Building a Social Contract? Understanding Tax Morale in Nigeria

Abstract

An important part of every country’s development process is the building of a social contract in which citizens pay tax and, in turn, receive public goods and services. Evidence suggests that this is associated with the establishment of a norm of tax payment and a belief that non-payment is wrong. We exploit a new, nationally representative, dataset to explore which factors are associated with higher tax morale in Nigeria. We find that a perception of higher penalties and greater difficulty avoiding taxes are both associated with higher tax morale. Tax morale is also higher the more people believe that other Nigerians pay taxes, the less frequently they have to pay bribes and the greater the trust they have in tax officials. However, we also find that Nigerians who believe that tax officials discriminate in their treatment of different ethnic, religious and gender groups have higher, not lower, tax morale. And we find no relationship between service delivery and measures of tax morale based on the respondent’s own behaviour, in contrast to the positive association found in the literature using broader measures of tax morale. This suggests that building a social contract based on taxation may be harder than previously thought.

1. Introduction

It is generally agreed that raising tax as a share of GDP is an important part of the development process in poorer countries and a key component of building an effective state (Akitoby, 2018). However, some poor countries collect very little tax. Nigeria, Africa’s most populous country and largest economy, is a case in point. Non-oil revenue in Nigeria has remained at around 3–4 per cent of GDP since 2011 (IMF, 2018). This is far below many other countries: South Africa and Brazil have non-oil revenues between 20 and 25 per cent of their GDP; Cameroon, Egypt, Ghana, Indonesia, Malaysia and the Philippines all collect between 10 and 15 per cent of GDP in non-oil revenues.¹

The collapse in oil prices in 2015, and more recently as a result of the COVID-19 crisis, led to the government of Nigeria making it a policy priority to significantly boost the share of non-oil revenue in GDP to 15 per cent by 2020 (Federal Republic of Nigeria, 2017).² One of the challenges in achieving this is the high levels of non-compliance in the tax system. There are around 57 million Nigerians who are economically active³; but very few pay tax. The Joint Tax Board in Nigeria claimed that there were 13.4 million individuals registered for tax in November 2016, but there is no publicly available information on how many of these individuals actually paid any tax.⁴ Rates of payment by registered taxpayers are low; the International Survey on Revenue Administration (ISORA) shows that there were only 761,000 registered to pay Personal Income Tax in 2016 (most of whom were formal sector employees paying by PAYE),⁵ but the survey suggests that fewer than 15,000 of these are active taxpayers (IMF, 2018).

Why do so few Nigerians pay tax? The obvious answer is that the quality of administration and enforcement is poor, particularly below the Federal level. However, evidence from numerous countries also points to the need to build popular support for taxation. This paper systematically explores the attitudes and perceptions of ordinary Nigerians about paying tax and the factors that may influence such attitudes. We use the phrase ‘tax morale’ in a broad sense to refer to attitudes about non-payment of tax. This is slightly different from the way in which the phrase is used in the literature. In particular, Luttmer and Singhal (2014)’s extensive review of tax morale defines tax morale as the non-pecuniary motives for compliance (Luttmer & Singhal, 2014). However, our principal interest is not in compliance, but only on the attitudes to payment and non-payment. Moreover, our aim is to explore the factors that are associated with attitudes towards tax payment in Nigeria and these may include pecuniary as well as non-pecuniary factors. Specifically, we examine five sets of questions:

• Enforcement and penalties: how are perceptions about the likelihood of being caught for non-payment and the size of penalty received associated with attitudes towards non-payment?

• Knowledge and complexity: is the level of knowledge of the tax system and its simplicity associated with higher tax morale?

• Social and individual norms: are people with strong religious beliefs more willing to pay tax; is a positive attitude to payment more likely if others in the same area share such views?

• Trust and fairness: do people have higher tax morale if they trust the Administration, and if they perceive that tax is collected fairly?

• Fiscal exchange/social contract: do people have higher tax morale if they feel that they are getting something in return?

We draw on a new, nationally representative survey of Nigerians’ attitudes towards taxation to answer these questions.

To briefly preview our results, we find that a perception of higher penalties and greater difficulty of avoiding payment are associated with higher tax morale. Tax morale is also higher among those that actually paid some income tax in the previous year. Surprisingly, we find little evidence for the influence of individual norms, including religiosity, on tax morale, but social norms do appear to have an influence. Trust in tax collectors and a lower perception of the frequency of bribes are both strongly associated with a more positive attitude towards payment. However, contrary to what might be expected, we find that Nigerians who believe that tax officials discriminate in their treatment of different ethnic, religious and gender groups are more, rather than less, likely to have higher tax morale. Also, in contrast to much of the literature, we find no association between the quality of services or service improvements and tax morale. We show how this may be due to the measure of tax morale typically used in such studies.

In the absence of a randomised intervention, or equivalent natural experiments, our results should be interpreted as associations, rather than causal relationships. However, we believe that elaborating the stylised facts around tax morale and its associations with key variables is important, since it provides the basis for formulating hypotheses for more formal examination.

Our paper contributes to the literature in two ways. First, although there is some excellent literature on taxation in Nigeria (reviewed below), this is the first time that a dataset has been available that is representative of the country as a whole allowing us to draw inferences that are applicable to a large and strategically important developing country. In addition, the size of our dataset enables us to examine the association between a wide range of different factors and tax morale. This is valuable conceptually, but also practically as it suggests areas where improvements might be most effective in shifting attitudes towards taxation in Nigeria. Second, while our findings confirm results found elsewhere about the correlates of tax morale, they also provide a significant challenge to the existing literature in the two areas mentioned above with potentially important implications for research and policy.

The paper is structured as follows. The next section briefly reviews the literature on tax morale both internationally and specifically in the Nigerian context. This is followed by a description of the data and the variables used along with their limitations. The results are then presented, including both the socioeconomic and geographical correlates of tax morale, as well as measures based on theories drawn from the literature. We conclude with implications for policy and further research.

2. Factors associated with tax morale

Both the neoclassical (Allingham & Sandmo, 1972) and behavioural literature on tax (Pickhardt & Prinz, 2014; Richardson, 2006; Torgler, Demir, Macintyre, & Schaffner, 2008) suggest several factors that influence tax morale. Like Ali, Fjeldstad, and Sjursen (2014), we prefer to treat both pecuniary and non-pecuniary determinants of tax morale together, since taxpayers themselves frequently appear to make little distinction between these motivations (Fjeldstad, 2004).

The literature generally suggests five dimensions that are associated with tax morale:

1. Neoclassical theory: The seminal paper by Allingham and Sandmo (1972) lays out the neoclassical theory of tax payment, arguing that payment depends on the size of taxes, the level of enforcement and the size of the penalties for non-payment. There is strong empirical support for the idea that stronger legal enforcement and more frequent audits increase tax compliance and lower evasion (Jayawardane & Low, 2016; Richardson, 2008). We expect that these factors will also be associated with tax morale as we have defined it.

2. Information and knowledge: A review of 45 countries found complexity as one of the key factors determining tax avoidance (Richardson, 2006). This effect on compliance and morale has also been confirmed for developing countries (Abukakari & Adafula, 2013): sharing more information on the taxation system can increase both (Ali et al., 2014; Nyamwanza, Mavhiki, Mapetere, & Nyamwanza, 2014). A 2014/2015 Afrobarometer survey in Nigeria showed that knowledge influences compliance with more than two-thirds of Nigerians finding it difficult to find out what taxes or fees to pay the government (Aiko and Logan, 2014). More information about taxes also increases the perception of fairness and influences an individual’s attitude towards their own and other people’s tax evasion (Eriksen & Fallan, 1996). Information may also be important to bridge differences in perceptions within countries. For example, Oberholzer and Stack (2014) found differences in perceptions between the four major ethnic groups in South Africa and suggest that tax education could help overcome these differences.

3. Social and personal norms: The literature emphasises that the most important norms strengthening tax morale are the ones internalised as personal norms (Jimenez & Iyer, 2016). Other, societal norms influence more indirectly (Bobek, Hageman, & Kelliher, 2013; Bobek, Roberts, & Sweeney, 2007; Damayanti, Sutrisno, Subekti, & Baridwan, 2015) and can also include negative drivers on tax morale such as the observation of wide-scale evasion (Dell’Anno, 2009; Frey & Torgler, 2007). The impact of religion on tax morale is also significant (Torgler et al., 2008); a review of 30 countries suggested that religiosity increases tax morale (Torgler, 2006). Another review of 47 countries finds that higher religiosity leads to lower tax evasion (Richardson, 2008). The process behind this is not clear, but one study concluded that the threat of shame coming from religious identity appears larger than the threat of embarrassment (Grasmick, Bursik, & Cochran, 1991). Nigerian researchers have also explored the impact of religious norms on payment. Fagbemi, Uadiale, and Noah (2010) found no difference between the attitudes towards compliance of Christians and Muslims, but substantial support for the statement that ‘tax evasion is ethical if the government discriminates against me because of my religious or ethnic affiliations’.

4. Trust and fairness: The literature emphasises that a perception of fairness of the tax system is crucial to both morale and compliance (Hennighausen & Heinemann, 2015; Richardson, 2006; Vihanto, 2003). Conversely, tax evasion is often considered ethical if the tax system is considered unfair (McGee & Benk, 2011). For example, Meagher (2016) found that poor women and poor non-indigenes in northern Nigeria were the most vulnerable to extortion and the least able to engage in tax bargaining or to complain against unfair practices. As she puts it ‘different social groups within the informal economy experience the legitimacy of taxation and its effect on political voice very differently’ highlighting the importance of differentiating by gender and ethnicity. More broadly, trust in government is an important driver of the perception about the fairness of the tax system (Jimenez & Iyer, 2016). Similarly, a general perception of corruption and a lack of accountability tend to lower tax morale (Bird, Martinez-Vazquez, & Torgler, 2008; Torgler et al., 2008).

5. Reciprocity and service delivery: The literature on social contracts suggests that there is higher tax morale and lower tax evasion when there is a stronger belief that tax efforts are linked to economic success (Hennighausen & Heinemann, 2015). Similarly, Cheeseman and De Granmont (2017) showed how the Lagos State government used adverts on electronic and print media as well as billboards to create consciousness of the civic obligation to pay taxes while publicising the dividends from public expenditures. Bodea and LeBas (2016) also looked explicitly at how tax influences the construction of the social contract in Nigeria. Their 2010 survey found that citizens that received more public goods and were more satisfied with the way in which the government had spent tax resources, were also more likely to have a tax compliant attitude. However, communities that were able to ‘opt out’ of government provision through the use of community-provided public goods were less inclined to comply. Studies in other countries suggest that having to pay for your own services damages the social contract – for example, having to pay for one’s personal security (through bribes) reduces tax compliance (Ali et al., 2014).

A key feature of the literature on tax morale is that there are a wide range of potential explanatory factors (Pickhardt & Prinz, 2014). In particular, we would expect tax morale to be associated with a variety of personal characteristics of the individuals and the households from which they come, such as the level of education, socio-economic status, income, employment status, ethnic origin and religious affiliation (Dornstein, 1987). The literature suggests that differences in income can explain divergent views on tolerance towards tax evasion (Richardson, 2006; Shafiq, 2015) and that, when income rises in developing countries, factors such as religion and gender appear to matter less (Shafiq, 2015). However, the effect of income on tax morale is contested. Some studies have found that less educated and poorer people are more opposed to tax evasion than more educated and wealthier people (Mcgee & Tyler, 2006), while other studies have found that educated people have higher tax morale (Cyan, Koumpias, & Martinez-Vazquez, 2016). This may be explained by regional differences, suggesting the importance of including geographical region as another variable, especially in a large federal country such as Nigeria.

3. Data

Most datasets which have been used to explore tax morale have been of two types. On the one hand are cross-country surveys that rely on data that are collected in several countries (e.g., Afrobarometer and the World Values Survey).⁶ These provide cross-country comparability, but lack a deep focus on tax morale, relying on a small number of general questions. On the other hand, there are single country studies that have focused on tax issues – including tax morale – but whose sample is not designed to be nationally representative.

We draw on a large, new, nationally representative dataset that was specifically designed to explore the attitudes and perceptions of Nigerians towards tax compliance. This consisted of a quantitative household survey as well as Focus Group Discussions (FGD) and In-Depth Interviews (IDI) with tax officials in selected states.⁷ We briefly outline below the composition of each of the sources of data.

3.1. Household/individual survey

In July 2018, a survey was conducted of 10,000 Nigerian adults across the country. A clustered, stratified, multi-stage random selection procedure was used to achieve a nationally representative sample. The sample is also representative of Nigeria’s six geo-political zones,⁸ as well as of both urban and rural areas. Households were sampled from all states (with probability proportionate to their populations). Because of the small sample size in each state, results are not statistically representative at the state level, except for the oversampled states (see below). The data were collected by in-home, face-to-face personal interviews in the language with which the respondent was most comfortable. Respondents were adult Nigerian males and females aged eighteen years and above, who had lived in the household for a period of not less than six months. An equal number of male and female respondents was selected. Since urban dwellers are much more likely than rural residents to have experience of the tax system, the sample was stratified to provide greater representation in urban areas: 70 per cent of individuals were selected from urban areas and 30 per cent from rural areas. Sampling weights were created to account for this design so that the results are nationally representative.

It was also desired to be able to provide representative results at the state level. Doing this for every state would have been prohibitively expensive. Therefore, six states were ‘oversampled’ – one in each geo-political zone: Ogun (South-West); Rivers (South-South); Abia (South-East); Nasarawa (North-Central); Kano (North-West); Bauchi (North-East). The sampling and data-collection procedure in each of these states was the same as that for the nationwide survey. In each of these six states, 1000 individuals were selected (again split 70% urban; 30% rural; and 50:50 men/women). This provided a sufficiently large sample in these six states to allow comparisons of results across these states. Sampling weights have been calculated so that it is possible to use the entire sample of 16,000 households as a single dataset, with weights reflecting both the urban/rural stratification and the much higher probability of selection of the respondents in the over-sampled states.⁹

3.2. Focus group discussions

Quantitative surveys can provide representative data on the perceptions of individuals and firms on tax issues, but it is sometimes difficult to know why respondents hold the views shown. We therefore complemented the quantitative surveys with an extensive set of Focus Groups Discussions (FGDs).¹⁰ Two FGDs were done with households in each of two states in every region – one with men and another with women. The design of the FGDs ensured that we obtained views from both men and women and in both urban and rural areas.

4. Variables and limitations

4.1. Variables

Our key variable of interest is tax morale, which we have defined as the attitude that people have towards non-payment of their own taxes. As noted above, the literature generally defines tax morale as the ‘intrinsic motivation to pay taxes’ (Feld & Frey, 2002, 2007; Frey & Torgler, 2007; Torgler, 2003, 2007), which is usually considered as separate from pecuniary motives for paying taxes. However, our focus is on peoples’ attitudes towards non-payment which may be influenced by both pecuniary and non-pecuniary factors. Specifically, our dependent variable question asked respondents to respond to the following statement: ‘I would not pay my taxes if I would not be caught’, where the options were: ‘Agree’; ‘Neither agree nor disagree’; and ‘Disagree’. These represent an increasing level of tax morale.¹¹

Our key explanatory variables follow the categories described above and are shown in Table 1.

Table 1. Key explanatory variables

Explanatory category/variable	Description
Neoclassical
Penalty when getting caught for not paying taxes (perception)	Perception of the size of penalty that a person would have to pay if they were caught: range from ‘ nothing’ to ‘ a heavy penalty.’
Difficulty of avoiding paying income or property taxes	Perception of how easy or difficult it is to avoid paying income or property taxes: range from ‘ Very easy’ to ‘ Very difficult.’
Lightness of tax burden	Perception of how light the tax burden is: range from not paying tax to ‘ a very significant burden’ .
Information and knowledge
Ease of finding out what taxes and fees should be paid to the government	Perception with range from ‘ Very easy’ to ‘ Very difficult’
Received communication from the government in the last year about what taxes to pay	Yes/no
Tax liability and payment
Required to pay personal income tax (PAYE) – awareness	Belief that you were liable for PAYE (Yes/no)
Paid some form of income tax last year	Whether you actually paid some form of income tax in the last year (Yes/no)
Required to pay property taxes – awareness	Belief that you were liable for Property tax (Yes/no)
Paid some form of property tax last year	Whether you actually paid some form of property tax in the last year (Yes/no)
Social and personal norms
Membership of religious group	Whether you are an active member of a religious group (yes/no)
Frequency Nigerians pay their tax (perception)	Perception of how frequently Nigerians pay their taxes (reversed scale of how often they avoid paying taxes with range: ‘ never’ to ‘ always’)
Community pay income tax	Whether community members actually pay full amount of income tax (‘ Never’ to ‘ Always’)
Community pay land us/property tax	Whether community members actually pay full amount of land use or property taxes (‘ Never’ to ‘ Always’)
Trust and Fairness
Do tax officials treat all ethnic groups the same?	Choice between ‘ treat all ethnic groups the same’ vs ‘ favour some ethnic groups over others’
Do tax officials treat all people from all religions the same?	Choice between ‘ treat people from all religions the same’ vs ‘ favour people from some religions over others’
Do tax officials treat all men and women the same?	Choice between ‘ treat all men and women the same’ vs ‘ favour one gender over the other’
Infrequency of tax officials taking a bribe	Perception of how infrequent it is for tax officials to ask for bribes (reversed scale of how frequently they ask from ‘ Never’ to ‘ Always’)
Trust in State tax officials to collect tax fairly*	Perception of how much people trust State tax officials: range from ‘don’t trust at all’ to ‘trust a lot’
General trust in government	Perception of how much people believe that government acts in the interest of all Nigerians vs acts in its own interests (sum of responses about trust in federal, state and local governments)
Approval of President Buhari	Approval of the President’ s performance: range from ‘ Strongly Disapprove’ to ‘ Strongly Approve’
Reciprosity and service delivery
Services satisfaction (index)	Average of satisfaction with six services (schools, health, local markets, roads, security, electricity) with range for each from ‘ Very dissatisfied’ to ‘ Very satisfied’
Quality of government services provided by state government compared to 3 year ago	Perception about governments services relative to three years ago: range is Worse, Same, Better.

*We look at trust in State tax officials because the main taxes imposed on individuals and households are levied by State governments.

In addition to the above variables, we examine the association between tax morale and a set of individual socio-economic and geographical characteristics including: age, level of education, employment status, gender, personal income after tax, whether they are in an urban or rural area, language group (a proxy for ethnicity), geopolitical region and religion.¹² Appendix Table A1 provides the descriptive statistics for all the variables, and Appendix Table A2 provides the proportions of responses to the survey questions used in the analyses.

4.2. The limitations of perceptions data

It is important to recognise that our data relate primarily to peoples’ perceptions about tax. Collecting and using perceptions data poses some challenges. Most obviously, it is important to be confident that the responses are truthful. Two potential problems may arise. First, respondents may be concerned that the interviewers are really from the government or the tax authorities. To mitigate this, the survey team were trained to emphasise that they were an independent survey research organisation with no connection to government or the tax authorities. Nonetheless, the suspicion may persist in the mind of the respondent which could bias responses towards higher tax morale. Second, responses may suffer from social desirability bias that is respondents give the answer that they believe to be most socially acceptable even if it is not their own view. Again, this is likely to bias responses towards higher tax morale. Given the low levels of tax morale actually reported in the survey, we believe that these biases are likely to be small.¹³

Another important challenge is that many of the factors that we believe may be associated with tax morale are also based on perceptions data. For example, we do not have concrete data on the actual likelihood of being caught for non-payment nor the size of actual penalties paid; we therefore use people’s perceptions of enforcement and penalties. Similarly, we do not have accurate data on the frequency of bribery, nor on the quality of services and so we rely on people’s perceptions of these variables. Almost all of our results are therefore susceptible to the possibility that there is some third variable that influences perceptions of both tax morale and the variable of interest. The associations that we find should therefore be treated as a starting point for exploring what may be driving the correlation observed.

5. Results

Tax morale in Nigeria is low. For our dependent variable, almost half of Nigerians (48.7%) agreed with the statement ‘I would not pay my taxes if I would not be caught’, while only a little more than a third disagreed. Nigeria also has low tax morale by African standards. Aiko and Logan (2014) find that Nigeria ranked 22nd for tax morale out of the 29 African countries surveyed.¹⁴ To understand the correlates of tax morale, Table 2 shows the pairwise correlations between each of the individual and geographic socioeconomic characteristics and tax morale. In addition, it shows the coefficient of an OLS regression of tax morale on that variable only.¹⁵ Table 3 shows the pairwise correlations between each of the variables in Table 1 and tax morale. The OLS coefficients in this case result from a regression of tax morale on that individual variable, controlling for all the socio-economic variables in Table 2.¹⁶

Table 2. Pairwise correlation and OLS estimation results: descriptive model with just the controls

Variables	Pairwise Correlation	OLS	OLS (SE)
Urbanisation	0.0190*	0.035	(0.039)
Female	−0.0415*	−0.075*	(0.029)
Employed	−0.0152	−0.027	(0.029)
Age (<25)	−0.0274*	−0.060*	(0.028)
Age (25–31)	0.0069	0.013	(0.023)
Age (32–38)	−0.0019	−0.004	(0.027)
Age (39–44)	0.0179	0.053	(0.035)
Age (45–51)	0.0075	0.026	(0.042)
Age (52 above)	0.0059	0.023	(0.052)
English	0.0211*	0.047	(0.040)
Yoruba	−0.0588*	−0.144*	(0.044)
Igbo	−0.0143	−0.043	(0.061)
Hausa	0.0767*	0.146*	(0.041)
Other	−0.0487*	−0.112*	(0.052)
No formal schooling/ Informal schooling	0.0087	0.026	(0.053)
Some primary schooling/ Primary school	−0.0124	−0.035	(0.043)
Some secondary school/high school/ Se	−0.0277*	−0.050*	(0.026)
Ordinary National Diploma (OND)/ Nation	0.0297*	0.070*	(0.035)
Higher National Diploma (HND)/Some univ	0.0101	0.024	(0.035)
None	−0.0156	−0.035	(0.044)
Less than 20,000 NGN	−0.0503*	−0.094*	(0.032)
20,001–40,000NGN	0.004	0.008	(0.035)
Over 40,000 NGN	0.0748*	0.176*	(0.036)
None	−0.0234*	−0.438*	(0.167)
Christian	−0.0182	−0.033	(0.038)
Muslim	0.0216*	0.039	(0.038)
Traditional	0.0141	0.194	(0.190)
South East	−0.0072	−0.020	(0.063)
North East	0.0657*	0.176*	(0.055)
South South	−0.1160*	−0.286*	(0.055)
North Central	0.0681*	0.173*	(0.064)
South West	−0.0535*	−0.121*	(0.044)
North West	0.0459*	0.096*	(0.047)

For OLS estimations, robust standard errors clustered at the enumeration area in parentheses.

*p < 0.10.

Table 3. Pairwise correlation and OLS estimation results: baseline models. The results for control variables in OLS estimations are omitted in the table

Explanatory Variables	Pairwise Correlation	OLS	OLS (SE)
Penalty when getting caught for not paying taxes (perception)	0.1154*	0.119*	(0.017)
Difficulty of avoiding paying income or property taxes	0.0687*	0.100*	(0.019)
Lightness of tax burden	−0.0172	−0.035	(0.026)
Ease of finding out what taxes and fees should be paid to the government	0.0378*	0.012	(0.022)
Received communication from government in the last year about what taxes to pay	0.0492*	0.063	(0.046)
Required to pay personal income tax (PAYE) – awareness	0.0375*	0.012	(0.037)
Paid some form of income tax last year	0.0844*	0.187*	(0.039)
Required to pay land use or property taxes – awareness	0.0268*	0.004	(0.041)
Paid some form of property tax last year	0.0200*	0.035	(0.052)
Membership of religious group	0.0004	0.009	(0.033)
Frequency Nigerians pay their taxes	0.0970*	0.098*	(0.018)
Community pay income tax	0.0154	0.014	(0.014)
Community pay land use/property tax	0.0351*	0.029*	(0.016)
Do tax officials treat all ethnic groups the same?	−0.0965*	−0.189*	(0.040)
Do tax officials treat all people from all religions the same?	−0.1046*	−0.185*	(0.040)
Do tax officials treat all men and women the same?	−0.1016*	−0.179*	(0.040)
Infrequency of tax official taking a bribe	0.0325*	0.042*	(0.022)
Trust in State Tax Officials to collect tax revenue fairly	0.0438*	0.054*	(0.019)
General trust in government	0.0128	0.008	(0.015)
Approval of performance by President Buhari	0.0307*	0.005	(0.011)
Services Satisfaction	−0.0166	−0.018	(0.020)
Quality of government services provided by State Govnmt compared to 3 yrs ago	−0.0246*	−0.007	(0.022)

For OLS estimations, robust standard errors clustered at the enumeration area in parentheses.

*p < 0.10.

5.1. Socioeconomic correlates of higher tax morale in Nigeria

Some studies of other countries have found that women have higher tax morale than men (Cyan et al., 2016; Torgler et al., 2008). Others, however, have suggested that the reverse is true in Africa (OECD, 2019). We confirm the latter result for Nigeria, where women have significantly lower tax morale than men. This may reflect the fact that women are less likely to be liable for tax. Age is not strongly associated with tax morale in Nigeria, except that the youngest adults have lower tax morale, again perhaps reflecting their lower liability.

We find a small and ambiguous association between education and tax morale, but, as expected, a positive association between income and tax morale. Those earning over NGN 40,000 per month have much higher tax morale than other income groups, while earning a very low income is negatively associated with tax morale. This is in line with focus group discussions that brought up poverty and lack of income as an important reason why avoiding payment is acceptable for some people. It is also consistent with other literature on the relationship between tax morale and income (Doerrenberg, Duncan, Fuest, & Peichl, 2012).

Our survey did not ask respondents about their ethnicity, but did record their main language. Here we find some major differences: Yoruba speakers have significantly lower tax morale than others, while Hausa speakers have much higher tax morale. This may reflect the legacy of colonial policies towards taxation in different parts of the country (Meagher, 2016). The correlations with language are consistent with those found by geopolitical region, where northern regions (which are largely Hausa speaking) tend to have higher tax morale than the South-West (mostly Yoruba speaking) and, particularly, the South-South. There may also be a political element to these variations – the current government is led by a President from the North and a Vice President from the South-West; the major group excluded from the current political settlement are people from the South-South and South-East. Another possible reason for the low tax morale of respondents in the South-South region may be the concentration of oil production in those states. The literature suggests that countries where the political settlement is reliant on rents from natural resources (or aid) find it harder to construct a social contract based on taxation (Elbahnasawy, 2020; Luciani & Moerenhout, 2020; Prichard, Salardi, & Segal, 2018). Our results suggest that the same may be true within countries too, with Nigerian states dependent on oil revenue having significantly lower tax morale than others. Interestingly, we find very little difference in tax morale between Christians and Muslims, again suggesting that it is historical and regional policies, rather than individual norms that are driving the differences.

5.2. Attitudes and beliefs associated with tax morale in Nigeria

Beyond the socioeconomic correlates, we find five factors that appear to be clearly associated with higher tax morale in Nigeria: actual payment of income tax; social norms; greater enforcement; higher levels of trust in tax officials; and better information.

First, the fact somebody has paid some form of income tax is strongly associated with higher tax morale. This is hardly surprising – we would expect people with higher tax morale to pay more tax – but equally the reverse might also be true that is participation in the tax system might raise tax morale. We highlight this because we find virtually no relationship between tax morale and people’s perceived liability for tax, only with their actual payment, which suggests that mere knowledge of an obligation may not change attitudes, whereas actual payment might. Much of the literature on the factors that lead to higher tax morale ignores the potential influence of actual payment. Our results suggest that future work could usefully explore whether compliance in and of itself affects an individual’s tax morale.

Second, social norms matter; we find a strong positive correlation between the belief that other Nigerians pay taxes and tax morale. However, this strong association is only found for Nigerians that do not pay some form of income tax. This suggests that people that do not pay income tax, either project their beliefs about the acceptability (or not) of non-payment onto the rest of the population, or their own beliefs are influenced by their perception of others’ payment. Conversely, those that do pay tax hold views about their own payment which differ from their views about the behaviour of the general population. For example, they may believe that they should pay their taxes, while simultaneously being of the view that most people avoid paying taxes (or vice versa). Again, this is suggestive evidence for the internalisation of a tax payment norm through the practice of payment.

Third, we find a strong confirmation of how neoclassical factors are positively associated with tax morale. We would expect that people who perceive a high penalty for non-payment would be more likely to pay (Allingham & Sandmo, 1972; Lopez-Luzuriaga & Scartascini, 2019) and, indeed, this is the case. However, we also show that people who believe that there will be a high penalty for tax avoidance have higher tax morale. This is in line with the experience in other developing countries such as South Africa (Mascagni, Nell, & Monkam, 2017) and Sri Lanka (Jayawardane & Low, 2016). This correlation is not dependent on income or whether people paid income taxes or not.

It is also the case that people who believe that avoiding taxes is difficult to have higher tax morale. In this case, the correlation is more pronounced among people who have not paid income tax or who have a lower income. As a result, while the perceived difficulty of avoiding payment is strongly associated with tax morale – it is not correlated with actual (self-reported) payment. The literature also suggests that people who perceive the burden of tax to be high are often more likely to believe that non-payment is acceptable, especially in the absence of perceived community payment (Górecki & Letki, 2020; Molero & Pujol, 2012). We find confirmation of this effect for those that pay income tax. We would also expect those who are less able to avoid tax (and therefore more likely to pay) to regard the tax burden as heavier than those who can avoid taxes more easily. This is also confirmed in our data with almost a quarter of those that paid income tax saying that the tax they are required to pay is a significant burden, against only 13 per cent of those that did not pay. In short – penalties, the likelihood of being caught, and the perceived burden of taxation have precisely the effect on tax morale predicted.

Fourth, we find evidence of a strong association between tax morale and trust in state tax officials. Over half of Nigerians say that tax officials ask for bribes (with around a quarter saying that this happens often or always). Unsurprisingly, 37 per cent of Nigerians do not trust state tax officials at all to collect revenue fairly, with a further 42 per cent that only trusts them a little. For those that pay income tax, we find a clear positive correlation between trust and tax morale. For those that do not pay income tax, there is a clear association between the frequency of bribes and low tax morale. Both of these findings are consistent with literature from other countries (Aiko and Logan, 2014; Jahnke & Weisser, 2019; Torgler et al., 2008). Our results are also consistent with the literature on the experience of taxation in Nigeria which suggests that it is a far from a transparent or fair process (Meagher, 2016). Respondents in our focus group discussions expressed frustration at the lack of transparency and fairness of in-person collection of taxes related to market fees, shop premises and property and land use taxes. The problem of trust in the tax administration is also consistent with experiences in other African countries (Ali et al., 2014; Fjeldstad, 2004; Isbell, 2017; Jahnke & Weisser, 2019; OECD, 2019).

Finally, better information about taxation – both the ease of finding out what taxes to pay and having received communication from the government about tax – are correlated with tax morale. However, in Nigeria this association only appears to hold for men and the overall association between information and tax morale is quite weak. The importance of tax knowledge has been found in other studies (Jayawardane & Low, 2016; Mascagni et al., 2017), including the suggestion that tax knowledge might be more important for male tax morale (Fallan, 1999).

The above results are all consistent with theory and with findings from studies in other countries. However, we also obtain two results which are quite different.

The first relates to fairness. There is ample evidence internationally and from Nigeria (Akdede et al., 2016; Bizhan, 2017; Meagher, 2016; Robinson, 2012) that tax officials do not always treat all taxpayers equally. Theory would suggest that favouritism towards some groups over others leads to resentment and therefore lower tax morale. We therefore asked respondents about their perceptions of differential treatment of three groups of people: different ethnicities; people from different religions; and people of different gender. To our surprise, we found a strong and consistent negative correlation between the perception that tax officials treat groups equally and tax morale. This is true for all three dimensions of difference – ethnicity, religion and gender. Why might this be? Part of the answer lies in Table 4 which shows the proportion of people in each group that believe treatment by tax officials favours one group over another, along with their level of tax morale.

Table 4. Discriminatory tax treatment

Group	Percentage who believe that tax officials favour some groups over others	Tax morale
Language	Favour by ethnicity
English	31.54%	36.85%
Yoruba	16.86%	30.40%
Igbo	50.77%	33.29%
Other	23.77%	35.18%
Religion	Favour by religion
Christian	35.96%	35.04%
Muslim	20.22%	34.09%
Traditional	32.89%	49.20%
None	38.65%	14.78%
Gender	Favour by gender
Male	27.35%	36.21%
Female	30.01%	32.90%

Notes: Tax morale is the percentage of respondents in the group that disagree with the statement ‘I would not pay my taxes if I would not be caught.’

Table 4 shows that more than half of Igbo speakers believe that tax officials do not treat people of different ethnicity in the same way – far more than any other group. However, Igbo speakers, together with Hausa and English speakers tend to have higher tax morale. Conversely, most Yoruba speakers overwhelmingly believe that tax officials treat all ethnicities equally¹⁷ but also have lower tax morale, giving rise to the negative correlation observed. Similarly, Table 4 shows that Muslims have a much stronger perception that tax officials treat people of different religions equally, than Christians or adherents to traditional religions. However, their tax morale is marginally weaker, which may partially explain the observed correlation. Finally, women have a stronger perception than men that tax officials do not treat men and women in the same way. However, as noted above, women in Nigeria have weaker tax morale, so this fails to explain the negative correlation between fairness of treatment and tax morale. In fact, the negative correlation between equal treatment of men and women and tax morale is more likely to be explained by religion. Only 22 per cent of Muslims believe that there is unequal treatment of men and women by tax officials, whereas a third of Christians believe this. This large difference, plus the slightly higher tax morale of Christians may help to explain the observed correlation. We also find that the correlation between unequal treatment and tax morale is much stronger for those who are members of religious groups, suggesting that religiosity may both prompt higher tax morale and a belief that people are treated unequally. However, these potential explanations should be considered as hypotheses for further research at best, since they rely on underlying and imperfectly explained correlations between ethnicity, religion, gender and tax morale.

The second unexpected finding is the almost complete lack of any association between tax morale and the perceived quality of service delivery and service improvements. Much has been written about how taxation may help to build the social contract in which citizens expect service delivery in return for paying their taxes (Ali et al., 2014; Moore, Prichard, & Fjeldstad, 2018; Prichard, 2015). If this was true in Nigeria as a whole, we would expect a relationship between tax morale and better services, either because higher tax morale gives rise to higher tax payment which enables better services; or perhaps because better services make people more favourably disposed towards tax payment. Previous work in Nigeria has suggested that state service provision leads to higher levels of tax morale and tax payment (Bodea & LeBas, 2016; Cheeseman & De Granmont, 2017). However, these papers used measures of tax morale that ask respondents whether they believe that citizens in general have an obligation to pay taxes, or whether non-payment by others is ‘wrong and punishable’. When we use such a measure, we also obtain a strong, positive association between service delivery (and improvement) and tax morale.¹⁸ However, our preferred measure of tax morale explicitly refers to the respondent’s belief about their own payment (or non-payment) of taxes. This does not appear to be associated with perceptions of service delivery or improvements. Thus, service delivery appears to increase the generalised norm that tax payment is a good thing if services are delivered – which is also the view expressed in our focus groups – but this does not necessarily translate into a belief that the individual themselves should always pay. Nigeria may have made less progress in constructing a social contract based on taxation than previously thought.

6. Conclusions and implications for policy and research

It is widely accepted that building an effective taxation system requires the inculcation of a norm of tax payment supported by a positive attitude towards payment that is high tax morale. As elaborated above, the literature has put forward several hypotheses regarding the factors that may be associated with higher tax morale including: better information, individual and social norms, trust in government and fairness of administration, and ideas of reciprocity and the social contract.

Using a new, comprehensive and nationally representative dataset, we confirm that tax morale in Nigeria is first and foremost associated with actual payment itself and better enforcement. We also find that it is associated with stronger social norms, trust in tax officials and better information. Although we are not able to prove causal relationships, these associations naturally suggest traditional policy interventions including broadening the tax base, improvements in enforcement, building greater social awareness of taxation, tackling corruption, and better communication.

However, our results also throw up two challenges both for policy and further research. First, we observe a strong association between the belief that people of different ethnicities, religions and gender are treated differently and higher tax morale. This is the opposite of the result that we expected and likely points to the different history and culture of taxation in different parts of Nigeria. While we provide a partial explanation, we look to further research to provide a richer contextual explanation for this finding. More generally, given the dramatic differences in tax morale by region and ethnicity in Nigeria, it suggests that policy should take more account of the way in which history, culture and religion in different places have shaped people’s understanding of citizen-state relations and how this is likely to affect the collection of tax.

Second, we find no connection between perceptions of service delivery and a measure of tax morale that is based on an individual’s own intention to evade tax (or not). Given the emphasis in the literature on taxation as a mechanism for building a social contract, this is surprising and slightly worrying. We find that, at least in Nigeria, the traditional association between service delivery and tax morale relies on the use of a measure of tax morale based on people’s opinions of what others ought to do, whereas this association disappears when individuals are asked to consider their own inclination to pay. It would be useful to explore whether this finding is replicated in other countries and to consider in more detail the implications for theories about taxation and the evolution of a social contract. This finding also has potentially important policy implications; if improving local services does not raise tax morale, policymakers may look to other mechanisms to boost revenue. Going forward, it will be important to understand the reasons why this relationship appears to be absent in Nigeria (for example, whether this is due to high natural resource rents which would imply the same result might be observed in other resource-rich countries) and what, if anything, might be done to establish a stronger connection between service delivery and the willingness of citizens to pay their taxes.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplementary Materials are available for this article which can be accessed via the online version of this journal available at https://doi.org/10.1080/00220388.2020.1797688.

Additional information

Funding

This work was supported by a grant from the International Centre for Tax and Development. The survey on which this paper is based was funded by the Bill and Melinda Gates Foundation in collaboration with the Nigeria Economic Summit Group and implemented by Practical Sampling International. The data can be accessed at https://www.ictd.ac/dataset/nesg-nigeria-tax-subsidy-perception-dataset/; the code for the analysis is available from the authors on request. Tom Moerenhout also expresses gratitude to the Swiss National Science Foundation (Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung). The views expressed are those of the authors and do not necessarily reflect those of any funder or supporter. All remaining errors are our own.

Notes

1. The countries are a selected sample of advanced, emerging and developing countries chosen by the IMF (IMF, 2018).

2. The government is still far from this target – non-oil revenue before the onset of COVID-19 was projected to be 3.7 per cent of revenue in 2020 (IMF, 2019).

3. This assumes a population of 190 million Nigerians; 54.3 per cent aged between 15-64; 52.65 per cent youth unemployment; 18.8 per cent unemployment and 21.2 per cent underemployment – see https://www.cia.gov/library/publications/the-world-factbook/geos/ni.html and http://www.nigerianstat.gov.ng/.

4. Of course, non-payment doesn’t necessarily mean non-compliance since there may be nil filers. However, a large majority of respondents in our survey have no connection with the tax system and most of those that do, do not file returns, so this distinction is less relevant here. Further work on administrative tax data would be useful to identify the scale of nil filing in Nigeria.

5. This figure probably reflects those registered to pay Personal Income Tax with the Federal Government; most PIT taxpayers are registered with State governments. The numbers paying Corporate Income Tax (CIT) are also low according to the ISORA survey; the IMF estimate slightly over one million firms are registered for CIT, of whom only 5.6 per cent are active taxpayers. See data.rafit.org for more details of the survey.

6. See http://afrobarometer.org/ and http://www.worldvaluessurvey.org/wvs.jsp for details.

7. The data collection process was coordinated by Neil McCulloch and Tom Moerenhout and funded by the Bill and Melinda Gates Foundation in support of the work of the Nigeria Economic Summit Group. The Nigeria-based survey firm Practical Sampling International conducted the data collection. The survey and FGDs were piloted in May 2018, and the data-collection exercise was conducted in July and August 2018. In addition, a survey was conducted on the attitudes of small firms towards taxation. The full data and documentation are available on https://www.ictd.ac/dataset/nesg-nigeria-tax-subsidy-perception-dataset/.

8. North Central, North East, North West, South East, South South, South West.

9. See the Field, Technical and Methodology Report available from https://www.ictd.ac/dataset/nesg-nigeria-tax-subsidy-perception-dataset/ for details, including the criteria for the selection of states within each region.

10. In addition to the quantitative surveys and FGDs, we also conducted a set of in-depth interviews with senior tax officials at the state and Local Government Authority (LGA) level. The structure and nature of these interviews, however, are not relevant to this particular analysis.

11. We also explored the use of the variable used in the Afrobarometer survey ‘Which of the following options is closest to what you think about people not paying taxes on income?’ with options: ‘Not wrong at all’; ‘Wrong but understandable’; and ‘Wrong and punishable’. Although this variable is commonly used to study tax morale, we are grateful to one referee for pointing out that it may relate to a different concept since it refers not to what the respondent does, but what they think about the actions of others. The Supplementary Materials show results using this alternative variable.

12. We also conducted analysis using type of job, in place of employment status, and state, in place of geopolitical region. Results are available on request.

13. We also made sure that our tax morale variable was not dichotomous. Respondents had the ability to ‘neither agree, nor disagree’ with the statement about potential non-payment of taxes.

14. Using the Afrobarometer definition of tax morale – see Supplementary Materials.

15. We report both because the correlation coefficient shows the tightness of the association, while the regression coefficient shows the slope of the relationship. Neither imply any causal association.

16. The Supplementary Materials also show these correlations and regressions for sub-samples broken down by: liability for income tax; whether the state is an oil state; gender; income; and whether the respondent paid income tax. We report the key results in the text.

17. Indeed, Grossman and Honig (2017) suggest that market traders discriminate more by class than ethnicity in Lagos.

18. See Supplementary Materials for details.

Appendix A.

Descriptive statistics

Table A1. Descriptive statistics of our main explanatory variables

Explanatory category/variable	Mean	STDV	Min	Max
Neoclassical
Penalty when getting caught for not paying taxes (perception)	1.04	0.94	0	3
Difficulty of avoiding paying income or property taxes	1.91	0.81	0	3
Lightness of tax burden	1.88	1.12	0	3
Information and knowledge
Ease of finding out what taxes and fees should be paid to the government	1.40	0.82	0	3
Received communication from the government in the last year about what taxes to pay	0.11	0.32	0	1
Tax liability and payment
Required to pay personal income tax (PAYE) – awareness	0.39	0.49	0	1
Paid some form of income tax last year	0.19	0.40	0	1
Required to pay property taxes – awareness	0.31	0.46	0	1
Paid some form of property tax last year	0.07	0.25	0	1
Social and personal norms
Membership of religious group	0.41	0.49	0	1
Frequency Nigerians pay their tax (perception)	1.57	0.96	0	3
Community pay income tax	2.52	1.22	0	4
Community pay land us/property tax	2.05	1.20	0	4
Trust and Fairness
Do tax officials treat all ethnic groups the same?	0.70	0.46	0	1
Do tax officials treat all people from all religions the same?	0.68	0.47	0	1
Do tax officials treat all men and women the same?	0.68	0.47	0	1
Infrequency of tax officials taking a bribe	2.17	0.96	0	3
Trust in Federal tax officials to collect tax fairly	0.85	0.86	0	3
General trust in government	0.83	1.15	0	3
Approval of President Buhari	1.44	1.49	0	4
Reciprosity and service delivery
Services satisfaction (index)	1.74	1.10	0	4
Quality of government services provided by State Government compared to 3 year ago	0.73	0.78	0	2

Appendix B.

Table A2. Proportions of responses to the survey questions used in the analyses

	Proportion	Std. Err.
Penalty when getting caught for not paying taxes (perception)
Nothing	0.293	0.006
Small penalty	0.410	0.007
Medium penalty	0.210	0.006
Large penalty	0.087	0.004
Difficulty of avoiding paying income or property taxes
Very easy to avoid	0.072	0.004
Easy to avoid	0.207	0.006
Difficult to avoid	0.549	0.007
Very difficult to avoid	0.172	0.005
Lightness of tax burden
A very significant burden	0.166	0.005
A burden but not a major one	0.324	0.006
Not a significant burden	0.217	0.006
Doesn’t pay tax	0.293	0.006
Ease of finding out what taxes and fees should be paid to the government
Very difficult	0.073	0.004
Diffucult	0.476	0.007
Easy	0.353	0.007
Very Easy	0.098	0.004
Received communication from government in the last year about what taxes to pay
No	0.831	0.005
Yes	0.169	0.005
Required to pay personal income tax (PAYE) – awareness
No	0.553	0.007
Yes	0.447	0.007
Paid some form of income tax last year
No	0.736	0.006
Yes	0.264	0.006
Required to pay land use or property taxes – awareness
No	0.646	0.007
Yes	0.354	0.007
Paid some form of property tax last year
No	0.905	0.004
Yes	0.095	0.004
Membership of religious group
No	0.592	0.007
Yes	0.408	0.007
Frequency Nigerians pay their taxes
Never	0.163	0.005
Rarely	0.317	0.006
Often	0.374	0.007
Always	0.145	0.005
Community pay income tax
Never	0.070	0.003
Hardly	0.102	0.004
Sometimes	0.319	0.006
Most of the time	0.250	0.006
Always	0.258	0.006
Community pay land use/property tax
Never	0.113	0.004
Hardly	0.180	0.005
Sometimes	0.367	0.007
Most of the time	0.212	0.006
Always	0.129	0.005
Do tax officials treat all ethnic groups the same?
Do not treat the same	0.303	0.006
Treat the same	0.697	0.006
Do tax officials treat all people from all religions the same?
Do not treat the same	0.327	0.006
Treat the same	0.673	0.006
Do tax officials treat all men and women the same?
Do not treat the same	0.307	0.006
Treat the same	0.693	0.006
Infrequency of tax official taking a bribe
Always ask	0.067	0.003
Often ak	0.189	0.005
Occassionally ask	0.282	0.006
Never ask	0.463	0.007
Trust in State Tax Officials to collect tax revenue fairly
No trust at all	0.365	0.007
Trust a little	0.415	0.007
Trust somewhat	0.167	0.005
Trust a lot	0.053	0.003
General trust in government
No trust at all	0.525	0.007
Trust a little	0.148	0.005
Trust somewhat	0.136	0.005
Trust a lot	0.191	0.005
Approval of performance by President Buhari
Strongly disapprove	0.382	0.007
Somewhat disapprove	0.198	0.005
Neither approve nor disapprove	0.081	0.004
Somewhat approve	0.204	0.005
Strongly approve	0.135	0.005
Quality of government services provided by State Govnmt compared to 3 yrs ago
Worse than before	0.473	0.007
About same	0.334	0.006
Better than before	0.193	0.005