1. Introduction

The economic analysis of government taxation and expenditure has had a major influence on policy debates and policy decisions. George Osborne’s decision as Chancellor of the Exchequer to lower the top rate of income tax, for example, was arguably influenced by James Mirrlees’ Nobel prize winning analytical framework [1], which precisely balances both equity and efficiency considerations. [2 p184–185] Similarly, reform of public expenditure is also influenced greatly by economic analysis. For example, fuel subsidies are argued to be fiscally unsustainable, inefficient, and inequitable [3]. But removal of these subsidies has led to what has been termed “IMF riots”. Similarly, consider the “gilets jaunes” protests in France:

“The diffuse interests represented in France’s National Assembly had agreed to increase gas taxes in 2014. But after farmers and their sympathizers closed down roads and took their fight to the cities, President Emmanuel Macron’s government backed down and rescinded the tax hike in 2018 [4].”

These examples, sharp and prominent as they are, only scratch the surface of the question of social acceptability of tax and transfer schemes which, on the face of it, pass the usual economic tests for efficiency and equity. Policy makers need to go beyond the calculations of conventionally economically rational actors, considering responses based on other perspectives and an altogether different calculus. In recent years the insights of behavioural economics have begun to illuminate alternative aspects of tax and transfer proposals. Similarly, cultural and historical factors are increasingly being recognized as constraints on the design of tax and expenditure frameworks.

This paper reviews the social acceptability of alternative tax and transfer schemes. Social acceptability tends to be much more culturally and contextually specific than conventional economic efficacy on efficiency and equity grounds. We draw on a range of literatures, including behavioural public economics; carbon tax plus compensation schemes; universalism versus targeting; as well as experiences in Latin America and elsewhere of implementing conditional cash transfers and other redistributive mechanisms. As we shall see, trust in government emerges as a key issue.

The structure of the paper is as follows: Section 2 begins with the standard “winners and losers” assessment of schemes based on narrow economic self-interest. Section 3 turns to the growing literature on behavioural economics applied to tax policy. Section 4 addresses the question of social acceptability in the context of the debate on universalism versus fine targeting of transfers for poverty reduction. Section 5 considers the broader question of ideas versus interests in animating major shifts in social protection. Section 6 concludes.

2. Economically Rational Private Interests and Politics

A conventional political economy framework for assessing policy reform is that of “the three I’s”—Interests, Ideas and Institutions. “Interests” comes closest to the conventional economic perspective, which is the impact of the proposed change on narrowly defined economic self-interest. “Ideas” encompasses a range of notions including ideology, information, and peer pressure. “Institutions” captures the rules of the game according to which the political contest on change is conducted—for example constitutional, procedural and other provisions. In this section we focus on rational private interests in the mediation of changes to tax and transfer schemes.

The usual way to think about the consequences of an alternative scheme of tax and transfer, or indeed any policy, is to delineate winners and losers from the change. For example, an export tax on food will hurt farmers and benefit consumers and the government Treasury. On the other hand, an export subsidy on food will hurt consumers, benefit farmers and will require fiscal injection. Clearly, those who lose will oppose the change. Examples of this, including those given in the introduction, abound. Whether they succeed in their opposition will depend on the institutional details of the political contestation, including, as in the case of the gilets jaunes, the role of street power and public disturbance.

An obvious answer to those who oppose a change on narrow rational economic self-interest grounds is to compensate them accordingly, so neutralising their opposition. The compensation makes sense from an efficiency perspective provided the aggregate gains would still exceed aggregate losses. If the losers are also at the bottom of the income distribution, such compensation is additionally beneficial from an egalitarian perspective. But within the political economy framework, the main reason for the compensation is to move the opponents into the supporters’ column.

A key issue then is to identify losers, to quantify the losses, and to devise compensation mechanisms. For those seeking to implement a carbon tax, this is an essential exercise. Such a tax would correct for the fact that the current generation have used carbon intensively, at the expense of future generations. But a carbon tax will also have distributional consequences in the present generation. There are many studies that provide such calculations. For example, Paoli and van der Ploeg use UK micro data to estimate the impact of carbon taxes along the income distribution [5]. They then simulate the effects of alternative compensation mechanisms through a recycling of carbon tax revenues—in other words, a balanced budget compensation framework. They conclude as follows:

“Since having a minority of large winners but a majority of small losers incites public antagonism towards a policy, it is crucial to study disaggregated measures of costs and benefits at the household level. Our main conclusion is that rebating carbon tax revenues through social security payments renders the policy progressive and benefits the highest share of households in our sample, at 35%.”

However, even with this compensation, the numbers show that “recycling does not ensure a majority of winners.” As such, the conclusion suggests that those looking towards compensation as a means of pushing through carbon taxes should be wary.

The difficulty of ensuring adequate compensation through existing instruments is not the only issue facing reformers. The question of credibility of compensation looms large, beyond the technocratic calculations of gains and losses and alternative compensation mechanisms. Recipients need to believe that compensation will actually be paid once the policy has been introduced. Consequently, trust in government promises turns out to be a key factor even in – and perhaps especially in – a system that emphasizes rational self-interest. Those considering how to overcome this would do well to look to the management of the energy subsidy removal in Iran in 2010. Here is how Devarajan [6] summarizes the problem and the outcome:

“In the mid-2000s, energy subsidies in Iran constituted an unsustainable 25 percent of GDP. But promises that subsidy removal would be accompanied by targeted cash transfers were met with skepticism….The government of Iran did something different. They handed out the cash transfers ahead of the subsidy reform. People could see the money in their bank account, but they could not spend it until the subsidies were removed. The reforms were implemented with hardly any protest.”

Experiences in Latin America, positive and negative, also highlight the importance of this trust. In most Latin American countries, tax reforms require new legislation, as opposed to compensation measures which can often be done by executive decrees. This is an important difference when trust is scarce. In 2003 a proposal was made by Mexico’s President Fox to remove exemptions to VAT, a permanent change requiring approval by Congress. The president pledged transfers to fully compensate households up to the fourth decile of the distribution (thus over-compensating those in the first three). Critically, however, these transfers would be set by presidential decree, requiring citizens to trust that future presidents would sustain them. In effect, it was as if the government was telling citizens: “We will change the law to increase your taxes but, don’t worry, trust us, we will compensate those in need.” Unsurprisingly, the asymmetry between a permanent increase in taxes written in a law, on one hand, and compensations of uncertain duration written in a presidential decree, on the other, led to the proposal’s defeat. Since then, Mexico has been unable to reform its VAT, and continues to experience the associated efficiency and revenue losses.

Chile provides an example in the opposite direction: a situation where trust led to a change that was very likely efficiency-enhancing and inequality reducing. During the presidential elections of 2013, then candidate Michelle Bachelet –who had been president before and was a trusted politician— ran on a campaign to increase taxes to fund education reform. In 2014, once elected, she carried out a tax reform of approximately 2% of GDP focused mostly on increasing income taxes. She then used the additional resources as promised to expand access to higher education; but with no compensations to any households [7]. This issue of trust and reciprocity between state and individuals as a determinant of the scope and nature of taxation has become central in the literature [8, 9, 10], and we will return to it throughout the paper.

The contrast between Chile and Mexico highlights two additional issues with the social acceptability of tax reforms aside from trust: the source of funds and their purpose. In the case of Chile, society was told that income taxes would be raised to widen access to higher education, a concrete issue which had been the subject of intense social mobilization in the two years prior to the presidential elections. In the case of Mexico, society was told that consumption taxes would be raised to improve social welfare, a vaguer issue where the benefits were unclear to many.

These experiences show that trust and societal valuations of the purpose of tax reforms are two factors that determine their social acceptability; factors not always captured in discussions of compensations. This is a valuable lesson that, mutatis mutandis, can apply to reform of energy subsidies, for example. Society needs to be told more than subsidies are inefficient or regressive and that the poor will be compensated; it also needs to be told that the resources obtained from removing these subsidies will be used for a worthier cause; and, of course, with credible commitment mechanisms.

3. Behavioural Economics and Tax Policy

Over the last three decades behavioural economics has questioned the standard model of individual behaviour, as embedded in the “rational actor” model of economic analysis. It is this model which underpins most standard analysis of the efficiency and equity of tax and expenditure schemes. Not surprisingly, an emergent “behavioural public economics” is providing new perspectives on a range of issues including conventional perspectives of taxation.

A number of overviews of the literature provide insights into the nature of responses to tax reforms [11, 12, 13, 14, 15]. In the most directly policy oriented of these reviews, Acheson and Lynch emphasize a number of implications, including the crucial importance of popular perception. In particular, there is significant misperception of the structure of taxes, and providing information can help correct this. For example, Chetty and Saez showed that giving personalized information on the direct impact of Earned Income Tax Credit (EITC) on income had a significant effect on hours worked [16].

Another example of the application of behavioural perspectives to tax policy is the recent analysis of the carbon taxation proposal in France by Douenne and Fabre [17]. They summarise their analysis as follows:

“Using a representative survey, we find that after the Yellow Vests movement, French people would largely reject a tax and dividend policy, i.e., a carbon tax whose revenues are redistributed uniformly to each adult. They overestimate their net monetary losses, wrongly think that the policy is regressive, and do not perceive it as environmentally effective. We show that changing people’s beliefs can substantially increase support. Although significant, the effects of our informational treatments on beliefs are small. Indeed, the respondents that oppose the tax tend to discard positive information about it, which is consistent with distrust, uncertainty, or motivated reasoning.”

The conclusions of Douenne and Fabre are consistent in spirit with the findings of the broader literature on behavioural economics and tax policy [17]. Namely, that beliefs are central, that they are difficult to change, and that responses to proposals for change are dependent on historical degree of trust in government. But the literature goes beyond these points to highlight a range of other factors, including the importance of salience, reference dependence, and bounded rationality.

The work of Douenne and Fabre is the most recent in a long line of research on public attitudes to carbon taxation [17], much of which has been reviewed recently by Carattini, Carvalho and Frankhauser [18]. Within this research, let us consider how public attitudes to carbon taxation depend on how the revenue is proposed to be used. The review draws three main conclusions from the empirical literature. First, formal earmarking of revenues to address environmental concerns, or provide redistribution directly, tends to increase public acceptability of carbon taxes. This relates back to the question of trust in government. Second, greater progressivity of redistribution of carbon tax revenues increases acceptability—although once more the issue of perception rears its head. Note that in Douenne and Fabre’s study [17], a uniform redistribution of revenues, which is progressive on standard economic definitions, was not considered to be progressive by the survey respondents. Third, offering compensation through cutting other taxes appears to be the least persuasive method of changing attitudes to carbon taxation. One explanation for this may again be the question of trust. Another explanation relates to the “mental accounting” discussion in behavioural economics. As Carattini, Carvalho and Frankhauser note: “…people do not make the link between a policy that is aimed at reducing greenhouse gas emissions and the desire to reduce taxes in a different area. They perceive these to be separate problems requiring separate solutions [18].”

Finally, the standard model of individual behaviour towards taxation does not allow for feelings or perceptions of reciprocity and solidarity. These are factors that can influence attitudes towards compliance and evasion, as exemplified by a case study from Argentina. In the city of Mendoza, home owners delinquent on their property taxes were sent tax bills breaking down the public services that would be financed with their taxes, but with particular emphasis on those provided to children. This measure resulted in an increase of up to 40% in the number of delinquent taxpayers that paid their bills, without any changes in enforcement [19]. This finding is remarkable, because Argentina scores low in terms of confidence in government and has experienced a long sequence of macroeconomic crises. Yet despite these adverse conditions, the appeal to reciprocity and social solidarity still received a positive response.

4. Social Constraints to Fine Targeting

A live and contentious issue, particularly in low-income countries, concerns the targeting of social protection and antipoverty transfers. Debates on the efficacy of universal basic income, or on the use of proxies to determine the magnitude of transfers, or on conditional cash transfers, all have a common set of conceptual and operational questions which have been addressed in the standard economic analysis framework, but on which consensus is difficult to find. At the same time, social constraints and public attitudes to alternative targeting and transfer schemes appear to be somewhat under-considered in the analysis.

The discourse on targeting is not new, of course. Besley and Kanbur reviewed the literature three decades ago and framed the debate in terms of the costs of targeting, accepting the clear benefits it offered in terms of the efficient use of fiscal resources for poverty reduction [20]. They identified three types of costs, under the headings of: (1) information; (2) incentives; and (3) political economy.

Information and administrative costs are clearest. In order to implement fine targeting such that transfers go to the poor alone (in other words, “inclusion and exclusion errors” are minimized) every potential recipient theoretically has to be evaluated for their income or consumption relative to the poverty line. This requires keeping, and updating, rolls of actual and potential recipients. The development of the literature and the practice of “indicator” (or “proxy”) targeting, relying on easily observable characteristics which are correlated with poverty status addresses this issue directly [21, 22].

The second issue with fine targeting identified by Besley and Kanbur is that of incentives [20], since by definition fine targeting requires high withdrawal rate of transfer as the poverty line is approached. In the extreme, perfect targeting requires an effective marginal tax rate of 100% on the poor. The incentive effects of this are calculated, for example, in Kanbur, Keen and Tuomala and it is shown that the optimal transfer schedule is some ways short of perfect targeting [23]. And in the current debates on the efficacy or otherwise of Universal Basic Income (UBI), the incentive effects of fine targeting do not appear to have been fully appreciated.

The third issue with fine targeting becomes clear when we take the poverty reduction budget as being not exogenously given but generated endogenously through political economy considerations. A more universalistic design ties together the interests of low and middle incomes, so that there is likely to be greater support for a higher transfer budget. A finely targeted scheme which makes a sharp delineation at the poverty line breaks that connection and isolates the poor, increasing the likelihood that the transfer budget would be lower. In the words of Gelbach and Pritchett, “more for the poor is less for the poor [24].” However, there is another social dimension which militates against this tendency. As noted by Fiszbein and Schady [25], such conditioning can serve to build up support from upper income groups who would otherwise be opposed to what they would consider to be “handouts.”

The above issues argue against fine targeting and toward more universalistic approaches. It has been argued that the availability of new digital technologies for information management can reduce the first of these [26]. Even if these issues were resolved, however, finer proxy targeting methods raise their own issues. These are more social in nature, and so perhaps not fully appreciated by economists. The basic point is that the sophisticated use of weighted proxies is difficult to explain on the ground to beneficiaries. It is especially difficult to explain to those left out of the particular scheme because they land just above whatever critical threshold is deemed appropriate by the technocrats, who have calculated that level as being the most efficient use of resources for poverty reduction.

One consequence of this difficulty is the creation of tensions and social discord which are not apparent to a technocratic mindset. Such consequences have however been identified in ground level qualitative assessments of transfer programs. Thus, Adato comments as follows on a transfer program in Nicaragua:

“The survey found that the program was well targeted, with under-coverage rates of 3 to 10 percent. The qualitative research found, however, that people saw themselves as ‘all poor’ and did not understand why households were selected into or out of the program, resulting in several types of stress and tension in the communities [27].”

And here is how one recipient perceived the outcome of targeting of the “Program Keluarga Harapan” transfer program in Indonesia:

“.. she was originally selected via a household survey where she was asked her name, house condition, how much land she owned and her employment. However, she said others got bigger allowances, some as much as 1 million [rupiah],” “because it was unfairly decided by the last kepala desa—you had to be connected to him.” “The last elections she has voted for a family member to ensure that she will benefit in the future.” [28 p29]

The kepala desa is the village head—the program which scores high on targeting clearly has unintended consequences.

Using narrow targeting in developing countries, where labor markets are segmented into formal and informal sectors, can also create difficulties. In some cases, cash transfers are distributed according to two conditions: being poor and not having a formal job. This double-targeting depresses formal employment precisely among those that need it most, and has been documented in the CCT programs of Argentina [29], Ecuador [30], and Uruguay [31]. In these cases, even though the targeting scheme may be socially acceptable, in part because informality is conflated with poverty, it nonetheless ends up locking poor workers in informal jobs; a phenomenon called at times an “informality trap.” This trap is conceptually distinct from more standard “poverty trap” as found in the analysis of Besley and Kanbur [20]. In the poverty trap the removal of transfers as incomes increase requires a very large marginal tax rate; in this case the subsidy is removed even if income does not change, but formality status does. That said, “poverty traps” can also be present when the cash transfer is high and sharply targeted on income levels, as evidence by Argentina’s CCT [32].

Overall, we have shown how there are at least four problems with fine targeting: (1) they can be perceived as unfair by at least some households; (2) they can lend themselves to political manipulation; (3) they can compress the available budget; and (4) they can distort labour market outcomes. These problems suggest the advantages of moving towards universal social protection schemes, particularly with regards to the provision of insurance. Such an insurance scheme would provide in-kind or in-cash transfers when health, death, disability or unemployment risks materialize, and in-cash transfers when individuals reach old-age. Aside from improving the functioning of the labour market and protecting all households from key risks, universality in social insurance could potentially be quite redistributive. Universality may not completely eliminate the need for some additional in-cash transfers among a subset of the population to support their current consumption. However, it would clearly diminish their relative importance, as substantial redistribution would already be occurring through insurance [33].

5. Ideas and Interests

We have discussed the social and political acceptability of various reforms of tax and transfer schemes. In most countries, the focus has been on a “scheme by scheme” approach, with incremental changes made to the economic system. However, sometimes a wholesale reordering of the entire system is called for. According to some, such large-scale reform would merely multiply the difficulties, rendering large-scale reform impossible. However, an alternative perspective is that an overall coherent vision would serve to convince the population and the polity that change is needed and possible. We illustrate this perspective with respect to reform of social protection in Latin America.

We begin by pointing out that there are three broad sets of social protection systems in Latin America. The first, focused on formally employed workers, consists of insurance against illness, disability and death risks, and old age pensions. It is financed from wage-based taxes paid by firms and workers, regulates dismissal standards, and sets minimum wages. The second set focuses on informally employed workers, and consists of health insurance and old age pensions, usually of lower quality or generosity than those in the first set and financed from general revenues. The third set is made up of targeted income transfers to the poor, often under the form of a CCT. Quite often, the last two sets are conflated because informality is confused with poverty, a situation that fails to recognize that in the region the majority of informally employed workers are not poor, although most poor workers—but not all—are informally employed.

In totality, these systems are not very effective. Insurance is erratic because workers change labour status during their life cycle, sometimes being protected by one set of programs and sometimes by another. Redistribution is weak and occasionally in the opposite direction of what is needed and/or desired. Further, the asymmetry in financing means that firms and workers are taxed when they contract formally but subsidized when they do so informally, disincentivising formal employment. Incentives are stacked against formality, a situation that aside from undermining the very objectives of social protection, is costly from the point of view of productivity, and erodes the tax base. Finally, there are untenable differences in terms of workers’ entitlements to social benefits as similar workers making similar efforts are not treated equally by the law [33].

In part, these problems result from the fact that these systems have evolved gradually in a “scheme by scheme” approach, designed in isolation, rather than part of a coherent total system. Consequently, there is little comprehension of how they interact with each other. Whatever the reason, these systems need to be more universal if they are to overcome their current shortcomings. However, a transition towards universality is impossible without increasing taxes and is thus inextricably linked with the social acceptability of taxation, which in Latin America is viewed with scepticism.

The reluctance to pay higher taxes in Latin America has many sources. Low quality of services, corruption, and weak accountability all translate into distrust of government. In some cases, the perception is that social spending keeps a segment of the population permanently living on public welfare. These are powerful forces arguing against more taxation, and interest groups that would be hurt by higher taxes often take advantage of them to sabotage tax reforms.

But these forces are not immutable. While some households that benefit from the status quo, particularly higher income ones, would likely oppose higher taxes under any scenario, middle- and low-income ones could be better off with a systemic reform of social protection, even if taxes on some of them were higher. This could happen if households understand that new taxes would not be channelled to fund more of the same, but to fund a universal system that provides similar and higher-quality benefits to all; that removes some of the obstacles that stand in the way of better-paid jobs; and that strengthens incentives to comply with the law, combats evasion and informality, and widens the tax base.

Ideas can change interests. At present, middle-income groups may oppose isolated reforms because they only perceive costs. They oppose eliminating generalized fuel subsidies because compensations are usually targeted at groups with lower incomes than theirs. They oppose tax increases because they perceive that they finance social protection programs that fail to address their needs. With these perceptions, reforms are very difficult, but an overall vision of the social protection system may change them. A proposal for a broad redesign might allow different groups to see what is in it for them and to understand that maybe they can be winners. Even though individual components of the proposal might be undesirable to some, the proposal as a whole may not. All in all, a broad proposal for systemic reform might then be more acceptable than isolated reforms. But for this to happen, members of society need to know where they are being led—what the endpoint looks like. They need to be presented with new ideas for them to consider that their interest might lie elsewhere.

Systemic reform of social protection does not imply that everything needs to be reformed at the same time. But it does imply that an overall vision of the desired social protection system is required, even if individual reforms are carried out gradually. In fact, one could argue that the absence of a vision of the desired social protection system is partly to blame for the current situation. Without a vision, social protection is little else than the inertia of long-dated policies and a continuous addition of new ones, sometimes in reaction to immediate needs, sometimes in compliance to the whims of the president in charge, sometimes in imitation of policies in other regions with very different realities, and sometimes in response to the latest fad in academic or international policy forums, with little coherence between them. In other words, the status quo. It is indispensable to have clarity about what needs to be done. Ideas are central.

The current architecture of social protection reflects ideas going back to Bismarck in the 19^th century. These ideas have had their day. There will always be interests opposing taxes, but new ideas can change their social acceptability. But for that to happen it is first necessary to change the nature of the political debate. No to more taxes for more of what has not really worked; yes, to more taxes for an option that can provide universal social protection and accelerate growth. No to isolated social protection reforms that deepen segmentations and perpetuate exclusions; yes, to a systemic reform to include everyone. Reforms based on new ideas may still not happen, but for them to have a chance of happening, the political debate needs to change.

6. Conclusion

All societies rest on taxation, and the distribution of its burdens and benefits among households is central to their welfare. In modern societies, however, taxes cannot be imposed arbitrarily by emperors, pharaohs or kings; a broad majority of citizens needs to agree with them and be willing to pay them.

In this paper we argued that the equity-efficiency framework usually used by economists to analyse taxation, based on the rational individual carefully calculating costs and benefits, is incomplete because it pays little attention to the social acceptability of taxation. This is not to deny that individuals pursue their own self-interest, but it is to say that their attitudes towards taxation are more complex than that. They also depend on factors like trust in government, perceptions of fairness, feelings of solidarity and reciprocity and, critically, their own ideas. These factors need to be factored in when policy proposals are made to reform specific taxes or subsidies, and even more so when they involve reforming various taxes or subsidies at the same time.

We documented situations where failure to consider the social acceptability of taxation led to the rejection of tax changes by individuals that, on the standard calculus, should support them. These failures can be very costly, and some countries are foregoing the benefits of tax and subsidy reforms that could increase societal welfare: they may not be taxing carbon when they should, as in France; or may be unable to remove inefficient VAT exemptions, as in Mexico. We also documented situations where building trust, or strengthening feelings of reciprocity and solidarity, led to successful reforms, like the removal of regressive energy subsidies in Iran, or increases in property taxation in Argentina.

Understanding the social acceptability of different tax and subsidy schemes is thus of the essence. The key point is to recognize that in addition to self-interest, individual behaviour is also determined by culture and ideas. The interaction between these factors is complex and, moreover, depends on their specific context. For example, in a given country, the same tax-cum-subsidy reform may be accepted or rejected depending on the government that is proposing it. It is also complex because the ideas or perceptions that individuals have about reforming taxes and subsidies sometimes depend on whether they are put forth on a “scheme by scheme” basis or in a more systemic context; an issue that is of great relevance for the reform of inefficient social protection systems like the ones observed in Latin America and other regions of the world.

The challenge ahead is thus to develop a richer framework for the study of tax and expenditure policies. This framework needs to give central consideration to issues of efficiency and equity as generally understood by economists, but also recognise that these policies do not operate in a vacuum; they operate in countries with specific histories, cultures and political contexts. The framework will be more complex, no doubt; but also much more useful.