Morten Jerven
Recent modern mainstream economic research on Africa has focused predominantly on explaining its relative lack of material wealth and on unearthing the cause of its seemingly chronic monetary poverty. Such a comparative perspective suffers from some weaknesses and shortcomings. This needs to be addressed by historical inquiry into the wealth and poverty of African nations. In this chapter I shall first provide a review of the contemporary economics and economic history literature, and then review the available evidence on relative wealth and poverty within the African continent, before finally looking at and suggesting interpretations of trajectories of income and wealth.
Why is Africa poor? This has been the central research question for much recent empirical work in economics and economic history (Acemoglu and Robinson 2010). In the mainstream literature it has largely been taken as a given that the answer to the question involves explaining relative poverty measured as lower GDP per capita “today” in African countries, compared to GDP per capita in other places, such as Germany and Canada. The empirical literature has found systematic correlations between plausible causes that could harm economic performance today – such as incidence of malaria, exports of slaves, and colonial regimes (for a review see Jerven 2015). Models that seek to quantify these historical events, and to isolate cause and effect, have argued that these and other historical events have determined the relative wealth and poverty of African countries as well as vis‐à‐vis the rest of the world. While the methods are new, some of these answers cohere with well‐known hypotheses in African history. The contributions have been cautiously welcomed as a “new economic history of Africa” (Hopkins 2009), yet its (a)historical approach has also been forcefully criticized for presenting a “compression of history” (Austin 2008).
It was often lamented that economists ignored Africa’s past when analyzing the causes of its economic performance (Manning 1987; Hopkins 1986). This was most evident perhaps in the debates on why Africa was growing slowly in the 1980s and 1990s, where one camp of scholars, mainly economists, argued that the disappointing economic performance in the 1980s and 1990s was internal to African economies. The position was most strongly associated with the International Monetary Fund (IMF) and the World Bank, and most clearly expressed in the Berg Report, issued by the World Bank in 1981 (World Bank 1981). According to the report, the causes of unsatisfactory economic performance were chiefly inappropriate economic policies. More specifically, excessive state intervention in the economy was judged to be the policy mistake, and consequently the policy reform enforced through structural adjustment lending prescribed liberalization and privatization, as encapsulated in what was called the Washington Consensus. On the other side, the externalist position argued that the IMF and World Bank diagnosis underplayed the strength of external factors in determining economic fortunes and that it overestimated the potential efficiency of markets in African economies. Disregarding what would have been the correct diagnosis, the internalist interpretation proved more influential and formed the basis for the treatment prescribed by structural adjustment in the 1980s and 1990s.
According to the available evidence, these decades did not see any marked improvement in economic performance, and perhaps even the contrary. The political scientist Nic van De Walle (2001) summarized the two decades (1979–1999) as the “politics of permanent crisis,” whereas the economist William Easterly (2001) summed up the same period as the “lost decades” and concluded that, despite policy reform, there was no growth. The lack of success in reforming led to change in some of the thinking around what matters for economic growth and therefore determines wealth and poverty in the long run. The paradigm changed from “getting the prices right” to “good governance,” reflecting a change in perspective from emphasizing the set of polices to the implementation and adherence to these policies.
Concurrently, there was a parallel movement in the empirical work on economic growth by economists. Elsewhere I have described this as the first and the second generations of economic growth literature (Jerven 2015). Following the seminal publication of a model that used cross‐country regressions with global data sets on growth in GDP per capita, and looked for correlations with other variables such as educational attainment, number of assassinations, black market currency rates and more, a literature developed that was aiming to find the global “determinants of growth” (Barro 1991; Jerven 2011a). When Barro had checked and controlled for a lot of different variables, one central finding remained. There was a large and significant African continent “dummy variable.” A dummy variable takes the value 1 or 0; in this case it took the value 1 if the country was situated on the African continent and 0 if it was not. In the regressions, the dummy variable remained significant. Barro’s interpretation of the dummy was that the analysis had not yet fully captured the characteristics of a “typical” African country (Barro 1991: 437).
Thus, a scholarly search for the right variables began and, while over the course of a decade of research many correlates with slow growth were found, the African dummy variable proved stubborn to remove. I have labeled the literature that shares this motivation the “quest for the African dummy” (Jerven 2011a; see also Englebert 2000) and, as I have argued elsewhere (Jerven 2015), the quest was ultimately both unsuccessful and highly influential.
In a review of the empirical growth literature,1 Durlauf and colleagues referred to this scholarly production over the following decade as a “growth regression industry” (Durlauf, Johnson, and Temple 2005: 599). As early as 1998, Pritchett observed that the growth experience of most developing countries had been characterized by instability rather than stable growth and warned that the “exploding economic growth literature” was “unlikely to be useful” (Pritchett 1998: 3–4). The general growth regression literature has been described as disappointing; one assessment concluded that the “current state of the understanding about causes of economic growth is fairly poor” and that “we are in a weak position to explain why some countries have experienced economic growth and others not” (Kenny and Williams 2001: 15).
Whereas the ultimate outcome of the search for causes of slow growth in Africa was inconclusive, an unintended consequence was the idea that African economies were captured in a chronic failure of growth became an accepted stylized fact within economics. Ultimately, this led to growth econometricians, equipped with new models and data sets, searching for the cause of lack of income instead of lack of growth. The main contenders in the empirical literature searching for a root cause of underdevelopment can be organized in three chronological strands according to whether they emphasize the negative effect of initial conditions particular to Africa’s geographical characteristics, the decisive impact of the slave trade, or the effects of the European colonization, as summarized in Battacharyya (2009) and Nunn (2009).2 Many of the contributions are variations of the same argument: that a particular historical event or factor endowment led to a particular institutional constellation that has had lasting economic effect. The big question in the literature remains which of these historical events had the decisive impact and, secondly, through which transmission channels it continues to have an effect. A divisive issue in the economics literature is the relative importance of institutions and geography or, as was succinctly phrased in debate, “institutions rule” (Rodrik, Subramanian, and Trebbi 2002) versus “institutions don’t rule” (Sachs 2003). The list of factors and variables are growing and, at times, they are in conflict with each other. However, they are also often listed as a growing body of evidence that indicates that “history matters.” A historian might reply: Which one? It appears that the approaches to the past are quite different between economists and historians.
The argument that Africa’s root of underdevelopment lies in its engagement with the West, and more specifically Europe, is not new (Rodney 1974). Nor has it gone unappreciated that the geography or factor endowments of Africa have been particularly challenging (Austin 2008). For historians and economic historians, it might be hard to accept that arguments that are not new seem to gain another level of influence, solely it seems, because they are delivered by econometric techniques. Moreover, it might be frustrating that new sets of concepts are not engaging already accumulated knowledge and concepts that are accepted in the discipline. One example is the pair of “extractive” and “productive institutions” set up in colonies, which according to Acemoglu, Johnson, and Robinson (2001, 2002), should explain relative income per capita today. That result, subject to caveats about the quality of the data, may be econometrically robust to 95 percent statistical significance, but it does not cohere very well (or even engage) with the existing categories of colonies in African historiography where the distinctions between settler and peasant colonies have been very important. For example, Bowden, Chiripanhura, and Mosley (2008) compiled wage data, poverty data, and real wage data for three settler economies (South Africa, Zimbabwe, and Kenya) and three peasant economies (Ethiopia, Uganda, and Ghana) and compare long‐term trends in trajectories and levels in living standards. Their main contribution was to challenge the thesis by Acemoglu, Johnson, and Robinson (2001) that suggest that higher levels of European settlement led to more productive institutions that predict better economic outcomes today. On the contrary, higher levels of poverty and lower wages (in particular) were observed in the settler economies.
As mentioned, economic historians of Africa previously lamented that economists took 1960 as their starting point, as if events preceding this point in time had no effect on current development (Hopkins 1986; Manning 1987: 50; Austin 2007: 12). There was an informal division of labor between the disciplines: economists studied the post‐1960 period, while historians were mainly occupied with the pre‐1960 period. Some time ago, Ellis (2002) issued a call for historians to write “the history of contemporary Africa,” which partly reflects that 1960 is becoming a more distant part of the past. The disciplines are thus finally intersecting, if only in terms of the time period studied.
There may therefore be good reasons to be excited about the fruits of cross‐disciplinary work. But there is also good cause to be aware of the potential for conflict and misunderstandings.3 Historians and economists differ in the types of questions they are interested in, how evidence is dealt with, the role of theory and models or, to put it simply, there are important methodological differences between economists and historians. I have discussed the main differences in approach between economists and historians as they interpret social and economic change in the African past elsewhere (Jerven 2011a); here I attempt to survey some of the new empirical work that has been motivated by the explicit or implicit aim of testing them. I am not referring here to competing big tests, such as a different type of specification that shows that it was the prevalence of malaria or of tsetse fly, the absence of cereal crops, the numbers of slave exports, or the numbers of settlers that has the largest explanatory power in a regression framework. I am thinking about the work that seeks to “decompress” history, and particularly work that seeks to move beyond explaining “today” as a static explained by some historical “event” some time ago, but rather seeks to analyze political and economic change and to explain trajectories of development.
Of course, it is difficult to decide whether these large narratives and their accompanying regression results are genuinely testable – it is not clear to me, for instance, first of all, whether they are making a theoretical claim such as “private property rights are important” or a historical claim such as “the failure of the king of Kongo to formalize private property rights for farmers explains why Angola is poor today.” In a critical review of the “reversal of fortune thesis” (Acemoglu, Johnson, and Robinson 2002), Austin (2008) suggested that it may already be the most discussed contribution to the study of the economics of growth and development since Arthur Lewis’s (1954) model of development, with unlimited supplies of labor. The comparison is instructive. In a dual sector model Lewis postulated that there was an untapped labor supply in the rural agricultural sector, and from this he derived a possible model for future economic development. This model has been applied empirically to explain particular African historical economic developments, such as in Barber’s study of Rhodesia (1961). That specific application of the model was subject to a sharp and still relevant critique by Arrighi:
Barber exemplifies the ideological bent of the anti‐historical approach which is the essence of modern economics. For in economics assumptions need not be historically relevant. In fact, they are often plainly untrue and recognized as such. Historical processes fall into the background and are summarized by statistical series of ex‐post data, the “stylized facts” as they are sometimes called, which by themselves reveal nothing about causation. Thus, all that Barber takes from the complex historical process which we have been analyzing are a series of real wages and a series of rates of African participation in the labour market. Causal relations, on the other hand, are not derived from historical analysis, but are imposed from without, that is, through a priori analysis: and a set of assumptions which yields the stylized facts is held to have explanatory value, irrespective of its historical relevance. But since there will be normally many such sets, this methodology leaves room for considerable arbitrariness of choice and therefore mystifications of all kinds. In view of this, the low scientific standards attained by modern “development economics” should surprise nobody. (1970: 227)
How do the recent contributions from development economics compare to classic models such as the model of unlimited supplies of labor? The Lewis model used a few simple stylized facts to gain insights and to derive policy implications regarding conditions for future economic development. The model was logically deduced from assumptions. These assumptions could be argued to hold or not, and the model could be deemed applicable or inapplicable to specific empirical studies. It was not offered as a grand explanation of historical development patterns, induced from empirical “evidence,” or presented as the outcome of a “natural experiment,” unlike many of the recent contributions from the new economic history of Africa.
Despite development economists’ repeated declaration of the mantra “history matters,” there persists a serious neglect of the qualitative historical literature and the historiographical lessons drawn from the discipline of African history, to the extent that some historians may regret what was called for in the first place. In the words of Robert Jenkins:
when faced with studies by economists who use history mainly as a source of data with which to advance unsubtle hypotheses concerning the causes of developmental outcomes, those who had earlier called for scholars to pay more attention to history may regret ever having voiced such a plea, and find themselves revisiting the proverb about being careful what one wishes for. (Jenkins 2006: 7)
The concern about compression of history is not only a disciplinary quibble on what constitutes a correct approach to history but may also have significant implications for policy. History certainly matters, but perhaps those cases where difficult historical legacies were overcome, even temporarily, could offer sources of useful policy advice. It would be very likely that a historian, and in particular an economic historian, would find it difficult to disagree with the proposition that history matters for political and economic outcomes. Partly because of this recognition, there are PhD candidates in North American economics departments who combine the use of economic models with careful archival work to further knowledge of Africa’s material past. And because of that we have accumulated some new evidence and knowledge; in particular we have more quantitative evidence on the colonial period than was available before.
African economic history is experiencing a renaissance (Austin and Broadberry 2014). One of the early and central contributors to African economic history, A. G. Hopkins (2009), called the new surge of literature on material change in African economies the “new economic history of Africa.” According to Hopkins, this new literature had appeared largely unbeknownst to historians, having originated in North American economics departments. Hopkins noted several shortcomings in the historical approach, yet invited historians to engage with the research. These seminal papers succeeded in shifting the focus in development economics from associating growth with policy, and turning attention toward the historical causes of wealth and poverty among nations (Acemoglu and Robinson 2012).
Whether this new economic history of Africa is conceptually and methodologically coherent with the “old economic history” of Africa has been hotly debated (Jerven 2011a). In particular, there was an exchange as to whether the new contributions should be called “causal history” (Fenske 2010) or whether the methods result in what Austin called a “compression of history” (Austin 2008). It is, however, beyond doubt that the seminal papers published more than a decade ago had an impact on the sheer amount of empirical work done in African economic history in recent times.
The contribution of this short section is to take stock of the many empirical contributions since the mid‐2000s. A wealth of new historical data has been unearthed, collated, and organized – a lot of it from colonial archives but some of it also from other sources such as military files recording the heights of soldier recruits and trading companies recording prices and wages. The new estimates of levels and trends in wages, growth, living standards, and taxes have, to date, been presented in a piecemeal fashion. This new empirical work has not yet been synthesized, and it is time to start thinking about how this new knowledge changes conventional knowledge, and to assess how this decade of research has challenged and refined those big research questions posed a decade or more ago.
The question has also gained some impetus from contemporary development in African economies. Africa has moved from being depicted as the “hopeless continent” to being described as the hopeful continent that is on the rise (Jerven 2015). In turn, there is an increasing demand for knowledge on the material history of change in Africa. Focus is shifting from explaining chronic growth failure in the so called “bottom billion” (Collier 2008) toward research that can shed light on the role of states in fostering industrial change, and situating the current period of economic growth, or “Africa rising,” in the history of economic growth in African economies.
The availability and the character of poverty numbers and economic statistics have shaped the historical analysis of poverty and growth in sub‐Saharan Africa. Quantitative economic histories of economic growth are written using time series data on economic growth. In this case year zero has been 1960. The availability of this evidence in part explains why economists have focused on the postcolonial period. If one writes histories of poverty, using the dollar per day metric, year zero has been 1990. Both are, of course, artificial starting points, but data availability restricts us to this time frame and to the kinds of questions that can be investigated.
This history of poverty in sub‐Saharan Africa has been shaped by definitions of poverty and the availability of information that fits those definitions. The narratives of economic growth in African economies change dramatically if the starting point of 1960 in the history of economic growth in Africa is rejected. Notions such as “chronic growth failure” and the “bottom billion” would not hold if we were to remove the short time horizon.
Similarly, narratives of trends in living standards in Africa are shaped by the current configuration of what constitutes poverty knowledge, particularly at the World Bank, and what does not. That may be a mistake because the short history of poverty by numbers in Africa is characterized by gaps and inaccuracies in the underlying data, and because it fails to contextualize the history of poverty in the 1990s and 2000s in a longer time perspective, and thereby persistent chronic poverty became the “imagined fact” that prompted the narrative on Africa and poverty.
One of the central challenges is to connect the colonial period to the postcolonial period. When this is done, there is a conspicuous absence in the central narrative of the economic history of twentieth‐century Africa. In previous analytical models, the postcolonial period has been treated as an “outcome” (and a dismal one), whereas a historical event in the colonial or the precolonial period has been treated as a root cause of this outcome. The aggregate pattern, particularly in terms of poverty and growth – though it is also evident in taxation measures – is a prolonged period of growth from the 1890s into the 1970s. This gave way to widespread failure and decline in the 1980s, followed by two decades of expansion since the late 1990s. Thus, it seems increasingly evident that growth failure in the postcolonial period has taken up undue space in explanations of African economic development. The old pattern in the literature was to regard the postcolonial period as a dismal outcome, as represented by a low GDP per capita today, which should be explained by a colonial or precolonial event or variable. With new data sets, the boundary of investigation is pushed backwards to allow the evaluation of trajectories of economic development across the colonial and postcolonial period.
I have argued that the interest in African economic history in recent influential economic research was motivated not by an interest in history as such, but by the fact that history provided the instrumental variables that econometric techniques needed to explain Africa’s relative poverty today. It combined evidence from the very recent past (typically GDP per capita today) with evidence capturing some historical event (slave exports, colonial settlers, and geographical variables) to find a root cause of the relative underdevelopment of African economies. Thus, this analysis of economic growth in Africa gave the impression that African economics have been stuck in zero growth equilibrium for centuries.
A longer time perspectives makes it clear that growth has been recurring in African economies (Jerven 2011a). New GDP estimates for the colonial period, and historical proxies for GDP growth, such as terms of trade, show that periods of economic growth are not new in Africa, and thus seek to place the recent decades of “Africa rising” in historical perspective. There is considerable regional variation but the presence of pockets of robust growth, in some places from the late 1890s and lasting well into the 1970s, actually puts recent research into a new perspective. The growth literature has been unduly influenced by the so‐called lost decades of the 1980s and the 1990s. This ahistorical perspective is worth correcting, particularly because it opens new research avenues, such as a comparative political economy of growth perspective from the colonial period, the postcolonial period, and more recent growth since the 2000s.
A general misconception is that our knowledge by numbers in the postcolonial period is overrated and this is mirrored in underrating the numerical basis of knowledge for the colonial period. The GDP estimates for the postcolonial period depend on a lot of assumptions to make up for the missing data (Jerven 2013b); the levels of GDP, in particular, are subject to large errors (Jerven 2015). There is relatively consistent data on government revenues and expenditures for the colonial period, as well as fairly reliable data on exports and imports, at least for the goods that left through ports (but less so for trade across internal borders). In one paper Jerven (2014b) uses these data to estimate GDP growth for the Gold Coast, offering a new GDP series for colonial Ghana for the period 1891–1957.
What are the broader implications of these time series? Current databases on GDP for sub‐Saharan Africa only go back to 1960 or 1950. Extending the data series for Ghana and for other countries in sub‐Saharan Africa will be of analytical benefit for several reasons. First of all, it is of great importance which year is treated as year zero in any investigation. A striking example of the power of data sets and the importance of starting points comes from the debate on global warming. Maslin (2004: 40) argues that it took so long to notice global warming because of weaknesses in the global mean temperature data set. From the viewpoint of the 1970s, it looked like the temperatures had been falling since the 1940s, but the current data set, covering 1860–2010, shows that the temperatures are indeed on a rising trend.
Similarly, while growth in Ghana slowed down in the 1960s, the general pattern in sub‐Saharan Africa was growth in the 1960s and the 1970s (Jerven 2011a, 2014a). In previous analysis, this growth has seemed like a short blip before the onset of stagnation and decline from the late 1970s. An expanded data set – with coverage for Ghana and beyond – provides a different perspective. First, it becomes easier to understand the current rise of Africa. Growth has been and is compatible with imperfect institutions, but sustained growth and diversification remain an elusive goal. Second, the availability of a GDP series for Ghana and for other countries allows a quantifiable perspective on the political economy of growth in colonial Africa. Recent work has quantified changes in real wages and living standards in Ghana and elsewhere (Austin, Baten, and Moradi 2007); as of the time of writing, it has been difficult to assert whether rates of change in living standards, observed improvements in heights, or trends in fiscal spending and taxation are simple correlates of economic growth, or whether different political economies (settlers versus peasants colonies) determined a particular distribution of rents and economic growth in the colonial era (Bowden, Chiripanhura, and Mosley 2008). These valuable prospects for comparative work on African colonies and colonies in other regions will be possible as verifiable estimates of population and GDP become available for scholars of economic history and students of the history of development.
In one sense, estimating GDP growth for colonial Ghana does not show us anything that is completely unexpected. The way in which the series is constructed for GDP is driven primarily by what happens in external trade (which we know increases rapidly for Ghana) and by the expansion of the public sector, and we know that the colonial administrations expanded in this period as well. When growth is computed this way, we may expect to see growth in practically all African colonial economies. However, care must be taken when interpreting the series. Interesting as it is, it is still based on the relatively limited available data. It has been argued that the level estimates that are derived from such sparse data cannot be readily compared with areas where more data are available (Jerven 2012), but it has also been pointed out that growth estimates are better than the level estimates (Blades 1980). As previously mentioned, estimates of contemporary GDP per capita in Africa and Ghana vary considerably according to the methods, data, and benchmark years that are in use (Jerven 2013a, 2013b).4
A key question in the literature on the cash crop revolution in general has been how to interpret the visible aggregate export data. How do they relate to development in the whole economy? Basic models of economic development used to assume that the opportunity cost of modern sector growth was zero. In the vent for surplus model both land and labor are assumed to be abundant (Myint 1958). In the classical dual economy model proposed by Arthur Lewis (1954) the basic assumption is that marginal productivity of labor is zero in the “traditional” or “rural” economy. In such a setting the modern sector growth was an absolute benefit. However, it was not true that the marginal productivity of labor was low or close to zero. In contrast, the vent for surplus model does not make assumptions about labor productivity, but instead assumes that there is excess capacity of land and labor, and thus the external market opportunity appears as a vent for surplus. Recent empirical quantitative research on the Gold Coast has supported the interpretation that export growth was associated with a net benefit and that until the 1920s, in particular, there was no trade‐off between food production and cocoa production (Moradi, Austin, and Baten 2013: 22).
This is not the consensus in studies on other countries. Scholarship has, in different ways, contested these assumptions and by extension the analysis using these models. Most importantly, it has been pointed out that labor was abundant only seasonally and very scarce in certain periods (and more so in areas outside of the forest belt), that the production of exports involved both innovation and capital (investment in new technologies), and that the opportunity costs were not zero (food quality and security, labor division, and manufacturing all changed considerably) (Tosh 1980). These and other empirical contributions all remind us that when we see aggregate modern sector growth it is not equivalent to observing aggregate economic growth. Thus, the assumptions made in these estimates do not hold indefinitely for Ghana; they need to be qualified for some countries and rejected for others. In short, these quantitative studies need to be supported by other evidence that can give us information about what happens in the unrecorded economy.
For Ghana, the GDP series brings support to the argument that the manifest growth in production, exports, imports, and increases in living standards in Ghana and other nonsettler economies in sub‐Saharan Africa during this period may be contradicting the claims put forward in the work of Acemoglu, Johnson, and Robinson (2001, 2002) and of Nunn (2007, 2008) and others who have suggested different variations of the proposition that these economies were stuck in a slow‐growth equilibrium during this period as a consequence of institutions. Certainly, this growth period was not indefinitely sustained in Ghana, but it should nevertheless indicate that the prevailing institutions in this period, be they rules of taxations, property rights, or general rule of law, were not incompatible with economic growth. Other work on various areas in the less developed regions of the world may confirm or reject whether Ghana was an exception or typical case of this period. Alexander Gerschenkron suggests that if estimates from quantitative data “are at variance with what we should expect from general historical knowledge, they should be rejected” (quoted in Platt 1989: 1). Thus, while this GDP series was prepared with parsimonious data and assumptions, it coheres with other findings. This points the way forward for other efforts of quantifying economic change in colonial Africa.
There is a long history of poverty in Africa. However, the most influential narrative of African poverty is a very short one. The history of Africa by numbers as told by the World Bank starts in the 1980s with the first Living Standards Measurement Study. It is also a very narrow one. In general, there is a disconnection between the theoretical and the historical underpinnings of how we understand and define poverty in Africa and how it has been quantified in practice. This chapter reviews how particular types of poverty knowledge have gained prominence and thus shaped the historical narrative of poverty in Africa. It summarizes recent work on living standards. New sources on real wages and evidence from anthropometric research provide perspectives on trends and relative levels of living standards between the 1890s and the present. This raises the possibility that the narrative of African poverty that was born in the 1980s is a historical anomaly. Such a perspective may also offer a better perspective from which to reach a historical comparative verdict on the more recent “Africa rising” narrative. First, there is nothing phenomenally new about growth in African economies; and, second, since evidence of economic growth leading to poverty reduction during the recent period is weak, the period of growth and poverty reduction may compare unfavorably to growth in the 1960s or even in the colonial period.
Bowden, Chiripanhura, and Mosley (2008) compiled wage data, poverty data, and real wage data for three settler economies (South Africa, Zimbabwe, and Kenya) and three peasant economies (Ethiopia, Uganda, and Ghana), and compared long‐term trends in trajectories and levels in living standards. Their main contribution is to challenge the thesis by Acemoglu, Johnson, and Robinson (2001) that suggests that higher levels of European settlement led to more productive institutions that predict better economic outcomes today, and that, on the contrary, higher levels of poverty, and in particular lower wages, were observed in the settler economies.
Their investigation shows that real wages stagnated in the settler economies, whereas in the peasant economies (Uganda and Ghana) real wages rose markedly from the 1910s into the 1970s. A similar pattern, without separating settler and peasant colonies, is found by Bourguignon and Morrison (2002) using a large database of compiled household surveys. They note that “in 1950, only 12 percent of world inhabitants with incomes of less than half the world median income lived in Africa. By 1992, 30 percent did” (Bourguignon and Morrison 2002: 739). A long‐term historical reading of poverty in Africa would raise questions about viewing poverty as an “African” structural problem, a point supported by Rönnbäck’s (2014) analysis of living standards in the precolonial Gold Coast.
Real wages are calculated by comparing annual nominal wages with the cost of goods that workers consumed at a given time. This process involves a series of assumptions and calculations (about annual household labor, consumption, income, and purchasing power) to determine if a representative household could meet its daily subsistence requirements. The argument is that when income was not enough to meet these requirements poverty is present. This method calculates household income from the records of colonial ledgers. These ledgers typically reflected daily wages, though they could be reported as averages or in terms of minimums and maximums.5 Still following Allen (2005), price data are then compiled for a “bare‐bones” subsistence basket of goods – the absolute minimum daily requirements (calories, protein, fuel, and other goods) for an adult male. It is then assumed that an average family of two adults and two to three children would require three of these consumption baskets, the equivalent of a household subsistence basket. Historical prices are compiled from a variety of sources (company books, trade statistics, government records) in order to calculate how much this household basket of goods would have cost at any given time. These figures are compared to the household incomes calculated above to determine a household’s subsistence ratio. A ratio of one indicates that household income is exactly what was required to purchase a household subsistence basket. Estimates less than one indicate that the household income was not enough to meet needs (poverty), while a figure greater than one indicates that the household is better off.
Standardizing income and needs in this way allows comparison of well‐being across both time and distance. Frankema and van Waijenburg (2011) use this approach in their analysis of living standards in eight British African colonies from 1880 to 1940. Specifically, they argue that real wages exceeded subsistence levels for this period and that wages in a selection of the colonies studied actually exceeded those in major Asian cities. Further, the authors highlight that a divergence of paths has occurred not only between Africa and these comparable Asian locations but also within the African continent. They point to the different paths that led to present‐day poverty and challenge the argument that colonial institutions and path dependency are at the root of poverty in sub‐Saharan Africa.
Making direct reference to, among others, Nunn (2007, 2008) and Acemoglu, Johnson, and Robinson (2001, 2002), Frankema and van Waijenburg (2011) gather evidence that poverty was more of an Asian problem than an African one until relatively recent times. In making this argument, they take a step to decompress history and to provide the nuance that a more diligent reading of history allows.
A more direct way of measuring living standards is to draw on the tools of anthropometrics, or the collection of measurements and statistics relevant to the human body. Specifically, researchers have used mean adult height as a measure of well‐being in a society by highlighting that stunted adult height is a result of the deprivation of essential needs in childhood. Such methods are prefaced on the argument that height is a measure of one’s physical quality of life and that it is positively impacted by the quantity and quality of childhood nutrition, housing, sanitation, and medical care. Conversely, conditions of childhood poverty (such as disease and physical labor) expend nutrients and have the ceteris paribus effect of stunting growth (Moradi 2009).
This method pre‐empts the critique that genetics impact height by using average population height and by pointing to the fact that height is normally distributed across a population. Thus, when one aggregates to the population level, the influence of genetics is cancelled out. One fact that is relevant here is that large height differences have been found to be more prevalent between rich and poor people from the same ethnic group than between rich elites of different groups (Austin, Baten, and Moradi 2007). At the macro‐level, it is suggested that the genetic pool can be assumed to be fairly constant given the relatively short period of time that such studies typically focus on (Moradi 2009).
It is suggested that the use of average height as a measure for quality of life has a variety of advantages. First, because height is an outcome (rather than an input, like income) of well‐being, this method does not discriminate between different varieties of input (avoiding the need to standardize consumption baskets) or make any assumptions about basic minimum requirements. The primary point of interest is whether an individual’s condition was sufficient to meet their needs. Further, height measures are applicable to the diverse socioeconomic conditions of developing countries (hunters, pastoralists, subsistence farmers, urban laborers) and thus facilitate comparison across groups. Mean height is also sensitive to inequality in that, as a consequence of the diminishing returns of nutritional input on height, redistribution from rich to poor will raise the mean population measure. Finally, it is suggested that height data allow for statistical insight into historical periods for which other measures of poverty are lacking (Moradi 2008).
Such methods are generally restricted to relative measures of well‐being. Specifically, mean adult heights are compared for different birth cohorts. If one group has a greater mean height than another, well‐being is assumed to have increased through an improvement in overall well‐being, a reduction of inequality, or both. The most readily available height data prior to the Living Standards Measurement Study are typically found in documents associated with the slave trade or colonial armies. Of course, these samples are not random and they may not always be representative. A preference for physically fit (taller) slaves could skew the sample mean, and military enlistment was often voluntary, paid, and subject to minimum height requirements.
Austin, Baten, and Moradi (2007) analyze regional inequality in Ghana from 1880 to 2000 and find that living standards improved during colonial times and then regressed after the economic crisis of the 1970s. In doing so, they employ data from a variety of times and sources. First, attestation forms facilitated the estimation of average heights for those born between the 1880s and 1920s. To account for minimum height requirements, the authors applied truncated regressions using a maximum likelihood estimator in order to establish mean heights that are closer to representative for the population. They then compared these averages with data taken from the 1987–1988 Ghana Living Standard Survey and Demographic Health Surveys (DHSs) from 1988, 1993, 1998–1999, and 2003 in order to evaluate the development of regional differences in Ghana over time.
Moradi (2008, 2009) follows a similar methodology in refuting the notion that contemporary poverty is rooted in colonial legacies. Using records from colonial armies, he argues that Ghana and Kenya experienced significant improvements in health and well‐being as measured by average population height. Moradi compares improvements in this metric with data from other developing countries to demonstrate that Ghana and Kenya were outperforming comparable international countries during colonial times (1920s–1970s).
It is beyond doubt that taxes matter for economic development and state formation. Not only is there an obvious link between the capacity of state to collect taxes and its ability to deliver development outcomes such as expenditures and investments, but, more subtly, there is indirect evidence supported by theory that there is some stimulation of economic and political development through taxation because it strengthens the contract between citizen and state (Besley and Persson 2014). There has been a longstanding, if sometimes disappointingly vague, conversation in development economics and development studies about how governance and institutions matter (Jerven 2015). Studying the institutions that tax and measure how much they collect seems like an obvious way to pin down a question of potentially great importance.
How have African states evolved over time? This chapter addresses the prominent, and sometimes disappointingly vague, debate in the social sciences on the role and persistence of institutions. Studying the fiscal institutions of African states provides an easily observable and measurable way of tracing the evolution of institutions through the precolonial, colonial, and postindependence eras.
Frankema (2011) provides an investigation into revenues and expenditures in British Africa from 1880 to 1940. Evaluating the divergence of fiscal policies in British West Africa and East Africa, he calculates gross public revenues and tax burdens for eight colonies. This information is compiled from colonial blue books and other archives, though it is not made explicitly available in the article. Frankema bases his tax burden measure on the fiscal revenue (per capita) divided by the daily wage of an adult male worker (2011: 139). This equates to the average number of male working days required to offset the tax burden imposed on the population for eight British colonies.
Building on this, Frankema and van Waijenburg (2011) compare fiscal policies in British and French colonial Africa from 1880 to 1940 (Table 17.1). Using data from various colonial records, the authors highlight that local conditions tended to trump metropolitan blueprints for colonial governance. The sample of countries covered is greatly increased from the previous article and is also organized according to colonizer’s identity, African region, and other geographical circumstances.
Table 17.1 Number of countries in comparison of fiscal policies in British and French colonial Africa, 1880–1940.
Source: Frankema and van Waijenburg (2011).
| East Africa | West Africa | Other | |
| Britain | 6 | 4 | 1 |
| France | 4 | 8 | 5 |
A significant finding of this article is that gross revenue per capita correlated negatively with the proportion of direct taxes levied on the population in colonial Africa. The authors conclude that, when states were able to raise sufficient capital through other means (taxing trade, for example), governments imposed relatively low direct taxes. One shortcoming of this research is that it details only a select number of years between 1880 and 1940 (1911, 1920, 1925, 1929, 1934, and 1937), providing snapshots in time from which to infer trends rather than consistent and transparent time series data.
Trends in taxation are not uniform, and some countries follow distinct paths. Before the continent‐wide data sets on levels and trends in taxation in African countries across the twentieth century, many questions remain unanswered. A database by Albers, Jerven, and Suesse (2016) presents measures of the fiscal capacity of African states in the twentieth century. A general continent‐wide trend in taxation (subject to a lot of country variation) indicates increases in taxation from the beginning of the nineteenth century through the 1960s and into the 1970s before declining in the 1980s. There is a marked increase in the late decades through the 2010s, yet this increase is driven more by indirect taxes than by direct taxes. The trends in taxation seem to support what is indicated in studies of growth and living standards across the twentieth century.
Since the late 1990s, African economic history has experienced a renaissance. A literature that quantifies development trends in colonial Africa has emerged in response to seminal contributions arguing that the colonial experience explains why some countries are poor today and others are not. We now have estimates of fiscal capacity, GDP growth, and real wages, and anthropometric measures of living standards. The literature has taken advantage of the existence of colonial records, and generally point to an external‐market‐led GDP growth that supported the expansion of fiscal states and were associated with improvements in real wages and human development as measured by stature.6
The seminal literature focused on establishing relationships between historical events and income levels today. This focus is a result of one basic data constraint. Estimates of national income and economic growth for African economies were available only back to 1950. As noted, this has resulted in the economic growth literature, which has been evaluating long‐run economic performance, often using evidence from the very recent past. Typically, the literature has focused on explaining variation in income per capita today, thus asserting causal relationships across time periods without being able to account for different trajectories of economic development, which results in what has been called a compression of history. Certainly, it would be an improvement to focus on explaining trajectories of African economic growth rather than the lack of economic growth. The literature on African economic development has already moved away from a compression of history toward focusing on analyzing and evaluating trajectories of economic development during the colonial and postcolonial period in Africa.
Recently, there have been efforts to provide historical national accounts data for a range of African economies. Particularly for the Gold Coast and Ghana, there is mounting direct and indirect evidence of periods of economic expansion and increases in living standards during the colonial period. Moradi (2008, 2009) uses anthropometric measures and reports that heights (taken from army recruits for both world wars) in cohorts born between 1905 and 1920 outgrew the cohorts born between 1880 and 1893 “by an astonishing 2 cm on average implying a considerable improvement in the physical quality of life” (Moradi 2008: 1113). Frankema and van Waijenburg (2011) collect nominal wages from colonial records and use price data from the same sources to get a measure of real wage developments. They find that real wages started growing in the 1910s and grew rapidly in the 1920s, and that growth was sustained into the 1960s. Bringing further supporting evidence of widespread economic development in the late nineteenth and the early twentieth centuries in the Gold Coast rather than a narrow export expansion, Austin (2008) argues that the expansion in production of cash crops did not cause a net deficit in food production in the area, and that the increase in output was accompanied by considerable capital investment, in form of planting and clearing land for cocoa. Moreover, Jedwab and Moradi (2011) have shown that cocoa production was stimulated by infrastructural investment, particularly in railroads, thus indicating that central government and large‐scale capital investment played a conducive role in this period of expansion. In sum, there is mounting evidence that questions whether colonial Ghana was stuck in a low‐level equilibrium. While a country study like this is not an explicit test of the causality suggested in the cross‐country studies literature, it suggests that further research can establish whether the claims that institutions had a causal effect on economic growth are consistent with trajectories of economic and institutional change.
A new description of growth and development since the nineteenth century underlines the need for new analytics. In his study of poverty in Africa John Iliffe (1987) speaks of conjunctural and structural causes of poverty. Conjunctural causes of poverty are events – failed harvests, disease, and so on – that cause a sudden change, whereas structural causes of poverty are changes in the political economy, such as property rights regimes or institutions of taxation. The analysis of trends in economic growth, taxation, and living standards can identify the relative importance of these types of factors, and thereby allow a return to assessing the relative importance of “policy” versus “shocks,” in order to explain recurring growth in African economic history (Jerven 2011a). Moreover, the combined availability of longer time trends on growth, taxes, and living standards allows for comparisons of the political economy of growth in the waves of external‐market‐led growth in the colonial period, growth in postindependence states, and finally growth in the poststructural adjustment period.
Both the concepts and the source material of this new empirical work focus on states and state records. My forthcoming book will focus on the central themes and questions that these state records can answer – and other issues such as agricultural productivity, land relations, and gender are largely missing, and that perhaps the colonial period has been given undue focus. One of the central lines of my previous research (as summarized in Jerven 2013b) is that what is visible in the state records varies across time and space. That may perhaps be viewed as a weakness of these studies. As a starting point, one should view the state record itself as a kind of fingerprint of the state and its activities: the record is in itself a lens through which we can gauge state activity and capacity. The statistical record that has been collected, systematized, and synthesized does provide a quantitative basis to challenge and to shape existing narratives on African economic development from the nineteenth century into the twentieth century. To push the boundary further, future research should move toward the use of other data sources to expand the new economic history beyond the colonial period and the use of the colonial records.