Estimating the Profitability of Colonial Ventures

There are competing theories claiming that the private return on foreign investment in general, and colonial investment in particular, ought to have been higher than the return on investment domestically (in the United Kingdom or in a wider group of core European countries). These include the classic theory that globalization opened up new, and potentially more profitable, investment opportunities in virgin areas of economic development (e.g. due to rich natural resources) and the neo-Marxist theory of super-profits from colonial exploitation. There are also theories suggesting that the private return on foreign investments might not need to be particularly high. These include a prediction of convergence of rates of return over time, different cultural values by different (‘gentlemanly’) investors and the fact that imperialism might reduce country risks (which would lead to a lower risk premium). It is important to note that these potential explanatory factors are not all mutually exclusive. They might reinforce or counteract each other. It is therefore by no means theoretically clear how high the return on foreign investments, and colonial investments in particular, would be compared to the domestic return on investment. The issue can therefore only be settled by empirical analysis.

What was then the actual historical rate of return on overseas investments compared to comparable investments in the United Kingdom? This issue has been debated in much previous research. In many monographs on particular companies, one can easily find information about how supposedly successful some of the more well-known companies operating in various colonies were. In the case of Africa, this includes companies such as the Suez Canal Company, the British South Africa Company, DeBeers Consolidated, Ashanti Goldfields or the Anglo-American Corporation (Chilvers 1939; Katzenellenbogen 1973; Hopkins 1976; Hansen and Tourk 1978; Cunningham 1981; Innes 1983; Afrifa-Taylor 2006; Tignor 2007; Newbury 2009; Bonin 2010). While these studies might shed light on the history of the individual companies, they are hardly enlightening on the issue of the general return on investment since there is a problematic selection bias of business history monographs. Those companies that have been the focus of such research were rarely chosen at random among all companies, but have been chosen particularly because they happened to be successful. For every successful company, one can also find failures. J. Forbes Munro has, for example, argued that many companies that invested in plantations in Africa failed economically, for various reasons (Munro 1981, 1983, 1984, 28–29; see also Mollan 2009).

There have been attempts to analyse the general return on foreign versus domestic investment, or the return on investment in particular colonies. One of the first and most influential studies, undertaken by Herbert Frankel, studied the return on investment in the South African mining industry from 1887 to 1965, comparing these to the return on investment in the United Kingdom around the same time (Frankel 1935, 1967). Frankel’s original study was based on a sample of 576 South African mining companies, but his later study on the return on investment was limited to a smaller subsample of more ‘important’ South African companies. The study collected data from the South African Stock Exchange for an analysis of what he calls the internal rate of return. The method is not described transparently but seemingly includes both net capital gains and dividends paid out, thus amounting to estimates of the total return on the investments, using annual data. The author does furthermore not report explicitly how the data from the individual companies was weighted in order to arrive at an average rate of return, but the estimates seem to have been weighted. Frankel’s results from these estimates suggest that the return on investing in South African mining was not particularly high if compared to the return possible from investing in British equity (Frankel 1967, 7–9). These results were later questioned by other scholars. Simon Katzenellenbogen studied the dividends paid out on South African mining equity, arguing that these dividends were extremely high on average (Katzenellenbogen 1975). Since Katzenellenbogen did not include capital gains or losses in his estimates, the figures are not comparable to Frankel’s estimates. Robert Kubicek too criticized Frankel’s estimates, arguing that the return on investing in South African mining actually was higher than the return on other comparable investments, but the empirical evidence in favour of this argument is not presented clearly and methodically (Kubicek 1979).

Michael Edelstein contributed with an often-cited study comparing the rate of return on British foreign and domestic investments. The study estimated the total rate of return, including both net capital gains and dividends paid out, for a sample of 566 ‘high-class’ securities. The study covered the period from 1850 to 1914 using data primarily from The Investors’Monthly Manual (a publication from The Economist started in 1864). Mining companies were explicitly excluded from the study—thereby excluding virtually all investments in colonial Africa—and only one single company (a bank) operating in Africa was included in the sample. Unweighted sectoral average total return on investment was calculated, and an aggregate estimate of the total return on investment was arrived at by weighting the sectoral averages by their shares of national income. Edelstein’s results show that the total rate of return was systematically higher on foreign investments than on domestic investments, even when adjusting for risk (Edelstein 1970, 1976, 1982). Important for the purpose of the present book, Edelstein did not distinguish between colonies and non-colonies in his aggregate estimates but tried to deal with this issue by separate case studies on investments in Canada, Australia and the United States.

The most often-cited studies on this topic have been the research undertaken by Lance Davis and RobertHuttenback (Davis and Huttenback 1982, 1986, 1988). In their research, the authors studied the return on British investments in colonies in particular. They compared this rate of return to the return on domestic investment, during the period 1860–1912. The authors make no distinction between different parts of the British Empire. The study is based on three different samples of companies merged together: one sample of 241 companies selected at random among companies traded on the London Stock Exchange, one non-random sample of 234 companies selected ‘because some portion of their original records exist and are available for analysis’, and one sample of seven British railway companies selected on unclear grounds and included ‘to gain a modicum of regional representation’ (Davis and Huttenback 1986, 81–82). Davis and Huttenback used data from the internal accounts of these companies to estimate the companies’ rate of reported profits relative to physical assets (explicitly refraining from estimating the total rate of return on the investments) and calculated an unweighted average rate of return for the merged sample. The conclusion they arrived at is that the return on investing in colonies was somewhat higher than the return on investing in Britain during the early part of their study, from the 1860s to 1880s. After this period, the rate of return from colonial investment was similar to (or even lower than) the rate of return on domestic investment (Davis and Huttenback 1986, table 3.15, 1988, figs. 3.3–3.5). Because of the different selection criteria, there was a high risk of bias in the selection. Some reviewers have questioned the representativity of the study since virtually all of the well-known (and hence potentially most profitable) colonial companies were missing from the samples (Hopkins 1988; Porter 1988).

Peter Svedberg studied the profitability of British investments overseas during the period 1938–1974 using data from the Bank of England on capital stocks and income from affiliates or branches of British-controlled companies. Comparing the profits from companies operating in British colonies to those operating in non-colonized Least Developed Countries during this period, Svedberg found that the colonial branches or affiliates were much more profitable for the controlling company (Svedberg 1982).

In more recent years, there has been much more work on the return on investment in various countries using updated methodologies and more representative samples of data. Many of these studies have studied developed countries in particular (e.g. Goetzmann et al. 2001; Dimson et al. 2002; Grossman 2002; Eitrheim et al. 2004; Acheson et al. 2009; Esteves 2011; Waldenström 2014; Jordà et al. 2019).

There is some recent research studying the return on overseas investments in less developed countries (i.e. many former colonies). In 1999, Philippe Jorion and William Goetzmann published a study of the return on investment in 39 stock markets around the world, including 15 markets in Asia, Africa and Latin America. Data was assembled from various sources including the International Monetary Fund (IMF) and League of Nations’ Statistical Yearbooks, for—at best—the period 1921–1996. For many of the developing markets, the time-series was much shorter, such as in the case of South Africa, where data was reported only for the period 1947–1996. Data on dividends paid out was furthermore missing for several of their series, essentially leading to a substantial downward bias of those estimates (Jorion and Goetzmann 1999).

In 2002, Elroy Dimson, Paul Marsh and Mike Staunton published Triumph of the Optimists: 101 Years of Global Investment Returns (Dimson et al. 2002). In the book, the authors created estimates for the total rate of return on investment in a number of countries during a whole century, from 1900 to 2000. Despite the subtitle, the data they present is not global in coverage. The data is reported by nation, but only one country in Africa (South Africa) and no Latin American or Asian countries were initially included in the analysis.Footnote 1 The data collection procedure varied between the countries and the time periods studied. For example, the data for the United Kingdom is based on a sample of 247 companies for the period 1900–1954 and of 604 companies for the period thereafter. The sample for the first period is made up of all companies present on the FTSE-100 list for six benchmark years (Dimson et al. 2002, 299–301). Since this list would include only the most traded and potentially most successful companies, it might introduce a considerable selection bias to these estimates. The dataset for South Africa is another example which is of interest as a point of comparison to this study. The dataset was spliced together from two different sources of data, based on a seemingly arbitrary weighting. This data might therefore also suffer from selection bias (Rönnbäck and Broberg 2018). The results that the authors arrive at suggest that the return on investment in a country such as South Africa was somewhat higher in both nominal and real terms than the return on investment in the United Kingdom during their period of study (Dimson et al. 2002, figs. 4–5).

Frans Buelens and Stefaan Marysse have published a study of the total return on investment in 158 companies operating in Belgian Congo, using the same basic methodology as Dimson et al. (2002), and compared the figures to total return on investment in Belgium during the same period (Buelens and Marysse 2009; see also Annaert et al. 2012). Their results indicate that return on investment in Belgian Congo was higher than the domestic return on investment in Belgium. This was true throughout the period except for the 1950s, the last decade of colonialism in Belgian Congo. If the 1950s are included in the estimate, the average return on investment in Belgian Congo decreases to low levels. The realization of country risk can thus have substantial impact even on the long-term average return on investment. In another study, Frans Buelens and Ewout Frankema studied return on investment in plantation agriculture in the Netherlands Indies using the same methodology. Return on investment there was in general somewhat higher than return on domestic investment, but investing in the colony was also somewhat riskier (Buelens and Frankema 2015).

Benjamin Chabot and Christopher Kurz published a study of the ‘foreign bias’ in English investment during the period 1886–1907 using monthly data for all securities reported in publications such as the Money Market Review and the Investors’Monthly Manual as listed on the London and US Stock Exchanges. These publications did not always manage to include all companies listed on the respective stock exchange but increasingly excluded smaller and less traded companies (Hannah 2018). All leading companies listed on the Stock Exchanges would have been included in the publications. Their data shows that high returns were not necessarily the only (or even the main) benefit of investing overseas. The real benefit, they argue, was instead that the diversification of portfolios from holding overseas assets increased the investors’ utility (Chabot and Kurz 2010).

The most recent study of the historical return on investment has been published by Richard Grossman, studying the total return on British investments both domestically and overseas in the period 1869–1928 using data on all companies reported in the Investors’Monthly Manual published by The Economist (Grossman 2015, see also 2017). In contrast to Dimson et al., Grossman’s data covered British investments in countries all over the globe, but a drawback is that the data was reported only by continent (with the exception of data for the United Kingdom in particular) and was based only on annual data. Grossman has received criticism for the data he had assembled (Hannah 2018), to which he responded by concluding that there seemingly are substantial problems with the estimated results, so that they might require revision (Grossman 2018).

Limitations of Previous Research in the Field

Table 4.1 summarizes the main previous research in the field.

Table 4.1 Summary of key previous research
Full size table

There has been substantial amount of research conducted in this field, but there are several limitations to this previous research. First, there are chronological delimitations. The older studies of the profitability of imperialism were all limited to studying the period prior to the outbreak of the First World War, and they are therefore not able to study what happened to the return on investment in the colonies once the colonial economies started to mature in the interwar period. Chabot and Kurz’s more recent study is similarly delimited to a short period prior to the First World War. Richard Grossman’s study covers a longer time period, including the years until 1928, in his study. Any development after this time is not included in the analysis. But, as shown by Dimson et al. (2002), in order to handle exogenous events and market volatility, the length of the data series is crucial when analysing the average return on investment. This issue becomes even more acute for the African context, where risk and volatility historically have been high. The case of Belgian Congo is illustrative, where the choice of limiting the data series to 1955 (rather than to 1960, when the country risk was realized under the process of independence) doubles the long-term average return on investment (Buelens and Marysse 2009, 152–53). Dimson et al., on the other hand, only started their study in 1900. For the purpose of our study, they thereby miss the important years of the ‘Scramble for Africa’, at the end of the nineteenth century. Apart from Buelens and Marysse’s study of Belgian Congo, there is no previous research trying to study the return on investment in Africa covering the whole of the colonial period.

Second, there are geographical limitations. While some of the older studies of the profitability of imperialism had a clear focus upon studying the return on investment in colonies, none of them made any distinction between different types of colonies or their different geographical locations and treated all colonies and dominions similarly. Due to the way their samples were constructed, they furthermore missed out key African companies. Buelens and Marysse’s study was, in contrast, limited to companies operating in one specific colony, Belgian Congo. Dimson et al. claimed to estimate return on global investments, but the only African country included in their dataset was South Africa. Belgian Congo and South Africa need not be representative of other African countries. One would expect that the historical return on investment could differ substantially between different colonies over time, as the economies in the colonies developed along different trajectories.

Third, there are issues of sample size and selection bias. As was noted in the previous chapter, older studies on the profitability of imperialism all suffer from a substantial selection bias. Edelstein totally excluded the main sector of investments in colonial Africa (mining). Davis and Huttenback created a sample based on three different (and not unproblematic) methods of selection. Dimson et al. based their study mainly on a limited number of larger companies in the mining and financial sectors, creating a potential survival selection bias. Grossman’s sample of companies operating in Africa is seemingly larger than that of Dimson et al.’s, but it is still comparatively limited (further limited by only using annual data).

The present study will, in contrast to the previous research, study a whole century, from 1869 to 1969. We can thereby cover much of the process of colonization and decolonization in several of the countries studied. Our study will furthermore be based on a larger sample of companies compared to previous research on investments in Africa. Though our sample is not completely free from selection bias, we believe it suffers considerably less from such bias than previous research—an issue we will discuss further in the following chapter. We also employ data of a higher frequency using monthly data, allowing us to conduct a more detailed and robust analysis of the rate of return on investment. Finally, the present study will have a broader geographical coverage, studying British investments in several different parts of Africa. Because of the overall nature of the flow of these investments, the sample is biased towards South Africa and Egypt, but it also contains a substantial number of companies operating in current-day Ghana, Nigeria, Zambia and Zimbabwe as well as a smaller number of companies operating in 14 other colonies/countries in Africa.