Why social tables remain the coolest tool for historical inequality studies

Studying economic inequality in contemporary settings remains a difficult thing to do. People are typically not too keen on letting you know how much they earn or how rich they are. We often under-report our earnings for fear of rebuke from tax authorities or because it’s a social taboo to speak about money and incomes. So modern economists are left with survey data that may either underestimate or overestimate inequality metrics. But despite these challenges the overall picture we get from many of these studies is more or less accurate. Now imagine looking at long-term inequality where the data situation is so much worse. Researchers busing themselves with historical inequality studies really have their work cut out for them considering that their data is often both difficult to retrieve and of questionable quality.

Therefore, writing a blog post about a meta-study on social table use in economic history research is challenging because one is torn between a feeling of genuine commitment to the rigour of academic discourse, on the one hand, and the show of respect to academic peers that often have spent years constructing these magical data grouping tools, on the other hand. Nevertheless, together with Dieter von Fintel (Stellenbosch University) and Erik Green (Lund University), I have worked on investigating the empirical soundness of the social tables approach for calculating historical income inequality.

We believe that data grouping in inequality studies have been useful in measuring income or wealth inequality for populations where information is scarce. In contemporary economics it is often deployed to comment on incomes for countries where large informal sectors are present, and data is scarce. This is exactly why it is so appealing for the historical settings, where micro-level income or wealth information is limited at best. Social tables essentially group a population into classes, to which a well-researched mean income is assigned and a between group income/wealth inequality number can be calculated. Based on this inference, we looked at a large sample of historical social tables used in the global economic inequality discourse fired up by Branco Milanovic in the early-2000s – with some inspiration from twentieth century Kuznets.

Before I venture into the mechanics of how our study is set up, it is important to say that our concern began with skepticism related to low inequality estimates for large populations where classes were few. Since Modalsli (2015) had beaten us to the punch with commenting on overlapping incomes between various classes, we decidedly shifted our focus. But, putting together this database of inequality studies would prove to be daunting, and while we are now finalising our paper, our efforts stretch as far back as 2018. I think part of the difficulty in collecting the data may have been due to our fellow economic historian’s initial skepticism to us writing a commission of inquiry type paper.

As good investigators do for an inquiry, we set out to build a comprehensive database for social tables that was current and detailed – only then could we really say anything meaningful. Some detective work and wonderful co-operation from UC Davis’ Global Price and Income History group has enabled us to construct a dataset spanning more than a thousand years and including 108 individual social tables. In a similar fashion to Milanovic, we survey these social tables which were set-up in meticulous ways by scholars from all over the world for places like England and Wales in 1688, China in 1880, France in 1788 and the Kingdom of Naples in 1811. Beyond looking at mostly Europe, North America and Asia we also expand our dataset to improve the representativity of Africa and South America. While Milanovic did include two African countries and three South American nations in his 2018 paper, we have upped the ante as we also included studies by AFLIT researchers for countries like Botswana and Ghana that have inequality observations for more than one year.  Our dataset also contains information on the specific country, gdp per capita, gini-coefficients, population shares attributed to the various classes, and the years for which calculations were done. By incorporating more information for countries of the Global South we get a fuller picture of the computing power of social tables and extend Milanovic’s 2018 analysis of just 41 social tables.

So, what exactly were the findings of our inquiry? Initially, we expected that precisely because many countries in Asia, such as Bihar India or China, had so few classes, their gini-coefficients had been under-estimated. Also, we thought that deliberate research interventions to create more variation in the data by creating more classes would bias the calculated gini coefficients.  Our results presented in Figure 1 shows that the association between social table attributes such as the number of classes and the percentage of the population in the top class matters for the eventual gini coefficients calculated. This is especially evident if you look at earlier meta-studies on social tables (Milanovic 2011) versus our own database with ‘all’ available social tables. At a first glance this may have confirmed our expectations, but it turns out that the number of classes is not the most significant determinant of bias in the gini coefficients obtained from social tables. Instead, our multivariate regression analysis shows that one must be cautious of large bottom classes in small populations, meaning countries with populations smaller than two million. This often leads to the bottom class pulling up the eventual inequality estimate. However, we infer that this bias in existing gini coefficient estimates does not come from researcher interventions but rather stems from a lack of data availability.

Figure 1: Social tables dataset comparison

On the upside, we believe that the piece-de-resistance among social tables studies has been the work done by researchers in the AFLIT network. Hence the relevance of this blog post. We generally find that their social tables for Africa tend to be of better quality and more rigorous in their methodology as they show much more attention to issues like sufficient variation in the data and present observations for more years. This is an improvement compared to studies that often only give one point estimate of a gini for a specific year. They are also much easier to compare across time and space.

In the end, I think the way forward for scholars of long-term inequality trends is pretty straight forward and intuitive, although I have three recommendations. When building social tables, make sure you have enough information that accurately reflects the structure of an economy and gives sufficient variation in the number of classes. Thereby you don’t end up having a problematic bottom class. Another thing may be to introduce a range estimate for incomes in your specific class as opposed to a restrictive point estimate. This could help in dealing with the overlapping classes problem. Third, try to have more than one estimate for one year in any study. Having a time trend often allows one to sense check your findings. But ultimately, the social tables approach has performed reasonably well in giving us a sense of historical inequality.

References

Milanovic, B. (2011). A short history of global inequality: The past two centuries. Explorations in Economic History, 48(4), 494-506.

Milanovic, B. (2018). Towards an explanation of inequality in premodern societies: The role of colonies, urbanization, and high population density. The Economic History Review, 71(4), 1029-1047.

Modalsli, J. (2015). Inequality in the very long run: Inferring inequality from data on social groups. The Journal of Economic Inequality, 13(2), 225–247

Collecting Archival Material: a Conversation with Historian Karin Pallaver

Maria Mwaipopo Fibæk talks to Karin Pallaver, a member of AFLIT and an Associate Professor of African History in the Department of History and Cultures at the University of Bologna. Karin’s research draws on a multitude of historical sources of which many were collected at national archives in United Kingdom and East Africa and the two discuss the collection of archival material for research.

Karin has a longstanding interest in African history and she has been part of a ‘Money in Africa’ project based at the British Museum. She has published work on the long history of changing currency regimes and money circulation in East Africa. Apart from contributing to a monetary history of Africa, her work has also shed light on critical topics such as pre-colonial commercial expansion, wealth and power imbalances, the use of money by the ‘ordinary’ African citizen, and the transitions to colonial currencies where African agency played a critical role in how the colonial currencies were introduced. For AFLIT, Karin is working on producing social tables for the two island economies Mauritius and Zanzibar.

Q: Karin for your research you have collected colonial material from the National Archives in the United Kingdom yet you have also been a frequent user of the national archives in East Africa. Can you tell us about your experience from collecting archival material in these places? For instance, are there are any differences in the material that is available in the United Kingdom compared to what can be found at the national archives of the former colonies?

A: Contrary to common assumptions that tend to represent archives in African countries as poorly catalogued and difficult to access, both the Kenya National Archives (KNA) and the Zanzibar National Archives (ZNA) are well organised, with detailed paper catalogues and, in the case of the KNA, a computer database. Clearly, you would need a few days to understand how the archives are organized, where the information you are looking for is located, and so on. However this is, I believe, valid for all archives, not only those in Africa.

The perspective you can get from these documents is rather different from that obtainable from material collected at The National Archives in the United Kingdom. For example, in Kenya local district and province administrators sent regular reports to the central administration of the colony – first located in Mombasa and later in Nairobi. Parts of these reports were then used to write general annual reports on the colony, which were then sent to London. Therefore, documents collected at KNA usually give much more in depth information on what was going on in the colony compared to the material available at the National Archives in London, that are summaries of the local reports.

Zanzibar was in the early colonial period the most important centre of the British administration in East Africa. For this reason, in the late 19th-early 20th century, reports from the East Africa Protectorate and Uganda were all sent to Zanzibar. Thus, the material in the ZNA allows researchers to get more precise information on what was going on in the region compared to the reports to the Foreign Office that were sent to London (and are today housed in The National Archives in the United Kingdom).

Q: Could you take us back to some of your early visits to the national archives, how would you usually prepare for a trip and how long would you stay at a time?

A: My first visit to an archive in East Africa was in 2005, when I did research in the Tanzania National Archives (TNA) in Dar es Salaam for my PhD thesis. To access the TNA (and in general for doing research in Tanzania) you need a special research permit from the Tanzanian government called a ‘COSTECH Research Clearance’ as well as a special one-year visa for researchers. As it might take some time to get the COSTECH Research Clearance (on their website they advise that it can take several months), I had to prepare my visit and relevant documentation well in advance. Usually, you need to find a host institution such as a university department or research institute and their details would need to be included in the COSTECH application form.

For doing research in the ZNA, at the time there was no need to apply for the COSTECH research clearance needed on mainland. At the time, however, you still needed to submit a research application to a local office and that process would take between two and three weeks to be approved. You also had to pay a fee that depends on the amount of time you intended to spend in the archives.

Doing research in the KNA is much more straightforward. You can just go to the archive reading room, bring a passport size picture and a reference of an academic sponsor based in Kenya, and then pay a small fee to become a member.

Before my first visits to these archives, I read the paper catalogues that are available in some libraries in Europe, for instance at SOAS Library in London. For example, a catalogue of the “German Records” at the TNA was compiled in 1973 and one of the KNA in 1968. Unfortunately, no comprehensive online catalogues are available, with the exception of the catalogues of some microfilmed material that is collected at some American universities, such as Syracuse University. Moreover, there are various articles written by scholars who report their experience with doing research in these archives, that I always found very helpful.

The length of stay depends on the research you need to do and the availability of material. But also on all your other commitments. For instance, it was easier for me to spend longer periods in the archives during my PhD when I did not have the administrative or teaching duties that I have now! But at the same time, over the years I have built a database with pictures of many documents from various archives, that I do integrate with new data and sources that I can collect during visits of three to four weeks.

Q: In your work, you often integrate various historical sources. For instance in your paper published in 2019 on the East African rupee crisis 1919-1923, you use an impressive variety of colonial sources such as formal colonial documents including districts’ annual reports, correspondence between colonial officers, and minutes from meetings. This material you integrate with historical newspapers and missionary sources. Could you tell us more about your efforts to integrate various historical sources? How do you find the material?

A: As all Africanists know, one common problem for reconstructing African history is the lack of sources. For this reason, I have been taught by my PhD supervisor to consult and use as many different archives as possible. For example, European colonial sources were framed by their writers’ assumptions about African societies and economic practices, but nonetheless they are often the most substantial source of evidence available.

For my study of the history of currency, missionary sources have revealed themselves as extremely valuable, as missionaries meticulously recorded the wages paid to their labourers, church offerings and prices on the local markets, as well as bridewealth payments in commodity currencies and in coins. Newspapers written during the colonial period by some groups and associations (for example the newspapers of the Indian community in Kenya (that I have used for the mentioned article on the East African rupee crisis) provide a different perspective from that offered by the European administrative sources on the events taking place in the colony. So my research practice is generally to scrutinize different possible sources and then carefully conduct archival work in as many archives as possible. Sources can then be compared to each other to provide different data as well as perspectives on the topics and events I do research on.

Q: Among the contemporary African Economic Historians, you are one of few researchers who have published work on the African precolonial era. To give an example, a paper of yours traces the use of currency transitions in Buganda (later known as Uganda) as far back as the 18th century. The reason why many scholars within African Economic History steer away from the pre-colonial era is probably the lack of archival material from the era. How and where did you find material from the period?

A: Sources on the precolonial period are not easy to find, but still some of them are there, and include early administrative records, missionary sources, travel accounts, commercial companies’ records, ship logs, personal papers and early ethnographies. The information is scattered across different types of archives and institutions, but when it is put together, it allows you to build a critical mass of sources. In my case, this has allowed me to get enough data to reconstruct the history of currency in precolonial East Africa, or at least a part of it. To this, we can add that interdisciplinary research has always been the matrix of the field of African history, especially when studying the precolonial past. Research in archaeology and historical linguistics has for a long time laid the foundations upon which scholars have studied later periods.

Scholars of Africa’s precolonial past, in particular, have a long history of interdisciplinary collaboration beginning with the birth of the field in the 1960s, when their intense collaboration combined archaeological, historical and linguistic sources to reconstruct an original and independent precolonial past, that was ‘usable’ to fill the artificial colonial borders with a national history the roots of which extended well beyond colonial rule. Interdisciplinary sources not only add information, but they actually complement and enhance each other while revealing their respective contradictions and inconsistencies.

Q: Apart from national archives, several other institutions house historical material. For instance for my research on post-independence Kenya, I have used the libraries of the Kenya National Bureau of Statistics and the London School of Economics. Moreover, during a stay in Oxford, I found the library at the Oxford Centre for Economic and Social History very useful and I know many historians also use the Rhodes House Library in Oxford. Do you have any experience collecting material from other places than national archives?

A: We have already mentioned missionary archives. I did research in the White Fathers Archive in Rome, in the Church Missionary Archive in Birmingham, in the Mill Hill Missionary Archive in Freshfield, Liverpool and in the London Missionary Archive in London. For the study of the colonial period in Kenya, I found the sources collected at the Bodleian Library in Oxford very useful. The Library holds many personal papers of British settlers, their associations, and British administrators in Kenya. For example, for my paper on the East African rupee crisis, I located the names of the administrators who were critical in the decisions and discussions of the time. This I could do thanks to their signatures in the official correspondence in the national archives. I also looked for their personal papers in the Bodleian library. Often the “official” perspective is quite different from the “personal” one.

Finally, doing research on the history of money, I worked in numismatic collections, especially in the Coins and Medals Department of the British Museum. The study of the materiality of coin objects (their material, form and design) is extremely useful in understanding why certain coins were accepted by colonial societies whereas some other were not; but also to provide information on through which symbols the colonial state wanted to represent itself on coins and banknotes.

Q: Finally, can you tell us a little bit about your upcoming work for AFLIT. I know that you and Ellen Hillbom are working on producing social tables for Mauritius and Zanzibar. How is the work coming along and what type of historical material/sources will you draw on?

A: During a talk at the World Economic History Conference in Boston, some years ago, Ellen and I started to discuss the possibility of comparing the history of inequality in Mauritius and Zanzibar. The idea started from the fact that the two places share some significant similarities. Differently from the large land masses of many of the mainland colonies, they were smaller islands in the Western Indian Ocean, and because of their location, they had long histories of being hubs for valuable international trade, for example in spices. In parallel with trade, they both developed into plantation economies (in the case of Zanzibar cloves, in the case of Mauritius sugar cane). They were both occupied by the British in the 19th century and remained under British control until the 1960s and they were both the destination of substantial flows of migrant labourers. Nevertheless, they also had important differences, for example, Zanzibar employed enslaved labourers in the 19th century, and migrant labourers during the British colonial period, whereas the labourers in Mauritius were for the majority indentured labourers from India.

We feel that the history of these small islands is rarely told within African Economic History compared to more well known countries such as South Africa, Nigeria, Ghana, or Kenya. We want to add these country cases to the history of the region. What were the consequences of their similarities and differences on inequality over the long term? Can we identify patterns that they shared with mainland colonies? We have almost completed the collection of sources such as censuses and various statistical reports in the archives in Zanzibar and Mauritius. We will soon start digitising the data and discussing on how to best frame this case study.

Notes to the blog:

Notes to the national archives mentioned are posted below.

The Kenya National Archives are located on Moi Avenue in Nairobi, Kenya. The opening hours are 8 am to 5 pm. See here, for more information on availability, categorisation, and condition of material available at the archives. At the time of writing, no research permit was required to access the archives.

The Tanzania National Archives are located on Vijibweni Street in Dar es Salaam, Tanzania. The opening hours are 8.30 am to 6 pm. See here, for more information on availability, categorisation, and condition of material available at the archives. At the time of writing a research clearance from COSTECH is needed to access the archives.

The Zanzibar National Archives are located on Zanzibar in Kilimani, about 1.5km from the centre of Stone Town in Tanzania. The opening hours are 8.30 am to 6 pm. See here, for more information on availability, categorisation, and condition of material available at the archives. As of March 2018, researchers must submit two separate applications to conduct research in Zanzibar: one to the Research Committee in the Office of the Second Vice President and another to the Office of the Chief Statistician. The first application must be approved before the second can be submitted. However, there is only one research permit fee.

Land, Labour, Legacies: Long-term Trends in Inequality and Living Standards in Tanzania, c. 1920-2020

In my dissertation, I seek to investigate the roots of poverty and economic inequality in Tanzania and analyse how growth, inequality, and poverty interact in the country in four interrelated papers. Throughout the dissertation, the colonial period takes a prominent role, since a better understanding of the trends in inequality and living standards under colonial rule are crucial to tackle the overarching themes.

To estimate inequality and living standards in the colonial era, social tables provided an invaluable tool. As in most sub-Saharan African colonies, the absence of comprehensive income tax records and household budget surveys for that period make the use of contemporary estimation techniques nigh impossible. Here, social tables provide a solution, as they allow us to estimate inequality using grouped income data, which is much more readily available.

The social tables method provided the starting point for the first paper, which analyses income inequality trends in the wage sector under British rule. The overall level of income inequality was relatively high, in line with levels seen in settler colonies like Kenya, and mostly driven by racial income differences between Europeans, Asians, and Africans (see Figure 1). The root cause of these income differences were of an institutional nature: racial discrimination and a lack of educational provision by the colonial authorities. These barriers also impeded African advancement more generally. While they were slowly being dismantled after World War II, they had a long-term negative impact on Tanzanian economic development.

The second paper investigates overall rural living standards in Tanzania under British rule as well as inequality dynamics amongst African cash croppers. We have seen in other cases, such as Ghana or Uganda, that cash crop production and an increasing commercialisation of agriculture led to both increasing living standards and inequality amongst African farmers. Tanzania was no different, although regional-level social tables show that levels and trends in income inequality varied depending on the underlying social and institutional structures in different regions. Higher levels of inequality did, however, not always preclude an increasing participation of smaller growers in the coffee economy or an overall increase in welfare for the entire population – in fact, it was often the areas without notable cash crop production that saw the least improvements in general welfare.

The third paper, co-authored with Maria Fibaek and Erik Green, widens the geographic scope and provides a comparison of rural real wages and their determinants in Tanzania, Kenya, and Malawi from 1920 to 1960. We investigate whether the presence of more settlers led to lower wages, as is often argued in the literature. We find that, while more settlers meant the implementation of more coercive labour policies, the continued expansion of settler agriculture in Kenya and Malawi led to a growing demand for labour and a long-term increase in rural real wages. In Tanzania, in contrast, the lack of colonial policy support for settlers resulted in stagnant employment numbers and stagnant real wages throughout the colonial period.

The final paper, co-authored with Morten Jerven, moves away from the colonial period and investigates the interrelations between growth, inequality, and poverty (G-I-P) from independence in 1961 to the present day. We observe two distinct growth periods in Tanzania, from 1961 to the late 1970s, and from the mid-1990s to the present. Both led to a decline in poverty, but the gains from the first growth spurt were all but reversed during the economic collapse of Tanzania beginning in the late 1970s. We also find that the inequality-growth relationship is not straightforward: policy regimes and development strategies played a crucial role in determining its precise nature. For example, the policy regime of African Socialism in the 1960s and 1970s led to a decline in income inequality both during the growth and economic decline, while the recent growth episode led to increasing inequality.

Overall, I find a complex relationship between growth, inequality, and living standards in Tanzania. Levels of economic inequality in Tanzania generally appear low, but this is primarily a consequence of a low level of economic development. The macro perspective also obscures high degrees of economic inequality in different parts of the economy like the wage sector and cash cropping. Attempts to increase the speed of development and/or reduce the level of inequality were rarely successful, as they often failed to overcome the fundamental cause of inequality and persistently low living standards: the lack of economic opportunities.
Since the turn of the millennium, however, there has been a change in the pattern. The available evidence suggests that, finally, economic opportunities have expanded significantly across the board, and that this expansion of opportunities is one of the main drivers behind the latest growth spurt and significant reductions in poverty.