Taxation and Development: Again

Contributor Notes

Author’s E-Mail Address:

Issues of taxation and development, which have long been a central concern of the IMF, have attracted wider and renewed interest in the last few years. This paper reflects on three broad lessons of experience: that developing countries differ vastly in tax matters, and in ways that are less than fully understood; that the history of ‘big ideas’ in guiding tax reform for developing countries is decidedly mixed; and that the value of the emphasis often placed in this context on ‘informality’ is decidedly limited. It also asks whether ideas of ‘state building’ emphasized in some of the recent literature are likely to lead to practical advice much different from that commonly offered now.


Issues of taxation and development, which have long been a central concern of the IMF, have attracted wider and renewed interest in the last few years. This paper reflects on three broad lessons of experience: that developing countries differ vastly in tax matters, and in ways that are less than fully understood; that the history of ‘big ideas’ in guiding tax reform for developing countries is decidedly mixed; and that the value of the emphasis often placed in this context on ‘informality’ is decidedly limited. It also asks whether ideas of ‘state building’ emphasized in some of the recent literature are likely to lead to practical advice much different from that commonly offered now.

I. Introduction

Interest in issues of taxation and development comes and goes. This is true of policymakers—in developing countries themselves and, especially, in donor countries—and among civil society and academics (with, it has to be said, a historically low level among the last of these). Now we are entering an up-phase of interest, not least from the donor community. At their November 2010 summit, for instance, the G-20 leaders stressed the importance of strengthening revenue mobilization in developing countries and asked involved organizations to report on how best they could help.1 The explanation is perhaps not hard to find. Many developing countries need substantial additional revenue to finance poverty-reduction—an additional 4 percent of GDP2 or so is needed in many low-income countries if they are to have a good chance of meeting the U.N. Millennium Development Goals—as well as pressing needs for infrastructure and adaptation to climate change.3 At the same time, the dire post-crisis fiscal position of many advanced economies is naturally focusing attention on the extent and effectiveness of the aid they provide to developing countries and on ensuring that it supports rather than discourages the latter’s own revenue-raising efforts. Hence, the renewed focus on, in the jargon, “domestic resource mobilization.”

This welcome resurgence of interest makes it timely to take stock of experience and lessons in the area, and to assess newer challenges to resource mobilization in developing countries, such as those from globalization. But that is not the purpose here; this chapter does not aim to provide a sweeping review of technical issues, largely because there is no shortage of surveys and there are quite a few books on resource mobilization and development.4 Nor is the aim to identify important areas for research or methodological improvement, though there will be some of that. (This volume itself is evidence of the importance and value of strengthening empirical work in the area). Instead the purpose of this essay is to reflect, no doubt idiosyncratically, on some wider issues in the practical advice that developing countries are commonly given on tax matters.

The theme, perhaps, is to caution against the over-simplification (at best) to which this area has been, and remains, prone.

Four topics are selected for discussion (perhaps more accurately, a bit of a rant), each with a view to informing the renewed focus on resource mobilization issues—or at least avoiding past mistakes.

II. Developing Countries Differ—Yes, and…?

The literature in this area is rich in papers and policy documents with some variant of “Taxation and Development” in the title. There is no harm in that, of course, and many of the more recent contributions—not least, those in this volume—go beyond generic information and advice to provide detailed case studies of the effects of taxation in various developing countries. But the recurrent title does reflect a search for generalization that, after decades of work in the area, one might have hoped to move beyond. By comparison, public finance specialists rarely set out to provide similarly generic treatments of taxation in advanced economies.

The point is not simply that developing countries differ greatly from each other. No one would say otherwise: even separating out the newer category of emerging market countries (itself ill-defined, and overlapping with those still regarded as “developing”) and the transition economies (yes, there still are some), considerable differences in physical characteristics, political structure and institutional history remain. Development economists have become increasingly sensitive to these differences, with a lively and sometimes heated debate on the importance and relative roles of institutions—shaped in large part by colonial histories and legal traditions—and geography, especially climatic conditions.5 The question for present purposes is whether such differences really matter for thinking about taxation.

Some aspects of geography clearly matter a good deal. Probably the single most important tax-relevant difference across developing countries—indeed, perhaps across all countries—is in natural resource wealth.6 This is far from entirely exogenous, of course, in that the level of exploration for resources is an outcome of economic decision-making,7 but also far less than fully controllable. Between the early 1980s and 2005, resource-rich countries in sub-Saharan Africa increased their tax–GDP ratios by about 7 percentage points; non-resource related tax revenue in the region, on the other hand, was essentially stagnant.8 For such countries, the key tax design issues are how to secure an acceptable share of the resource rents in a way that ensures proper exploitation of those resources.9 In the extreme—the Timor-Lestes of the world—the question is in effect whether to have a domestic tax system at all. Even in more moderate cases, the role of the non-resource tax system can be quite different from that in resource-poor countries: largely a means of diversifying the revenue base and, perhaps, of increasing government accountability, rather than primarily a matter of raising revenue. Tax design for such countries becomes in large part one aspect of the wider issue of resource management—involving issues of transparency, macroeconomic management, and savings decisions—that have wider implications for the political and economic future of the country. Too long seen as an area for dull specialism, issues of resource taxation in low-income countries, including the treatment of such exhaustible resources as fisheries and forests, have been left for far too long to sectoral specialists.

Other aspects of geography also matter for tax design. Smaller countries, especially rugged, distant islands, can impose taxes at the border much more easily than can large landlocked ones, so it is not surprising, for instance, that smaller countries tend to have more efficient VATs10 and indeed seem less inclined to adopt a VAT, given the relative ease of raising substantial amounts by customs duties.11 Even shape may matter: The Gambia’s long, thin structure has contributed to extensive re-export activities and hence—with tax not being remitted on exports, some of which are presumably controlled at the border—to unusually strong revenue from a single-point sales tax.

Nor is there any doubt that politics, in the broad sense, are important. This is not simply a matter, as we tend to think of it in advanced economies, of building some minimal consensus for tax changes. Political instability, for instance, lowers incentives for the incumbent government to invest in developing administrative capacity; Acemoglu (2005) and Besley and Persson (2009) explore these incentives more generally (as discussed further below) and, as a fascinating example, Aizenman and Jinjarak (2008) show that political instability is associated with reduced effectiveness of the VAT.12 Still more obviously, some national governments (perhaps increasingly many) simply do not have full control over all their territory. In this case, the standard prescription for the revenue-desperate of relying on customs revenue and aiming to move to a value added tax—most of the revenue from which is collected at the borders—is of limited use. More generally, tax policy is in such cases likely to be constrained by the need to avoid worsening internal conflict; levels of mistrust and frank dislike, for instance, can make reallocating tax powers across levels of government effectively impossible even when it is demonstrable (so far, at least, as these things ever can be) that all could gain by doing so. In this and many other ways, tax reform can get very personal. Not least, its momentum and effectiveness can depend on the presence of one or two “champions”—the risk then being of backsliding once they have departed the scene.

Less immediately evident, and under-studied, is the question of whether colonial histories continue to make a difference to the nature of, and possibilities for, domestic revenue mobilization. Certainly there are particular instances where this seems to be the case. In both India and Pakistan, for instance, a key obstacle to arriving at coherent VATs has been a constitutional restriction originating in the 1935 Government of India Act, which allocates the powers to tax goods and services uniquely to distinct levels of government13—a distinction running counter to the appeal and logic of the VAT, which must apply on an integrated basis to both. The same act also allocated the taxation of agricultural income to the states/provinces, another source of continuing difficulty. There are also clear differences in tax design between Francophone and Anglophone countries in sub-Saharan Africa. The former, for instance, have traditionally used a “complementary” income tax (a progressive tax levied on the sum of net incomes after application of a series of schedular taxes), seem to make more use of VAT-withholding and advanced collection schemes,14 and have been inclined to use a territorial approach to the taxation of foreign income. These features (with the possible, and possibly telling, exception of withholding, which is discussed later) are all echoes of tax practice in France—or rather of the France of many years ago. Similarly, Lusophone African countries have tended to follow peculiarities of tax rules inherited from Portugal.

The distinction between the Francophone and Anglophone experiences is potentially a particularly interesting one. In the wider development debate, it has been argued that the latter (or, somewhat different, those with the British common law tradition) have fared significantly better in the post-colonial period. A natural question is whether that is true in the area of taxation. The results in Table 1 suggest that it might. The first column reports the results of a standard, very parsimonious “tax effort” panel regression, relating the ratio of tax revenue (other than from natural resources) to GDP (NONRES) in 39 sub-Saharan African countries to the key usual suspects in such equations—(the natural log of) GDP per capita (lnYPC) and openness (OPEN)—and to a dummy (in level and interacted) indicating the presence or absence of a VAT (VAT). The broad results are in line with previous work, and need not detain us further.15 What is of interest, however, is the consequence of adding (also both in level and interacted with the VAT dummy) in Column 2 a dummy, ANGLO, taking the value unity for Anglophone countries and zero otherwise. There emerges a strongly significant difference between revenue performance in the two countries: non-resource revenue is higher, all else equal, in Anglophone countries, and the VAT performed better.

Table 1.

Tax Performance in Sub-Saharan Africa

article image
Notes: Dependent variable is non-resource-related tax revenue (so far as it can be identified) relative to GDP). The dataset is described in Keen and Mansour (2010a); 17 of the 39 countries are classified as Anglophone. The dependent variable is revenue other than from natural resources. Estimation is by ordinary least squares, with country dummies. Robust standard errors are in parentheses; asterisks indicate significance at the 1 percent (***) and 5 percent (**) levels.

This result is more of a question than an answer. The challenge is to identify precisely what it is picking up, potentially including: immutable differences of geography; detailed aspects of tax design that are in principle easily changed; deep differences in legal and other aspects of economic arrangements, that translate (how?) into tax performance; or something else. There is surprisingly little work on the economics of comparative tax systems16 in developing countries to help us resolve these issues (in part, no doubt, because, with very rare exceptions, those working in the area tend to know well only countries in one or two language groups). The issues range from the very broad to the quite specific. VAT withholding is an example of the latter. Does it have a substantial revenue impact (not just directly, but through its impact on the wider VAT system)? And since it is not a tradition in France itself, why is it far more common in Francophone countries? Perhaps there is a post-colonial transmission mechanism in fiscal matters not from the colonizing country but between those with a common colonial heritage; in this case, regional integration, often grouping countries with similar colonial pasts, will preserve or amplify these differences across the wider set of developing countries. Better understanding of such issues will not resolve all the problems of framing good tax advice. But it can surely help—not least, perhaps, in coming to terms with the lesson of experience that change can be much harder to bring about than many have often liked to believe.

That geographic and institutional differences matter for tax design does not mean, it is worth stressing, that no general principles can apply. It is sometimes said, for instance, that the tax advice given by the IMF is “one size fits all.”17 But in some respects one size does fit all: we generally think that the decimal system works pretty well, for instance, as does dividing the day into 24 hours. The examples are flippant, but—without claiming for them the same universality—prompt the thought that some considerations in tax design also apply rather generally. These might include: the importance of effective withholding, on wages and at least some forms of capital income, for effective income taxation; the recognition that differences in statutory tax rates (whether across individuals, commodities, countries) invite evasion and avoidance; that import tariffs can protect inefficient producers; and the fact that turnover taxes create distortions that a well-functioning value added tax can avoid. We do have powerful guiding principles, perhaps most notably the Diamond-Mirrlees theorem on the desirability of production efficiency, which provide guidelines for tax design. We could do with better, more practicable ones, especially in the developing country context. But we should not pretend that we lack guiding principles, or that broad commonalities of tax design and advice are necessarily inappropriate.

III. The Checkered History of Big Ideas

Fashion and fads are a part of life in many areas of economic policy advice. But advice on revenue mobilization and development over the last sixty years or so has been especially strongly marked by the focus on a succession of big ideas, each seen in turn as potentially critical to a sustained improvement of performance. In practice, however, their records have been mixed:18

Direct Taxation

Starting with the Shoup mission to Japan in 1949—the birth (at least for the modern era)19 of the industry offering expert foreign tax advice to developing countries—the early focus was on the development of some more or less comprehensive income tax as the centerpiece of a modern tax system.20 Celebrated reports by Nicholas Kaldor for India and Ceylon retained the focus on progressive taxation at individual level, but aimed instead at progressive taxation of consumption rather than income.

There can be little doubt that progressive personal taxation has not made anything like the contribution to resource mobilization in lower income countries that had been hoped (as has also been true, incidentally, in Japan, where its yield remains low, both absolutely and as a share of all tax revenue). The personal income tax commonly accounts for less than 10 percent of all tax revenue in low-income countries—compared to an average of more than 25 percent in OECD countries—and is widely recognized as essentially a tax on the labor income of those working in the public sector or large private enterprises. This point is nicely put by Zolt and Bird (2005, p. 1694): “…in most developing countries, the global progressive personal income tax long advocated by experts is in fact neither global or progressive, nor personal, not often even on income.”

The reasons for this limited success (at best) of the personal income are less well understood than they should be, but no doubt reflect both political and technical failures. Entrenched power structures and corruption are powerful obstacles to taxing elites and many high income/wealth individuals—perhaps more so than in the early days of income taxation in current high-income countries, given greater opportunities (from resource wealth, for instance) for rent-seeking and the concealment of income by placing it offshore. But it also reflects administrative weaknesses and errors of design, importing inappropriately from the experience of developed countries. Thresholds may have been set far too low, for instance, forgetting that many developed countries initially began with a very narrow set of taxpayers; and the comprehensive income tax has proved overly ambitious even in many developed countries. More fundamentally, effective progressive personal taxation of non-wage income requires some form of self-assessment (perhaps bolstered by withholding and third-party reporting) that continues to elude many tax authorities, which remain wedded to more direct methods of assessment. Whatever the reason, however, the fact is that direct personal taxation has not proved a route to either a sustained or a particularly fair enhancement of tax capacity in lower income countries. Current advice tends to be much more modest, building on what are currently in effect schedular tax systems toward something approaching a dual income tax, with a progressive tax on labor income and a flat (or at least simple, more uniform and fairly low) tax on capital income; indeed the dual income tax is arguably better suited for many developing economies than for more advanced ones, in that its traditional Achilles heel—the difficulty of disentangling the labor and capital income of the self-employed and close companies—is less of an issue given the much larger difficulties in levying any kind of sensible tax on them at all.

The Value Added Tax (VAT)

Sometime around the late 1970s, practitioners—notably in the Fiscal Affairs Department of the IMF—made the great intellectual leap to a belief that the VAT, then firmly established in Europe and a handful of other countries, could, if kept sufficiently simple, be an effective and (by dint of judicious exclusions and a fairly high threshold) reasonably fair source of revenue even in countries with limited administrative capacity. Adding to this direct revenue appeal was the thought that by catalyzing changes in ways of doing tax business, the VAT—which can only work effectively if implemented by self-assessment—could pave the way to the elusive strengthening of income taxation. Built on such reasoning, adoption of the VAT became the centerpiece of tax reform in the developing world over the past thirty years: by around 80 percent of countries in sub-Saharan Africa, for instance. The VAT has become the norm, typically raising about one-quarter of all tax revenue.

Has this particular fashion lived up to the claims made for it?21 Clearly the VAT remains controversial, but the only systematic evidence that we have suggests that in broad efficiency terms it has: all else equal, countries with a VAT raise more revenue—a sign of having a more efficient tax system—than do those without.22 The gains are not always large, however, and are less apparent in sub-Saharan Africa than elsewhere. In equity terms, although the VAT is often thought of as a regressive tax, most studies in fact show it to be more or less distributionally neutral,23 and “the evidence is…that the VAT is likely on the whole to be less regressive than the trade and excise taxes it has replaced. Furthermore, in at least some developing countries, the VAT may be about as progressive as the income tax” (Zolt and Bird, 2005, p.1639). What ultimately matters, of course, is the distributional impact of the full tax-spending system. This has been little studied in developing countries, an important exception being the work of Muñoz and Cho (2004) on Ethiopia, which concludes that the net impact of a uniform rate VAT with proceeds allocated to education and (especially) health can have a strongly progressive impact.

Quite why the VAT nevertheless has such a bad name with many observers of tax policy in low-income countries thus remains a bit of a mystery. No doubt it has sometimes been oversold, and those who have been doing the selling may be too defensive. Clearly too the VAT causes distortions that should be of concern;24 but so does any tax. On its record, as we understand it so far, it is hard to see the VAT as a failure. The issue, surely, is not whether to remove it (replacing it by what?), but how to improve it.

Revenue authorities

As a way of addressing the corruption and political interference seen as fundamental obstacles to effective and fair taxation in many low income countries, considerable effort has been put into the creation of quasi-independent revenue authorities. Spurred in part by encouraging experiences in Latin America, starting in Peru in 1991—where non-compliance fell dramatically over the next fifteen years or so—the thought was to reform tax administration by placing it in the hands of well-paid officials, working in a well-financed institution protected against pressure from the highly-placed. Revenue authorities possessing such features—to quite varying degrees—are now in place in more than forty countries, including nearly all of Anglophone Africa.

While there are no formal evaluations, there seems to be a consensus that experience with revenue authorities has been mixed—or at least that there is no firm evidence to conclude otherwise.25 There was perhaps never much prospect of discouraging corruption by salary increases that remain dwarfed by the proceeds of collusion or extortion; turning a blind eye to a container of cigarettes, for instance, might save tax of around $500,000. Still more fundamentally, perhaps, there are often few obstacles to carrying out many of the key reforms associated with revenue authorities within existing organizational structures and public service rules. And, although easy on paper, in practice the formation of revenue authorities can be a painful and even paralyzing process (involving the exhausting and occasionally litigious process, for instance, of having employees reapply for their own positions). While there are more encouraging signs in the last few years—the Uganda Revenue Authority is now often held up as a model—it remains the case that revenue authorities have not always lived up to the high expectations held by some.

Large Taxpayer Offices (LTOs)

Securing the remittance of tax by the largest enterprises is critical everywhere, but likely even more so in lower income countries, given both the weakness of self-assessment by individuals and, perhaps, a missing middle in the distribution of firm size:26 the largest 1 percent of companies commonly account for 75 percent or more of all tax payments.27 To ensure proper control of such enterprises (which will generally include, for instance, banks, large foreign investors, and resource companies) the creation of dedicated LTOs within revenue administrations has been another focus over the last fifteen years or so. No systematic count is kept, but most Anglophone Africa countries, for instance, now have an LTO.

While there again are no formal evaluations,28 the stress on the LTO has been at least relatively uncontroversial. Any sensible allocation of administrative resources, when revenue collections are weighed against administrative and compliance costs, would seem to call for a particular focus on the largest payers: the gain from having the banks or main telecom and resource companies not delay remitting tax by a few days, for example, can dwarf that from expanding the net of small taxpayers. If there is one recommendation in tax matters for which one size pretty much does fit all, it would perhaps be the creation of some form of LTO.

This is not to say, however, that the focus on LTOs creates no difficulties. There can be an incentive, for instance, for enterprises to find some way to remain outside the LTO, by avoidance or evasion or by genuinely limiting the scale of their activities. Moreover, governments can have an incentive to adopt anti-competitive and other measures that shift tax base (not just profit, but also payroll and domestic sales) into the more easily-taxed enterprises.29 And even in the absence of these effects, focusing on any subset of taxpayers risks creating production inefficiencies—an important reminder that the size distribution of firms should not be taken as exogenous. In particular, in so far as there is a missing middle of firms in lower income countries, this may reflect the inverse U-shaped pattern of effective rates of both taxation (large firms get exemptions, small ones evade)30 and—in terms of profitability—bribes (more profitable firms may be less vulnerable to extortion in the form of having their tax liability credibly over-stated, less profitable firms have less to pay).31 Little is known about the practical significance of these concerns. But they stress that while controlling large taxpayers is a prerequisite for effective tax administration it is, at best, a first step toward that goal.


One conclusion to draw from all this is surely the need for more effective evaluation of initiatives in this area. The lack of careful studies partly reflects weaknesses of data—even on basic revenue information—and, perhaps, a lack of interest and/or awareness in the academic community. It also reflects an absence of controlled experiments, partly because they are intrinsically difficult to design for many major reforms—it is hard to introduce a value added tax in only part of a country, for instance, and impossible to do it only for a few sectors—but also because evaluation has not been given great weight relative to the potential benefits (and perhaps limited downside) of acting quickly: donors naturally want to get things visibly done. Evaluation is now being given more importance by donors, but as yet perhaps without full recognition that doing this to the same standards increasingly expected in other areas of public policy will require building it more purposively into project design. There are now instructive examples of feasible and useful experimentation in tax issues for developing (or at least emerging) economies: those of Pomeranz (2011) in collaboration with the Chilean tax authority, for instance, cast new light on how enhanced audit at one point in a VAT chain affects compliance elsewhere in the chain. More immediately encouraging is the increased use of microdata to address issues in developing countries—such as Kleven and Waseem (2011) on the impact of discontinuities in Pakistan’s income tax, Goyette (2012) on the structure of VAT audit probabilities in Uganda, and, not least, several examples in this volume.

A second lesson is that there are no quick fixes. But while the slow progress of tax reform in many countries over the last decades has been a disappointment, the wider considerations discussed above suggest that this should not, in retrospect, come as too great a surprise. One of the risks in focusing on big new ideas is that it can detract from the dull but critical job of moving beyond the moment of innovation to the hard work of implementation. The task of moving toward an effective VAT is not done, for example, simply by introducing it and surviving the political fallout. Years of further work can be needed to put in place the registration, audit, and other administrative capacities needed for its full potential to be realized—a task that in many countries still remains very far from complete. It can be easier to move on to more dramatic initiatives. More generally, focusing on grand innovations risks distracting from the less dramatic but important areas in which progress has been made; for example, the role of plain vanilla excise taxes has arguably received too little attention in many countries.

All this suggests that, important as it is to come up with new big ideas, some caution in their application—and the maintenance of some balance with less dramatic efforts at improvement—is appropriate. Three such ideas are currently at the fore.32 One is the development of offices devoted to the needs of medium-sized taxpayers, with small taxpayer offices following on; this builds on the perceived success of the LTO. Another is the notion that large sums can be recovered by addressing evasion by the rich through tax havens and by stemming profit-shifting by abusive transfer pricing, financial structuring, and other devices; here the issue is less the precise magnitudes of the sums involved, which remain uncertain but are no doubt substantial,33 than the extent to which lower income countries can effectively address problems that even the most sophisticated administrations struggle with. The third is an emphasis on the links between taxation and the building of accountable, transparent, and efficient governments; this is discussed in Section V below.

IV. Informality is Not the Issue

Discussions of tax challenges in developing countries conventionally emphasize early on a massive extent of informality, often around half or more of measured GDP. As Zolt and Bird (2005) stress, the impact on measured GDP itself is likely less than this, in that estimates of informality are typically of gross activity rather than of value added. But the more fundamental limitation of this focus on “informality” is conceptual: What exactly does it mean?

The wider development literature does not provide much help. Kanbur (2009, p. 2) concludes that the term “…has the dubious distinction of combining maximum policy importance and political salience with minimal conceptual clarity and coherence in the analytical literature…” and that “…the literature as a whole is in a mess” (p. 2). Much the same is surely true of the use of the term in discussions of taxation and development. It simply fails to evoke with much accuracy the key issues of compliance that are the real challenge. For instance

  • Many micro and small enterprises—street traders and something more—are naturally thought of as “informal” in some vague sense of operating essentially untouched by government restrictions. But in tax terms, it is far from clear that they pose any particular problem: a balancing of traditional concerns, weighing the revenue foregone by excluding them from tax against the administrative and compliance costs (and the gain in production efficiency) from including them, would almost certainly imply that they should not be taxed (except indirectly through taxes, such as the VAT on their purchases, that are remitted by others—and perhaps by small fees that, many would quietly say, are there for show rather than as an attempt to raise serious revenue). That is, the optimal tax to be remitted by such operators themselves may very well be zero.

  • Many of the most serious instances of under-payment of tax, on the other hand—in terms of both revenue loss and damage to the fairness of the overall tax system—are by professionals: doctors, lawyers, architects, and the like.34 Highly qualified, and subject to a range of professional restrictions, these are not the type of business operators one would naturally call “informal.”

The real issue then is not in any very useful sense “informality.” The term “hard-to-tax”35 is much closer to the mark, though perhaps carrying an excessive air of inevitability and, in so far as it has come to be taken as largely synonymous with dealing with small businesses, abstracting even from the differences between types of small enterprise—micro versus sole professionals, for instance—and consequently the key question of whether there should be any serious attempt to tax them at all. The point, in any case, is not to try to change usage—no doubt the term “informality” is here to stay—but rather to stress that the real issue is non-compliance and its endogeneity to both tax design and implementation. What is needed is to go beyond the comfort of broad labels and probe deeper into the anatomy of non-compliance, and how tax design and implementation should reflect and address it. These are complex issues, but facing them head on at least points to potentially fruitful areas of inquiry and action.

It makes a good deal of difference, for instance, whether non-compliance is in the form of “ghosts” (operators who should register for tax purposes but simply do not do so, and so remain unknown to the authorities) or “icebergs” (registered but under-paying taxpayers). In the former case, identification and registration of taxpayers is a critical first step; in the latter, audit and enforcement are key. Such information as we have suggests that while ghosts are by no means unknown in advanced countries (non-filers may be 7 percent of all potential taxpayers in the U.S.),36 they are much more prevalent in the developing world: both Gauthier and Gersovitz (1997) for Cameroon and Gauthier and Reinnika (2001) for Uganda find that around 50 percent of firms were failing to pay anything at all for at least one tax for which they were legally liable. The term “ghosts,” however, may suggest an invisibility that is far from reality: a substantial number of the evaders identified in Cameroon had had contact with the tax administration in the previous year. Finding potential taxpayers, it seems, is only part of what has to be done.

Focusing on a non-compliance agenda also helps put efforts to deal with the “hard-to-tax” in a wider context. Simply from the perspective of efficiently allocating administrative resources, for instance, does it make sense to devote more effort to taxing smaller businesses rather than, for instance, improving refunding for exporters under the VAT or tackling aggressive tax planning by large multinationals? Is there even much of a real trade-off, given the very different skill requirements? In many countries, the large number of low-skilled employees working on small business taxation is perhaps best interpreted as a workfare scheme—which may make sense, but casts quite a different light on the exercise.

Importantly, strengthening compliance is not a matter only of tax administration. In a telling and justly famous remark, Casanegra de Jantscher (1990) claimed that “in developing countries, tax administration is tax policy.” But the opposite, of course, is equally true: “Questionable options in…tax policy sometimes lead to equally questionable administrative practices.” (Bodin and Koukpaizan, 2008). Exemptions, for instance, not only pose control problems but create opportunities for corruption. And setting too low a VAT threshold can face tax administrations with essentially insuperable implementation difficulties. Indeed it is in the proper partitioning of enterprises for different forms of taxation reflecting their book-keeping ability in a way that is both practicable and avoids significant distortions37 that the need to combine policy and administrative considerations is most inescapably evident—and that remains one of the most challenging issues in the whole field.

Another key area in which tax design can affect compliance is through the use of withholding taxes and advanced collection schemes of various kinds; not just to collect, indirectly, at least some tax from those who may not be fully compliant with their own legal responsibilities but also—to the extent amounts withheld are creditable against taxes they are failing to remit—tilt the balance of their calculation toward becoming more fully compliant. This is part of the logic of the value added tax: a retailer who chooses not to register, for instance, escapes all tax38 under a retail sales tax, but only the tax on their own value added, so long as tax is charged by their suppliers, under a VAT. Much the same argument applies to other forms of withholding: on imports, for example, as additional charges (by the buyer or seller, sometimes sector-specific) under the VAT, or even on some consumption items (such as mobile phones in Pakistan). The argument is far from watertight, of course: the same considerations can imply pressure on the withholder to be less than fully compliant, with the possibility of ‘bad’ chains forming under the VAT for this reason stressed by de Paula and Scheinkman (2006). Securing the first step in the withholding chain is thus likely to be critical—and so too is the effectiveness with which crediting is implemented.

Views on withholding and advanced collection schemes differ widely. Some see them as a useful tool to encourage compliance, others as a pernicious distraction from the core task of improving administration. Importantly, they are to a large degree “home-grown” innovations, and often run counter to external advice—suggesting that they require more careful thought and analysis than they have often received. Withholding is a front-line tool in addressing not some imprecise notion of informality but very specific issues of non-compliance.

A refocus from the defeatism of ill-defined impressions of informality also leads to simple but important insights in relation to agriculture, which still accounts for 20 percent or more of GDP in lower income countries.39 Agriculture features prominently in any listing of the “hard-to-tax.” Historically, however, it has in fact proved surprisingly easy to tax. The Mughal Empire, for instance, may have collected about one-sixth of national income through the land tax; more recently, the agricultural sector accounted for about 40 percent of total tax revenue in Argentina, but only 15 percent of GDP (Skinner, 1991). In much of the post-colonial era, indeed, the concern has been that the sector was over-taxed, through a combination of export taxes, controlled prices, and over-valued exchange rates.40 Export taxes, however, are now much less widespread, and liberalization has largely eroded implicit taxes on the sector. (The most important exception is probably the taxation implied by agricultural subsidies in the advanced economies). Focus thus shifts to other measures of explicit taxation, addressing the conclusion of Skinner (1991, p.143) that “the problem with the land tax”—and the same is true of agricultural taxation more generally—“is not so much its ability to collect revenue [as] its ability to do so fairly.”

It is important to remember that most small farmers will fall into the category of those who on standard criteria would not be brought into taxation at all—whether for personal income tax or the VAT. But land holdings remain highly concentrated in many developing countries, being much more unevenly distributed, for instance, than income itself.41 Unfortunate though that is in many respects—the evidence being that greater land inequality is associated with slower growth—it is a reminder that there are, in most countries, large farmers and agricultural enterprises, which should be just as capable of complying with tax requirements as other operators of similar scale. Plantation agriculture, as Rajaram (2004) points out, is pretty easy to tax. If it has proved hard to tax agriculture in a way that is fair and productive of significant revenue, this is often not because the sector is intrinsically hard-to-tax but because of its political clout. The failure of the agricultural income tax in Pakistan, for instance, surely has more to do with the power of elites than with intrinsic technical difficulty. The politics of agricultural taxation can be complex,42 and there are no generic, easy solutions: whether allocating taxing rights to lower level jurisdictions gains more in improved local information than it may lose through increased corruptibility, for instance, will be context-dependent. It is important too to understand what types of simple presumptive taxes43 could play a potentially constructive role in the agricultural sector as they do for small enterprises in general. But the first order of business, as for large enterprises in general, is to ensure that the core tax base is properly taxed.

V. State Building and Taxation—What’s New?

One currently popular big idea—mentioned earlier, in several of the official documents cited at the outset,44 and notably stressed by the OECD Development Assistance Committee (OECD, 2008)—is that fuller account needs to be taken of the importance of the links between taxation and state building.

Here there are two strands of recent literature, both taking a broad view of revenue mobilization within the historical sweep of state development. One strand stresses that the capacity to collect tax revenue reflects prior investment decisions, and explores the way in which these are shaped by such considerations as political stability, the extent of common interests—warfare being a leading instance—and the degree of political consensus (Acemoglu, 2005; Besley and Persson, 2009 and 2010). More evident in recent policy documents, however, has been the second strand of literature: the “new fiscal sociology.” This has many similarities to the first, but differs not only in methodology but also in focusing more directly on the idea that taxation is critical to developing good governance more widely, in the sense of building state institutions that are responsive, accountable and competent.45 Expressions of this idea vary, but a common theme is that taxation fosters state building both by providing a focal point for bargaining between the state and citizenry and through the development of high quality institutions for tax collection (Bräutigam, 2008). The argument draws on a range of historical examples, from ancient China, through (somewhat ad nauseam) the taming of the Stuarts and the emergence of a well-financed powerful state in Britain over the latter seventeenth century, to the experience of tax revolts in post-colonial Africa. These provide often compelling (and always fascinating) illustrations of the general and surely non-controversial point. But practical people working in the business of tax design will want to know: “What difference does this perspective make to what we should be doing or advising?”

Views differ even among advocates of the state-building perspective. Referring to the orthodoxy of the IMF and others, OECD (2008, p. 22) concludes that “…this agenda broadly serves state building as well as economic policy objectives.” To Bräutigam (2008, p.33), however, “[a] reform agenda focused on issues of state building in the poorer countries would look substantially different.”

Some of the factors that we know have had a critical role in developing and shaping the tax bargain between rulers and ruled—external threats (or ambitions), the balance between presidential and parliamentary powers—are effectively off-limits for detailed tax advice. The special governance challenges for resource-rich countries, which the state-building literature particularly highlights, have been recognized for many years, with a focus on the use of funds to avoid squandering of resource wealth now widespread, initiatives underway to foster transparency in the design and implementation of resource tax regimes,46 and a continuing debate on whether to hand out resource wealth directly to citizens. The problems are far from solved, of course, but few new tools seem to have emerged.

In relation to non-resource taxation too, much of what is now standard seems relevant to wider state-building issues, and indeed is often motivated by similar concerns. Core elements of standard advice include, for instance, reducing exemptions—often as much on the grounds of the abuse and non-transparency they invite as anything else—and publishing tax expenditure budgets, eliminating discretionary powers, moving to self-assessment and other devices to separate the assessment and collection functions (minimizing opportunities for corruption), developing taxpayer services, consulting with the private sector and (an increasing focus) building tax policy units capable of informing, proposing and taking ownership of major reforms. The possibility of improving accountability by decentralization of tax powers, which the state-building literature appears to favor, has also been a feature of advice for many years, though views continue to differ as to whether or not this favors accountability and good governance as much as one might like to hope.47

While there is thus much overlap between the emerging state-building agenda and current day-to-day advice, there are perhaps two main areas48 in which one can see potentially significant differences.

First, a key theme of the state-building literature is the importance of taxpayers seeing their money going to some worthwhile use, and whether there might thus be a case for more purposive use of earmarking (a thought carefully kept alive for many years by Bird (1997), for example). Generally decried by public finance economists, earmarking is in fact relatively uncontroversial in some areas, including social contributions, resource funds (emphasizing here the ‘relatively’), and even in the allocation of particular revenues to particular levels of government. The issue is whether its use should become more widespread. There are examples of reforms that might have proved difficult without earmarking—Ghana raised the standard VAT rate from 10 to 15 percent in recent years by dedicating the additional revenue to education and health—and certainly a theoretical case can be made for earmarking funds in the face of untrustworthy politicians (Brett and Keen, 2000). The difficulties remain, however. Earmarking creates pots of money that can invite corruption and, unchecked, it can lead to a plethora of small nuisance taxes—as indeed seems to have been the experience in some developing countries. If it acts as genuine constraint on spending on particular, it can lead to harmful inflexibility; if it does not, then it is “an exercise in…misleading taxpayers rather than expanding democracy” (Institute for Fiscal Studies, 1993, pp. 64–65).

The need to establish better links between the pain of paying taxes and the enjoyment of public spending remains, nonetheless an important point. Much of the answer surely lies in improving public financial management and transparency. But it is also a fair criticism that advice on, say, broadening the VAT base, has often been accompanied by generic references to dealing with any adverse distributional consequences on the spending side rather than by recommendations in terms of specific instruments (actual or potential).

A second area in which the state-building perspective can lead to a differing perspective is in the taxation of small and micro enterprises, and it is here that the differences could prove most profound. In broad terms, the contrast is between an emphasis on, on one hand, the limited revenue potential and high administration and compliance costs associated with such traders and, on the other, the potential benefit from purposively including them in the tax system so as to foster their inclusion in the wider tax bargain by, for instance, demanding accountability from those taxing them. Of course, even the most vehement advocates of focusing on larger taxpayers usually see a role for some simple tax, perhaps a license fee, for smaller traders. And the gain in production efficiency from imposing some tax on such traders—who would otherwise be able to survive at lower levels of efficiency than the fully taxed traders with whom they compete—means that it can be welfare-improving to do so even if the direct revenue gain does not cover the administrative costs of imposing it.49 The question is precisely how much further, and in what directions, a state-building concern leads.

Some seem to attach importance to having as many people as possible to remit tax: OECD (2008), for instance, argues that “the challenge for poor countries is…to tax a larger number of citizens and enterprises more consensually.” This presumably should not be taken too literally, although the account of poll tax riots in Tanzania and Uganda in Fjeldstad and Therkildsen (2008) reminds us that the best way to encourage the exercise of voice in tax matters is by taxing badly. Moore (2008), on the other hand, takes a less extreme view, arguing that all can benefit from greater accountability and responsiveness when only a subset of citizens directly remit tax. Certainly there seems little hope of encouraging improved compliance by smaller taxpayers if larger ones are manifestly escaping tax, legally or otherwise. Another vein in the state-building literature stresses formalization rather than the simple remittance of tax:50 “the goal of MSE taxation should be broadened beyond simple cost-recovery to include the benefits of having more formalized firms in the economy” (Everest-Phillips, 2008, p. 6). This also seems to be recognized in established advice (as for instance in ITD, 2007)—or at least in standard aspiration, since this is hard to do. The point, perhaps, is to understand better the various external benefits—beyond standard production efficiency—associated with regularization, and the role of taxation in fostering it. The possibility of low-level compliance traps, for instance—no one complies because no one else does—is widely recognized (Cowell, 1990). But the theory then also tells us that bringing taxpayers into compliance one-by-one may be doomed to failure: some form of “big push” may be needed. Some also argue, for instance, that keeping proper records for tax purposes facilitates access to credit markets; but, if so, why don’t enterprises do it without any need for intervention—such as the design and provision of accounting software—by the tax administration? There is abundant evidence too that that transactions and informational costs are at least as substantial a deterrent to regularization as taxation. What may be needed is not so much clever tax redesign as a greater emphasis on taxpayer education.

There are other important and wider questions raised by the state-building literature. One is a recognition that foreign aid may have some of the same debilitating effects on tax development as do resource rents, in which case the issue is whether aid conditionality can be configured to mitigate these effects, for instance, by some degree of matching to domestic efforts (an obvious difficulty then being the Samaritan’s dilemma of finding some way to credibly commit to limit support if targets are not met). Another is the nature of tax structures that lend themselves to the most constructive bargaining. For example, high visibility (such as with explicit listing of the tax content in retail purchases) and remittance (the literal payment of tax) presumably both have some significance. Similarly, the perceived fairness and distortionary costs associated with any tax are critical—with this factor again pointing to the awkward and presumably repugnant conclusion that it is unfair and inefficient taxes that are particularly likely to prompt citizens to call governments to account and limit any tendency toward inefficiently rapid growth.

Whether this newer literature on state building will bring substantively new insights, or lead to substantially improved practical advice, thus remains open to question. All to the good, however, is that the emergence of the “new fiscal sociology” has widened the perspectives and skills brought to the debate. Not the least of the merits of the state-building literature, however, is to emphasize the sustained, long-term nature of the effort needed to develop effective tax systems, while at the same time not simply implying, as some of the recent literature tends to do, that nothing can be done until effective institutions have, somehow, been developed. This literature not only offers no quick fixes, but can be in itself a reminder that, as the history of big ideas above suggests, there are no quick fixes.

VI. Conclusions

The renewed interest in resource mobilization by developing countries is cause for optimism. Potentially the most fundamental and lasting consequence may be the increased use of micro data—with several notable examples in this volume—to take understanding beyond the professional hunches of practitioners, often driven by the serendipity of their own experiences and idiosyncrasies of their interests. Such work has fundamentally improved the quality of tax analysis in advanced economies, and, as more and better data become available (not least from tax administrations themselves), has tremendous potential for developing countries. Certainly there is a wide range of critical technical issues hardly touched on here (but taken up, for instance, in IMF (2011)), on which substantial progress hinges and on which firm evidence, and careful reasoning, remains in short supply.

The point stressed here, however, is simply that the practical advice which emerges from analysis and practical experience needs, if it is to be genuinely effective: to be ambitious for near-term gain but distrustful of fads; to recognize that progress is both contingent on, and can be a key factor in, long-term state development; and to be aware that fundamental strengthening of revenue collection will be largely a matter of persistent and unspectacular effort.