Why are there big variations in the proportions of GDP that national governments collect as public revenue? Why do some raise so little? What is their capacity to raise more? And can we in any meaningful sense define and measure that capacity? These largely unanswered questions are attracting a great deal of contemporary interest. Globally, governments are getting deeper into debt. It is hard to see a way out that does not involve higher taxes. Simultaneously, and especially in the global south, governments are not collecting the revenues needed to support the spending that experts believe is needed for developmental purposes.

Lucy Martin’s relatively precise cut into this sprawling terrain is welcome. She sets herself the task of explaining recent variations in the proportions of GDP that African governments collect as public revenue. Underpinning her explanation is a model of the politics of governments’ choices about how much tax to raise that is in some respects original. It is not so slender that it could meaningfully be summarized here. The essence is encapsulated in the phrase strategic taxation. The theory is that, when they tax, governments act not as tax maximizers but as rent maximizers: “… a rent maximizing government wishes to extract as much revenue as possible only to the extent that it allows them to increase spending for their own priorities and consumption” (p. 5). If governments perceive that by raising more tax, they become more accountable, and cannot siphon off any additional rent for themselves, they will forego the extra revenue. Martin never explains how she would operationally define or measure the rent element of tax revenue. Her test of the theory is whether it appears consistent with a range of more specific empirical findings. That is something of a problem, not least because some of those empirical findings are themselves very problematic. I will examine two of them.

Martin assumes that her dependent variable—the ratio of tax collection to GDP, often labeled the “tax take”—is determined entirely in the political realm. It reflects the willingness of governments to push collection in the face of political constraints. This is to ignore a long-standing and continually updated research literature, produced mainly by economists and economic historians, telling us that levels and variations in national tax takes are consistently and very significantly shaped by the structure of national economies. In sum, the tax take is larger in economies with (a) higher-income levels, (b) higher ratios of imports and exports to GDP, and (c) lower levels of agricultural activities. The proposition that the proportion of national income collected in taxation rose as incomes increased was originally formulated as Wagner’s Law in 1877.

In the same way that she ignores the role of national economic structures in influencing the level of the tax take, Martin fails to consider the possibility that the (variable) structures of national tax systems help shape tax politics. She rather adopts a simple model of national tax systems that is invariant across countries and bears little resemblance to any actual national tax system in Africa. There are two core—but largely unvoiced—empirical features of her median taxpayer-citizen-voter model.

One is that the bulk of central government revenue comes from taxes levied on the generality of ordinary (adult) citizens. The population categories “taxpayers,” “citizens,” and “voters” are implicitly, by repeated usage, assumed to overlap substantially. The second feature follows directly: most (adult) citizens or voters are assumed to pay taxes to central government and therefore have a direct interest as taxpayers in national fiscal policy. Tax politics are modelled as if they revolve around interactions between a single coherent state authority and the generality of undifferentiated taxpayer-citizen-voters. Those two assumptions are to some extent valid for the United States and other high-income economies that lack great natural resource wealth. Most rich country governments get most of their revenues from a combination of personal income taxes and sales taxes (mostly VAT) levied on the final consumers. But for most of Africa these two assumptions are seriously awry.

First, in Africa, it is typically large companies, not citizens, who are the largest single government revenue source. Rwanda provides detailed and reliable revenue statistics. In the 2019–2020 fiscal year, just 375 large Rwandan taxpayers, accounting for 0.06% of all registered taxpayers, remitted 58% of all tax revenue. 843 medium taxpayers (0.36% of the total) remitted 12% of revenue, and the 99+% labeled small and micro taxpayers remitted 30% of revenue. Second, very few Africans pay taxes to the central government. In Rwanda in 2019–2020, the total number of registered taxpayers was equivalent to about 3% of the total adult population. The figures we have for other African countries are generally in the same ball-park. African tax administrations and governments concerned about increasing their tax revenues have little incentive to bother much with smaller taxpayers. In large part, the smallest taxpayers simply cover the salaries of the tax collectors who bother to chase them. Governments pay most attention to small numbers of large companies.

Martin is not the first researcher to be seduced, through misleading survey results, into believing that most adult Africans pay taxes to central government. Over a series of rounds in more than 30 countries, starting in 2011–2012, a small module of tax questions was inserted into the regular Afrobarometer survey. The questions and countries covered have varied over time. Most commonly, respondents were asked a few attitudinal questions and to state whether they personally were “required to pay” any of several taxes listed. They were not asked to identify the government agency that required them to pay. Martin, like some other researchers, has (a) interpreted “required to pay” as “actually paid” (in the survey reference period) and (b) assumed that all these taxes were paid to central government. Using these interpretations of (ambiguous) survey data, Martin produces a table on p153 that suggests that about 67% of African adults surveyed by Afrobarometer were central government taxpayers. That is a world away from the actual figures cited above. Why the big difference? It is likely a combination of three reasons. First, the Afrobarometer survey questions are ambiguous. Second, the results of other surveys indicate that African are generally reluctant to tell casual interviewers that they are not reliable taxpayers. Third, most of them do indeed pay taxes but not to central government. More pay taxes to sub-national than central government, and even more pay “taxes” of some kind to a wide range of semi-state or non-state agencies, ranging from community, ethnic, and locality associations to service providers of various kinds to bandits and protection rackets.

Martin’s core project is well worth pursuing. But this iteration provides a warning about becoming too immersed in the politics of taxation without also understanding the character and diversity of tax systems.