services (scored 1 if the respondent is satisfied with at least 6 out of 10 services), and (2) whether the respondent received social benefits (scored 1 if the respondent received at least one social benefit).