The Politics of Taxation and Tax Reform in Times of Crisis: Covid-19 and Attitudes towards Taxation in Sierra Leone

Abstract The pandemic has had significant fiscal implications around the world. A key question facing governments is how the pandemic has shaped taxpayer attitudes and what that means for the prospects for tax reform and new revenue raising. We aim to understand the impacts of the pandemic on attitudes toward taxation in Sierra Leone with novel survey data, collected before the pandemic, shortly after the pandemic’s onset, and for almost a year afterwards. Four key findings emerge. First, immediately after the crisis onset we see increased support for taxation in Freetown, despite escalating economic challenges. Second, at the same time that taxpayers show greater general support for taxation they become more likely to believe that one could refuse to pay taxes if government fails to deliver services in return. Third, while we lack baseline data on support for progressive taxation, we find rising and sustained support for it over the course of the pandemic. Finally, although we see an initial increase in willingness to pay more taxes for services, that support erodes over time. These findings have significant implications for understanding both immediate responses to the pandemic, and the broader politics of taxation and tax reform.


Introduction
The Covid-19 pandemic has had significant fiscal implications around the world, with substantial increases in expenditures to finance health systems and social protection and large decreases in tax revenue resulting from the economic slowdown.In low-income countries the fiscal situation is particularly pressing given relatively weak fiscal foundations, limited borrowing options, and significant revenue losses from the pandemic (e.g., ATAF, 2021; Bachas, Brockmeyer, & Semelet, 2020;Mascagni & Lees, 2021;Ng'etich, Kilonzo, Oure, Njagi, & Kabiru, 2021).
Against the background of the urgent revenue pressures facing lower-income countries, a key question facing governments is whether and how the pandemic has shaped taxpayer attitudes and prospects for tax reform.On one hand, crises have sometimes catalysed expansions of taxation, and of more progressive forms of taxation in particular (e.g., Aidt & Jensen, 2009;Gupta & Jalles, 2022;Hartley, Sheffrin, & Vasche, 1996;Waris, 2021).This reflects that fact that crises can generate political space for reform: they can put pressure on governments to increase revenue while simultaneously driving greater public acceptance of taxation amidst an enhanced sense of social solidarity and tangible examples of the importance of public revenues.On the other hand, amidst protracted economic hardship taxpayers may become more resistant to tax demands, sparking re-distributional conflict.Both dynamics may be relevant in the case of the Covid-19 pandemic, which has sparked sharp new revenue pressures and explicit international discussions of 'solidarity' taxes but has also generated severe economic hardship, particularly in lower-income countries.
We aim to understand the impacts of the Covid-19 pandemic on attitudes toward taxation in a low-income context and, in turn, to unpack what the crisis reveals about the dynamics and politics of taxation more broadly.Using two existing sampling frames in Sierra Leone, we conducted regular phone surveys with over 1,700 respondents beginning shortly after the onset of the pandemic and continuing for almost a year.We build on a baseline survey conducted in Freetown shortly before the onset of the pandemic, which contained explicit questions about attitudes toward taxation and thus allows us to directly compare attitudes before and after the beginning of the pandemic.
We first use this unique dataset to shed light on taxpayers' varied crisis experiences, focusing on the economic impacts of the crisis and the scale and reach of government relief efforts.Three key messages emerge.First, we find large and negative impacts on economic well-being across the population.Second, we highlight the relatively limited reach of government relief efforts and, in particular, the limited extent of tax relief.Third, we document that access to reliefand, in particular, to relief through the tax systemhas a strong elite bias, in part reflecting the narrow reach of the tax system and its consequent limitations as a means to deliver broad-based support to citizens.
This provides an empirical foundation against which to assess and understand changes in attitudes toward taxation.We highlight three interconnected messages.First, immediately after the onset of the crisis and despite escalating economic challenges, we see evidence of increased support for taxation in Freetown: support for the government's right to tax remains constant, trust in local governments' ability to collect taxes fairly increases, and willingness to pay more taxes in return for greater services increases.However, taxpayers also express increasingly conditional attitudes toward taxationthat is, at the same time that they show greater general support for taxation, taxpayers become more likely to believe that they could refuse to pay taxes if government fails to deliver services in return.This in line with recent evidence of changes in taxpayer attitudes during the pandemic in Rwanda (Mascagni & Santoro, 2023).Second, although we see an initial increase in willingness to pay more taxes for services immediately after the onset of the pandemic, we find evidence of that support eroding over time, potentially reflecting a combination of continued economic hardship, declining feelings of social solidarity, and some disappointment with government taxation.Finally, while we lack baseline data before the pandemic on support for progressive taxation, we find rising and sustained support for more taxes on the rich over the course of the pandemic.
These findings have potentially significant implications for understanding both immediate responses to the pandemic and the broader politics of taxation and tax reform.While there is some evidence of increased support for taxation initially after the onset of the pandemic, two years later general support for taxation looks similar to the pre-pandemic period.Yet two shifts appear more durable.First, particularly in Freetown there appears to have been a sustained shift in the extent to which taxpayers hold conditional attitudes toward tax compliance, with increased expectations of effective service delivery.Second, we see a sustained shift toward greater support for taxing the rich and larger companies.This indicates that post-pandemic tax strategies may receive greater support if they emphasise comparatively progressive taxes and The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 197 seek to explicitly demonstrate connections between new tax revenues and expanded service delivery.
More broadly, these results are suggestive of a simple but compelling story about the complex politics of taxation in times of crisis.We see evidence of the emergence of a political window of opportunity for tax reformand progressive tax reform in particularimmediately following the onset of the pandemic.We speculate that this is attributable to an initial sense of social solidarity in the face of a national challenge and increased expectations about what the government should do in times of crisis.However, that window of political opportunity was not unconditional: amidst economic challenges taxpayers appear to have been willing to support the collective good, but only if they saw evidence that tax revenues were being used to provide a meaningful response to the crisis.In practice, that window of opportunity appears to have been fleeting, as by the end of the pandemic public support had largely dissipatedthough with a sustained increase in support for taxing the rich.This experience largely echoes the broader literature on the politics of taxation (e.g., Prichard, 2019): moments of crisis can play a critical role in opening windows of opportunity (Sanchez, 2006), but those opportunities are often briefand exceedingly difficult for governments to 'harness' in a short period, given the practical challenges of passing and implementing tax reforms quickly.Evidence from elsewhere suggests that successful reform depends on governments acting decisively to build political coalitions for reform, including by seeking to demonstrate fairness and equity in new taxation (Fairfield, 2013) and strong connections between revenue and service delivery (Jibao & Prichard, 2015), though there remain questions about how to sustain support for reform in contexts where public spending is severely constrained, particularly in times of crisis when public finances are at their most strained.
Nevertheless, evidence of a sustained increase in support for progressive taxation and higher expectations of service provision in Sierra Leone sheds light on potential channels to overcoming low tax-low capacity traps and barriers to contractual taxation.As Mascagni and Santoro (2023, p. 813) note, shifts towards a more contractual view of taxation may be seen as 'a first step to spur a potential virtuous cycle whereby taxpayers are more willing to pay tax, but also demand better services in return'.We explore these possibilities in the remainder of the paper.

Background: tax reform in the context of crisis
In high-income countries significant shifts in attitudes toward taxationincluding increased willingness of citizens to accept new taxes and support for progressive tax reformshave been linked to periods of crisis, including wars and economic recessions (e.g., Aidt & Jensen, 2009;Emmenegger & Lierse, 2022;Hartley et al., 1996;Limberg & Seelkopf, 2022;Pollack, 2014;Scheve & Stasavage, 2016;Waris, 2021).In the context of Covid-19, there is some evidence from high-income countries that the associated economic upheaval can shape individuals' views of taxation and make them more supportive of temporary progressive levies (e.g., Klemm & Mauro, 2022).More generally, there has been greater attention paid around the world to wealth taxes as a means of financing pandemic relief and recovery, sometimes termed a 'solidarity tax' (e.g., UN Committee of Experts on International Cooperation in Tax Matters, 2021).
While experience in wealthier countries thus points toward the possibility that the Covid-19 pandemic may drive significant shifts in attitudes toward taxationand toward progressive taxation in particularwe know little about these potential dynamics in low-income countries.These dynamics may be different in lower income contexts given colonial histories of taxation, the relatively limited reach of direct taxation, and generally lower trust in governments.On one hand, we may see positive shifts in attitudes toward taxation amidst an expanded sense of social solidarity and evidence of the importance of public spending in responding to the crisis.There is some evidence of this, with Gupta and Jalles (2022, p. 4) finding that within 45 low-income countries 'past pandemics have propelled countries to implement tax reforms', particularly in relation to corporate income taxes, excises and property taxes.On the other hand, significant economic hardship, coupled with relatively limited pandemic relief, or concerns about government performance, could lead to reduced willingness to pay taxes, reduced support for taxing others, and even civil unrest or social backlash, as was recently seen in the case of Colombia (Janetsky, 2021).This may be more likely where there is limited trust in the state and where elites contribute to social welfare efforts through informal taxation and other forms of nonstate social welfare.
Meanwhile, there is limited understanding of taxpayer perceptions of and support for progressive tax reform in low-income countries. 1We may expect popular support for progressive taxation and redistribution in lower-income countries, given high income and wealth inequality and generally weak taxation of wealth.However, evidence suggests that there is often limited support among poorer people for redistributive policies from which they would benefit (Holland, 2018;Hoy & Mager, 2021;Kuziemko, Buell, Reich, & Norton, 2014;Roemer, 1998) and a general lack of popular concern about inequality despite its increase (Mijs, 2021).One possible reason for these apparent puzzles is a general lack of trust in the state and in the fairness of the tax systemin Sierra Leone, illustratively, the majority of citizens believe that elites are easily able to avoid taxes (Afrobarometer, 2018). 2

Data
To gain a deeper understanding of the impact of the Covid-19 pandemic on attitudes toward taxation we build on two existing surveys.The first captured a random sample of property owners in Freetown and included several questions about attitudes toward taxation. 3This survey was carried out shortly before the onset of the pandemic in Sierra Leone, thus allowing us to track changes in responses in the pre-and post-pandemic periods. 4The second survey, Gilchrist, Grieco, Meriggi, Mobarak, and Voors (2020), was a study of the impact of electrification on welfare indicators and captured a representative sample of households in 94 rural communities across all 14 districts in Sierra Leone outside of Western Area. 5 This survey did not include questions about taxation at baseline, and so only allows for tracking changes in experiences and attitudes after the onset of crisis in March 2020.Unsurprisingly, these two populations differ with respect to demographic characteristics (age, education, occupation, wealth) and knowledge about Covid-19 (Appendix, Table A1 and Figure A1).
We used a combination of these sampling frames to collect data throughout 2020 and 2021.Data was collected via phone interviews, 15-30 minutes in length. 6Interviews were conducted periodically, with respondents called between every four to six weeks, and targeted heads of households or individuals who were most knowledgeable about household affairs.Owing largely to respondent fatigue, not all respondents were successfully surveyed in each survey wave.We correspondingly average responses by respondent for four time periods, as follows and as illustrated in Owing to administration challenges only half of respondents were attempted during Round 4, with meaningful differences between respondents who were surveyed during this phase and The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 199 those who were not. 7As a consequence, we focus our analysis on data from the first three rounds, though we present data from the fourth round throughout. 8 466 Freetown respondents (out of 901 baseline property owners) and 233 provinces respondents (out of 3,075) were interviewed at least once in each of the periods and were included in the data analysis. 9The averaging of responses within each period appears to pose limited risk to data quality given that we are primarily exploring slow-changing attitudes, and that each round corresponds to a key phase of pandemic dynamics.Recognizing that previous tax payment experience may affect some of the perceptions questions we are interested in, we capture respondents' previous tax paying history.Unsurprisingly, tax payment at the local government level is much higher than to the national government, and tax payment is higher in Freetown than in the provinces.When asked what taxes they normally pay (outside of the pandemic context), 35% and 92% of the sample in Freetown indicated that they paid some tax or fee to the national and local government, respectively; these figures were much lower in the provinces: 20% and 50%, respectively. 10

Taxation, Covid-19, and access to relief in Sierra Leone
We focus on the case of Sierra Leone because of access to unique tracking data before and during the pandemic, but also as the government had recently undertaken significant tax reforms making it an interesting context within which to explore perceptions of taxation.In recent years, the Sierra Leonean government has made major improvements in tax administration and collection which have improved revenue collection (IMF, 2019(IMF, , 2021)).While its tax-to-GDP ratio was still low at 12.3% in 2021, this is in line with the average for sub-Saharan Africa (12.85%) and low-income countries (11.54%) (UNU-WIDER, 2022).Personal income taxes are now the most important source of taxation in the country (Appendix, Figure A2) and on average from 2010 to 2021 represent a higher percentage of GDP (3.30%) compared to the lowincome country average (1.96%) and average for sub-Saharan African countries (2.96%) over the same period (UNU-WIDER, 2022).
Alongside tax reforms at the national level, the Freetown City Council launched a property tax reform program in 2019, which dramatically increased the progressivity of the local tax system and resulted in a roughly threefold increase in revenue in the first year of implementation (Wilson & Eleveld, 2022).The reform was accompanied by major outreach and engagement efforts, including the introduction of participatory budgeting, meaning that our surveys in Freetown capture changes in attitudes in the context of the pandemic as well as an increase in the political salience of taxation and tax burdens on our survey respondents.
Despite these relatively recent fiscal improvements at both the national and local level, state institutional capacity remains low while aid dependency remains high. 11The country has a large informal economy with few registered taxpayersin 2021, there were only 8945 registered taxpayers in the country (ATAF, 2022)which is likely to be a considerably higher figure than active taxpayersout of a population of 8.6 million and a labour force of 2.8 million (World Bank, n.d.)and even fewer active taxpayers: 3241 personal income taxpayers, 923 corporate income taxpayers, and 1911 VAT taxpayers (ATAF, 2022).Meanwhile, survey data from three districts in the provinces confirms that direct payment of taxes at the national government is low, with only 4 percent of respondents had paid any direct taxes to the central government in the previous year (van den Boogaard, Prichard, & Jibao, 2019). 12Overall, local government tax revenues in rural areas amounted to an average of only US$0.02 per capita from 2005 to 2017 and only US$0.40 in urban areas (van den Boogaard & Beach, 2023, p. 11), with the incidence of property tax payment particularly low outside of Freetown.In this context of relatively weak fiscal foundations, informal taxes have long been a way of filling gaps left by the state in essential public goods provision (van den Boogaard, 2020; van den Boogaard et al., 2019).
The Covid-19 crisis and related containment measures have had a significant impact on domestic revenue mobilization.While case numbers remained relatively low, the government enacted strict containment measures from mid-March through July 2020, though the stringency of these measures quickly lessened (Figure 1).The fiscal deficit almost doubled in 2020 as a result of a combination of revenue shortfalls and pandemic-related spending increases, alongside lower-than-expected external support (IMF, 2021;World Bank, 2021).
In aggregate the crisis led to significant economic impacts, while the weak fiscal landscape limited pandemic relief.Given the focus of containment measures, relatively few businesses had to close because of lockdown measures, while restrictions on businesses were largely inconsequential by the fall of 2020 (Appendix, Figure A3).Nevertheless, businesses were negatively affected economically, largely as a result of restrictions on travel and trade, reporting depressed revenues throughout much of 2020, increasing only in the fall and new year (see Supplementary Materials, Table SM1).
The economic impacts of the crisis are more broadly evidenced by a large majority of respondents reporting food insecurity, though at a decreasing rate over time in the provinces and with higher rates of food insecurity in Freetown (Appendix, Figure A4).By the fall of 2020, 97 percent of respondents reported being unable to meet financial obligations and essential needs.These findings are in line with evidence that the pandemic has increased poverty, food insecurity, and vulnerability and reduced living standards in Sierra Leone (Conteh, Sirah Kamara, Saidu, & Mustapha Macarthy, 2021;Egger et al., 2021;IMF, 2021;World Bank, 2021), with the largest increase in poverty seen in urban areas, particularly Freetown (World Bank, 2021).They are particularly stark given that prior to the pandemic, 57% of households already lived below the poverty line (Statistics Sierra Leone, 2019).
The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 201 Amidst this economic hardship the Sierra Leonean government and its development partners mobilised relief in a few key areas, building on an extremely limited social protection system. 13 New interventions included additional spending on health supplies, health care workers, quarantine measures, and an awareness campaign; one-time cash transfers of USD 135 to 29,000 households with informal workers, mostly traders, who were deemed most vulnerable to the pandemic; and support to people with disabilities during the lockdown (IMF, 2021).Meanwhile, the NRA instituted tax deferment procedures on customs duties and a 3-month tax deferral on income tax, focusing on industries most affected by the pandemic, while the FCC reduced late payment penalties on property tax payment.
Despite these efforts, we find that receipt of direct economic or tax reliefs was limited, with individuals instead more likely to receive support related to preventing the spread of Covid-19 (e.g., PPE or handwashing stations).Over the course of our data collection period, 35 percent of households and 31 percent of businesses reported receiving some type of relief, mostly in the form of sanitation supplies (of those that received relief, 86% received handwashing stations, known as Veronica buckets, and 46% received soap) and personal protective equipment (PPE) (of those that received relief, 42% received masks).By contrast, only a small proportion of those that received any relief received food aid (16%) or cash transfers (3%).This limited reach of reliefs, particularly of cash transfers, is in line with evidence of the challenges of rolling out new cash transfer programmes in the context of high informality (e.g., Adu-Ababio, Osei, Pirttil€ a, & Rattenhuber, 2021; Gallien & van den Boogaard, 2021).
Perhaps unsurprisingly given the low levels of registered taxpayers in the country, access to tax reliefs was limited, with only 29 percent of respondents receiving formal tax relief.The vast majority of this relief came in the form of delayed payments (96%), with only 8% seeing their tax payments waived and 3% having the amount reduced.Meanwhile, access to local tax relief was almost non-existent.This is in line with findings from elsewhere in sub-Saharan African showing that general tax-benefit systems failed to provide social welfare protection (Mascagni & Santoro, 2023), in large part because the large size of the informal economy means that 'only those in the formal sector, usually situated in the upper part of the income distribution, benefit from this mechanism' (Adu-Ababio et al., 2021. p. 2).
Overall we also see evidence that higher income individuals were more able to access relief, either because of a stronger ability to navigate government relief programmes, more extensive networks within government or their inclusion in the formal tax system (van den Boogaard et al., Forthcoming).Meanwhile, we see evidence that individuals across the country faced new informal taxes to help finance the pandemic response, reflecting the limited reach of state relief efforts (van den Boogaard et al., Forthcoming).Taken together, our data points towards sharp limits to the reach of economic relief.

Taxpayer attitudes and perceptions
With this understanding of the impacts of the pandemic, we turn to exploring how the pandemic shaped attitudes towards the government and taxation.First, we ask how attitudes toward the government and taxation changed between the pre-pandemic period and the period after the onset of the pandemic.Second, we ask how attitudes toward the government and taxation evolved over the course of the pandemic.Finally, we consider the impact of the pandemic on attitudes toward taxing higher income groups specifically.

Initial impacts of the pandemic on taxpayer attitudes
Using baseline data in Freetown alone, we compare 'before' and 'after' data in relation to four key indicators: unconditional tax morale, trust in local tax collection, willingness to pay more taxes for improved services, and tax morale conditional on representation. 14While imperfect, these questions provide a basis for painting a broad picture of changing attitudes in response to the pandemic in Freetown.We run fixed effect regressions on the four key indicators based on the following equation: where Y it is one of the key indicators, round1, round2, round3 and round4 are time dummies, income is self-reported wages or business profits and Zi are unobserved time-invariant heterogeneities such as gender and education.Fixed effects allow to not only control for all time-invariant omitted variables but also, by including time dummies, to eliminate omitted variable bias by excluding unobserved factors that may evolve over time but are constant across respondents.The coefficients of interests are b 1 , b 2 , b 3 and b 4 and are graphically represented in Figure 2 with their 95% confidence interval and includes a reference dotted line at zero, allowing us to clearly distinguish statistically significant changes from baseline levels.
Two key messages emerge in the first phase of the pandemic.First, we see overall steady support for general taxation ('Unconditional tax morale') despite the large-scale economic hardships caused by the pandemic among property owners in Freetown.As shown, the level of unconditional tax morale among respondents remains essentially unchanged immediately after the pandemic and thereafter. 15As a strong majority of respondents indicate support for unconditional tax moralein line with high unconditional tax morale at the national level (Afrobarometer R8, 2021)there is a ceiling effect on any further increase.This sustained high level of support can be taken as a strong signal of sustained support for tax compliance.This is reinforced by a large and significant increase in trust in the tax system immediately after the onset of the pandemic.The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 203 Second, and somewhat in contrast, we see a significant and large immediate increase in the extent to which property owners in Freetown view tax compliance as conditional on reciprocal delivery of benefits by government ('Conditional tax morale', 'Willing to pay for services').Taken in concert with sustained or increasing general support for taxation, this suggests a shift in the character of the relationship between taxpayers and the state.In response to the pandemic, taxpayers appear to experience an increase in their general level of support for taxation, consistent with narratives of social solidarity in response to crisis, though support for taxation becomes more conditional on government performance.While our data do not allow us to say what is driving the latter shift, it is consistent with the idea that taxpayers, confronted with both acute societal needs and significant economic pressures, increasingly focused on the responsibility of government to deliver concrete benefits with public revenues and, perhaps, also became more aware of failures to do so.More abstractly, this overall shift in attitudes appears to be reflective of taxpayers adopting a more 'contractual' view of taxation (Prichard, 2022), in contrast with recent ideas about 'tax aversion' being inherent in African states (Robinson, 2022).

The evolution of taxpayer attitudes over the course of the pandemic
Having captured a snapshot of changes in attitudes in response to the onset of the pandemic we then explore the evolution of taxpayer attitudes over the course of the pandemic.Looking first at general attitudes toward tax compliance we see evidence of a return to baseline over the course of the pandemic.As reflected in Figure 2, in Freetown support for the view that citizens should always pay taxes ('Unconditional tax morale') remains consistently high throughout the pandemic, with no evidence of decline over time.By partial contrast, after witnessing a significant increase in trust in the tax system immediately after the onset of the pandemic in Freetown we see levels of trust in the tax system return to baseline as the pandemic progresses (Figure 2, 'Trust in local tax collection').The most straightforward interpretation is that while general support for taxation remains universally high throughout the pandemic, initially high hopes that the pandemic would bring increases in the fairness of the tax system are to some extent unrealised over time.
We then consider the extent to which taxpayer attitudes become conditional on government performance over the course of the pandemic (Figures 3 and 4).We see striking patterns in Freetown, with a more immediate, but ultimately smaller, increase in beliefs that tax compliance should be conditional on government representing the interests of taxpayers, and a slower, but ultimately larger, increase in beliefs that taxpayers could refuse to pay taxes if they are not receiving public services of adequate quality.As discussed above, there seems to be evidence of the emergence of a more conditional, or contractual, attitude toward tax compliance in Freetown.We do not see any increases in the conditionality of attitudes toward compliance in the provinces over the course of the pandemic, but in the absence of baseline data prior to the pandemic, cannot draw any conclusions.
We further ask whether respondents would be willing to pay additional taxes to receive additional services (Figure 5).In Freetown we observe a small increase in reported support for paying additional taxes immediately after the onset of the crisis, while we see similarly high initial levels of support for paying additional taxes in the provinces.However, in both Freetown and the provinces we see a substantial decline in willingness to pay additional taxes over the course of the pandemic.The sharpness of the decline in Freetown is likely driven to some extent by the parallel introduction of increased property taxes during this period.But this is unlikely to be the only driver of this pattern: virtually all taxpayers would have been aware of the property tax reform prior to the first round of the survey and the majority of respondents would have also received a bill prior to the second round, when support for paying additional taxes remained relatively high.It correspondingly seems that after an initial bump in willingness to pay additional taxes in exchange for services, this support progressively erodes over time.
In seeking to explain declining support for paying additional taxes in exchange for greater services there are three interconnected possibilities.First, amidst increasing economic hardship declining willingness to pay may be driven primarily by an inability to pay more.Second, it  The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 205 could be a reflection of increasingly conditional support for taxation and concerns that tax revenues were not being adequately translated into public benefits.Our data provide mixed evidence of this possibility.Overall public satisfaction with the government response to the pandemic remains high and even increases over time (Appendix, Figure A5).However, as described above, in Freetown trust in taxation decreased over time (Figure 2), while perceptions of the fairness of both taxes and user fees declined in both Freetown and the provinces (Appendix, Figures A6 and A7).Third, it could be the result of a broader decline in a sense of social solidarity.Consistent with that possibility, we find that in the last round of our data collection 68% of respondents believed that people were less likely to help each other than prior to the pandemic and only 9% believing that people were more likely to help.In turn, 38% of respondents indicated that they felt less proud to be Sierra Leonean than previously.Overall, the most likely conclusion appears to be that the decline in support for paying additional taxes is a result of economic hardship, a related sense of declining social solidarity, and some disappointment with government taxation.

The evolution of attitudes toward progressive taxation
Finally, we explore taxpayer perceptions of various options for funding the recovery, in particular through progressive taxation.Our analysis is limited by the absence of baseline responses to our questions.However, we believe that changes over the course of the pandemic are nonetheless instructive, given that around the world debates about appropriate tax responses to the pandemic tended to occur after the onset of the pandemic.In all survey rounds we asked taxpayers how they think that the government should fund the pandemic response and recovery, whether through more taxes on everyone, more taxes on the richest, more taxes on large businesses, more taxes on properties, and additional foreign aid or loans.Unsurprisingly, in times of economic crisis reliance on external financing (aid and loans) is the preferred option (Figure 6).Of greater interest, however, is clear evidence of sustained increases in support for progressive taxation, 206 N. Orgeira Pillai et al.
through taxing the rich and taxing large businesses, relative to broad-based taxation ('more tax from everyone').In line with evidence from Rwanda (Mascagni & Santoro, 2023), over the course of the pandemic support for 'more tax from everyone' increases relatively modestly (though statistically significantly with a 5.2 percentage point increase, representing a 22% increase from round 1 levels), as compared to larger increases in support for 'more taxes on the rich' (20 percentage point increase, or a 37% increase from round 1 levels) and 'more taxes on large businesses' (11 p.p. increase, 28% increase from round 1 levels).
To further interrogate these findings we added an additional question during round 3 of our data collection asking respondents to choose which one of these tax types should be prioritised.Again, we see a preference for taxes on the rich and large businessesand also on properties in the provincesrather than a preference for broad based taxes (Figure 7). 16This trend is particularly striking in Freetown, where only 3% of respondents favoured expansion of broadbased taxes.We can imagine two possible explanations.First, the experience of a broadbased, yet progressive, property tax reform in Freetown may have fed support for even more progressive but more narrowly focused taxation among our sample of property owners.Second, support for taxing the wealthy and big businesses may be somewhat more muted in the provinces because of a lack of trust in the potential for redistributive taxation and stronger status quo bias.However, we are cautious not to overinterpret this data: in the absence of baseline data, or data about changes over time, these responses cannot tell us anything directly about the impact of the pandemic.

Conclusions
The Covid-19 pandemic and resultant fiscal pressures raised questions about whether the crisis may create political space for the expansion of progressive taxation.Yet there has been limited The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 207 exploration of attitudes toward progressive taxation in lower-income countries and drivers of changes in taxpayer attitudes over time.The data used here is imperfect in various respects, reflecting the empirical difficulty of capturing sudden changes in attitudes in the midst of crisis; it nonetheless begins to address these gaps and tells an intriguing story about the impacts of the pandemic on attitudes toward taxation.
In the first phase of the pandemic, we see sustained, or even modestly increased, support for general taxation, despite the economic challenges.However, that was accompanied by a sharp increase in the extent to which support for taxation was conditional on government reciprocity; that is, we see increasing expectations of reciprocity and the adoption of a more contractual view of taxation.Over the course of the pandemic, we see some erosion of that support: particularly in Freetown attitudes toward taxation become increasingly conditional, while willingness to pay additional taxes in exchange for services declines.We also observe a durable shift in attitudes toward progressive taxation, with increases in relative support for taxing the rich and large businesses over the course of the pandemic.
Our findings begin to tell a broader story about the politics of taxation in times of crisis.Consistent with some evidence from lower income countries (e.g., Gupta & Jalles, 2022), in Sierra Leone the pandemic seems to have created an initial 'window of opportunity' for strengthening the tax system and, in particular, for more progressive taxation.However, that window of opportunity also appears to have been both short-lived and conditional: amidst significant economic hardship willingness to pay taxes became more conditional on the government demonstrating that tax revenues are used effectively.Declining support for taxation as the pandemic progressed suggests that the government was not able to fully seize that window of opportunity for reform, amidst a government response that was able to provide relief to only a small part of the population.This reflects some of the broader challenges of harnessing windows of opportunity for reform, given the practical challenges of passing and implementing tax reforms quickly.
That said, there are some signs for optimism that the pandemic may open space for more effective, progressive or accountable tax systems.We see a sustained increase in support for progressive taxation, while unconditional tax morale does not decline despite substantial economic challenges and hardship.Perhaps of greater interest, the fact that citizens came to hold an increasingly contractual view of taxation, with expectations of service provision in exchange for tax payment, may in itself be a positive long-term outcome for the development of stronger fiscal contracts, with the potential to spark virtuous cycles of tax payment and service delivery.

Notes
1. Key exceptions are Sands and de Kadt (2020), who find that exposure to inequality was positively associated with support for a tax on wealthy individuals in South Africa and Mascagni and Santoro (2023), who find that the Covid-19 pandemic shifted taxpayers' attitudes towards more conditional notions of tax compliance.2. In 2018, 79 percent of individuals believed that a rich person could pay a bribe or use personal connections to get away with avoiding paying taxes they owe to the government, while only 29 percent believe the same to be true of ordinary people (Afrobarometer, 2018).In 2021, 64 percent of individuals believed that people in the country can often or always 'avoid paying the taxes that they owe to government' (Afrobarometer, 2021).3. The FCC collected information on all properties in Freetown and used it to calculate tax rates for each property, with many property owners included in the tax register for the very first time.These property owners include those whose contact details were available to allow enumerators to call them prior to the interview to increase the likelihood of being able to survey the property owners.As such, respondents were more likely to be readily available and may differ from respondents who were not available during the day/whose contact details were missing, implying that the sample may not be fully representative.4. Surveys before and after the start of the pandemic included tax morale questions that were phrased the same, allowing for comparability. 5.After a household census was carried, a representative sample of respondents was drawn from each village and the head of household or person most knowledgeable about household affairs was interviewed.6. Poorer baseline respondents and those who may not own phones and/or who live in remote areas with limited phone connection may be underrepresented, particularly in the provinces.Questionnaire is available upon request.7. Tables available upon request.8.We also replicate the analysis below by combining the first two 'During' periods into one from May 12 -October 26, which increases the sample in both Freetown (from 466 to 615) and more particularly the provinces (from 233 to 1,106).Not all respondents were successfully surveyed in the first round (1,102 out of 3,316), particularly in the provinces (only 483 out of 2,565 vs. 619 out of 751 in Freetown), which explains the final sample size and interview success rates.With this replication, we receive similar results to those reported below.
Tables and figures available upon request.9. We compare mean differences between respondents in Freetown who were only surveyed during the baseline with those who were surveyed in other subsequent rounds, finding that farmers and self-employed respondents were less likely to be surveyed while wage-employed and self-employed were more frequently interviewed.We also compare mean differences between respondents who were only surveyed during Round 1 and those who were surveyed at least once in Rounds 1 to 4. Broadly, poorer respondents were less likely to be interviewed in all survey rounds.Tables available upon request.10.This question was asked in the first round of data collection (May to July).National taxes paid include income tax and rental tax; local taxes include the local (poll) tax, property tax, market fees, business licences, mining licenses and power saw/ timber fees.11.Countries receiving more than 10 percent of gross national income are considered high-aid countries, reflecting aid dependency; since the end of the civil war in 2002, Sierra Leone's aid dependency rate has been on average almost twice that at 19 percent (van den Boogaard, 2020, p. 231).12.A large majority pay taxes to the local government on account of there being an annual poll tax (92 percent of respondents paid this poll tax), though this is collected by chiefs on behalf of the local government, limiting taxpayers' direct engagement with state institutions (van den Boogaard et al., 2019).13.Prior to the pandemic less than 10,000 people benefited from any form of social protection support from the government (Kamara, 2020), though over one million people (16% of the national population) are estimated to live in extreme poverty (Statistics Sierra Leone, 2019).The support that is delivered 'is not only inadequate and untimely, but also unpredictable', while there is 'no comprehensive database on the number of vulnerable people living in Sierra Leone' (Kamara, 2020) 14.Survey questions for each were phrased the same way 'before' and 'after' and are as follows: Unconditional tax morale: Do you agree or disagree that citizens should always pay taxes?Trust in tax collection: Do you agree or disagree that local government can be trusted to collect taxes fairly?Willingness to pay for services: Would you be willing to pay more taxes in order to receive improved services?Conditional tax morale: Do you agree or disagree that citizens should only pay if the government represents their interests?We use an indicator of trust in the local government rather than the central government in order to keep the question consistent with that used at baseline.Empirical evidence shows that Sierra Leoneans are more likely to pay taxes to and interact with the local government rather than the central government (e.g., van den Boogaard et al., 2019), with a The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 209 significant proportion of individuals being unable to accurately identify the government body providing services.Given these realities, we believe that using a measure of trust in local government is justified.15.Prior to the pandemic, 89 percent of individuals nationwide agree or strongly agree that 'the tax authorities always have the right to make people pay taxes'.16.While we are limited in our ability to explore how these perceptions vary by income (on account of a relatively small share of respondents reporting positive mean wages or profits and the limited number of wealthier individuals and businesses in our sample), we find that support for taxing the rich is lower among the richest and largest businesses, while support for taxing everyone is lower among the poorest (see Supplementary Materials, Figure SM1).The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 213   The politics of taxation and tax reform in times of crisis: Covid-19 and attitudes towards 215  216 N. Orgeira Pillai et al.

Figure 1 .
Figure 1.Data collection rounds, alongside new Covid cases and Covid stringency index.'New cases' are drawn from the John Hopkins University CSSE Covid-19 database and reflect the number of new confirmed Covid-19 cases per million people; what is shown is the rolling 7-day average.The number of confirmed cases is lower than the number of actual cases; the main reason for that is limited testing.The 'stringency index' is a composite measure based on nine response indicators including school closures, workplace closures, and travel bans, rescaled to a value from 0 to 100 (100 ¼ strictest).If policies vary at the subnational level, the index is shown as the response level of the strictest sub-region.Data is drawn from Hale et al. (2021).

Figure 3 .
Figure 3. Citizens should only pay taxes if the government represents their interests.

Figure 4 .
Figure 4. Taxpayers could refuse to pay taxes if they are not receiving public services of adequate quality.

Figure 5 .
Figure5.I would be willing to pay additional taxes in order to receive improved services.

Figure 6 .
Figure 6.Options for funding the recovery.

Figure A1 .
Figure A1.Difference between Province and Freetown respondents.We graphically coefficients and 95% confidence intervals of a regression of various respondent characteristics on a dummy variable to represent respondents from the province.The reference dotted line at zero allows to clearly identify statistically significant differences between Freetown and Province respondents.The sample size in the provinces and in Freetown are provided in the parenthesis.

Figure A4 .
Figure A4.Food insecurity (by round, surveyed in each round).Households are considered food insecure if they responded positively to any the following questions: 'Has your household: Eaten foods that you ordinarily would not eat i.e., 'less preferred foods'?Asked for help from relatives or friends?Reduced portions/quantities served per meal for adult males?Reduced portions/quantities served per meal for adult females?Reduced portions/quantities served per meal for boys under age 10? Reduced portions/quantities served per meal for girls under age 10? Eaten fewer times per day than normal for this time of year?Spent whole day without eating?'

Figure A6 .
Figure A6.Perceptions of fairness of formal taxes.

Figure A7 .
Figure A7.Perceptions of fairness of user fees.