THE LAND-LABOUR HYPOTHESIS

While traditional frontier literature finds a positive correlation between land size and demand for children, the rich literature on population and agrarian change in pre-industrial societies gives reasons to believe that it is equally plausible that the relationship was negative. A cornerstone of this body of literature is Boserup’s seminal work on population pressure and changes in agricultural systems. Where frontier literature explains how land availability affects fertility, Boserup develops a model to explain the effect of increased population pressure (caused by increased fertility or migration) on farming practices. Boserup’s work is often cited as a response to Malthus’s pessimistic account of the effect of population growth on output and rural welfare.

Malthus argued that increased population pressure would lead to a decline in production per capita and consequently – in the long run – a reduction in population size.Footnote ⁷ The reason is that without so-called positive or preventive checks, the population would increase geometrically while food production would increase arithmetically. An increase in per capita income above an equilibrium level of consumption leads to population growth. Since land is assumed to be a fixed factor of production, continued increase in population implies that marginal lands will have to be opened for cultivation. This leads to a slowdown in agricultural growth and hence diminishing output per capita. Mortality levels would increase until the per capita consumption falls back to its original equilibrium. This is often referred to as the Malthusian trap or low-level equilibrium.Footnote ⁸

Boserup challenged the Malthusian relationship between population and agricultural output. In short, Boserup claims that the rural population would respond to increased population pressure by adopting more intensive farming practices and increasing the frequency of cropping. In this way, output per capita could be kept intact or even increase with population growth. Land-use intensification, by way of shorter periods of fallow, for example, leads to increased yields per unit of land, but also requires more labour and/or capital. In the pre-mechanization period, land-intensification thus came at the cost of declining labour productivity. In contrast with the land-labour hypothesis, under these conditions, farmers can react to shrinking land availability by having more children, who can be put to work in order to maintain output per capita. We expect to find such a relationship only in cases where farmers mainly rely on the use of family labour, as farmers will employ this form of labour even when the marginal productivity is decreasing.Footnote ⁹

But why would farmers increase labour input instead of moving towards more capital intensive farming systems? One of the most influential theories of technical change (change in farming practices and/or technologies) is Hayami and Ruttan’s model of induced innovation.Footnote ¹⁰ They treat technical change as endogenous. In brief, they argue that as agriculture develops over time, particular resources become scarce, which affects the relative factor prices of inputs. In cases of land abundance and labour scarcity, farmers would seek innovations that increase land productivity, i.e. yield-increasing technologies. In this regard, their model is similar to a Boserupian interpretation of agrarian change. But while Boserup focuses on the land–population relationship, induced-innovation focuses more broadly on all factors of production, including both embodied and disembodied technologies.

Critics have pointed out that Hayami and Ruttan’s model depends on the existence of private or public organizations that respond to the changing factor prices by developing new innovations and making them available for farmers.Footnote ¹¹ In our case, such organizations were clearly missing. The best way to access new technologies was either to produce them on the farm, or travel to the market – a journey that could take several weeks. Equally important, and as pointed out in the extensive research on agricultural development in currently developing regions, farmers must have the means to purchase the innovations.Footnote ¹² In a case of fragmented markets and limited purchasing power, farmers may not respond to changing relative factor prices in the way that the model presupposes. More precisely, farmers may respond differently to changing factor prices according to their wealth and access to markets.

Let us consider an agrarian settler society consisting of two groups: one wealthy group and with the means to develop more capital intensive farms and/or employ additional family or non-family labour, and a poor group that faces capital scarcity and relies solely on family labour. The latter group would respond to shrinking land availability by having more children, who could be put to work. For this group, we expect a negative correlation between land availability and family size. This relationship may be temporary as Boserup suggests, or long-term as was the case in pre-industrial Asia and other parts of Africa.Footnote ¹³ The wealthier group would respond by either employing additional labour and/or developing more capital intensive farming methods. For this group, we expect to find no correlation or a positive correlation between land availability and family size.

This means that the wealth of farmers has both a direct and indirect impact on the relationship between land availability and family size – direct, as the level of wealth affects the means available to respond to shrinking land availability, and indirect as the wealth affects the type of labour employed as farm workers. In a context where the majority of farmers were poor, we expect the closing of an agrarian frontier to be associated with an increase in the number of settler children present in farming households.

Although the literature on farming systems and labour is mainly concerned with crop farming, it also has a bearing on pastoral farming. Intensification is a relative measure and while it is true that crop farming in general is more labour intensive, some pastoral farming systems require more labour input than others. In times of shrinking grazing land availability, pastoral farmers can react in different ways to maintain output and/or income. Wealthier farmers could capital intensify, for example, by investing in more productive, but also more expensive, livestock or invest in transportation to increase market access. Farmers that lack the means to take such measures could invest in less productive and less land-demanding but more labour intensive livestock. Even in pastoral communities we therefore expect that the wealth of farmers would determine whether the correlation between land availability and fertility is negative or positive. We explore this relationship further in the context of the closing Graaff-Reinet frontier.

THE GRAAFF-REINET FRONTIER

The Cape was initially settled by the Dutch East India Company (VOC) as a refreshment station for passing ships. Gradually, the number of European inhabitants increased, and by the early eighteenth century, the Cape had established itself as a settler colony, supplying Cape Town and passing ships primarily with wheat, wine, fresh produce, and meat. Most European farmers settled in the southwestern part of the Cape, but by the mid- eighteenth-century, expansion into the eastern and northern interior was well underway (see Figure 1).Footnote ¹⁴ During this time, the European population at the Cape was doubling approximately every generation.Footnote ¹⁵ The district of Graaff-Reinet, on the eastern frontier of the Cape Colony, was established in 1786, accounting for around a third of the total extent of the colony and covering an area of approximately 50,000 square miles.Footnote ¹⁶ There, agro-climatic conditions favoured pastoralism, and stock farming quickly became the dominant economic activity. As early as 1770, two thirds of inland farmers were estimated to have been engaged in livestock pursuits rather than as cultivators.Footnote ¹⁷ Unlike the frontier expansion in North America, the expansion eastward in the Cape saw movements of people followed gradually by commercial development.Footnote ¹⁸

Sons of the poorer arable farmers of the southwestern Cape likely viewed stock farming on the interior as an attractive means of survival, since entry requirements were low and the product could walk to market. During its formative years, it was possible for a settler with little to no capital to move into the Graaff-Reinet district and set up a farm. In the late 1770s, a local farmer’s wife remarked to the Swedish traveller Sparrman,

Figure 1 The Graaff-Reinet district expansion from 1795 to 1825.

Some doubt has been cast on whether it truly was this easy for poor frontiersmen to set up as productive stock farmers;Footnote ²⁰ nevertheless, the steady inflow of new settlers to the district continued into the nineteenth century and the best grazing lands were quickly snatched up. A farm of some 6,000 acres could be acquired under the land grant system with relative ease.

The extent to which frontier households were isolated and independent from trade with the domestic market has been the subject of some debate in the South African historiography. While some paint an image of a wayward, solitary, semi-nomadic, and self-sufficient brand of frontiersman,Footnote ²¹ others maintain that households on the frontier remained relatively orthodox and geared towards production for the market in order to provide for their daily needs.Footnote ²² While scattered evidence of the former can be found, the consensus is that although these households were able to meet some of their basic needs themselves, they were “never entirely cut off from the exchange economy of the south-western Cape”.Footnote ²³ On the inhabitants of the Graaff-Reinet district, it has been remarked that “highlights in the lives of the stock farmers were occasional treks to Cape Town, to pay their recognition fees, to marry, to baptize children, to obtain provisions, to redeem their bills. On such occasions the BoersFootnote ²⁴ took cattle with them to sell in Cape Town, and commodities such as butter and soap”.Footnote ²⁵ Indeed, Guelke acknowledges that the expansion of this group of frontier famers could not have taken place without “guns, gunpowder, wagons and other manufactured items obtainable only in exchange for the produce of the interior”.Footnote ²⁶

While the long distance to major markets in the southwestern Cape limited the degree of commercialization, the sale of cattle and sheep still constituted the main source of cash income.Footnote ²⁷ One contemporary traveller, a German doctor of medicine and philosophy, formerly in the Dutch service at the Cape of Good Hope, Henry Lichtenstein, wrote optimistically on large profits to be had from stock farming in this region. He remarked that sheep in this district were favoured above all others in the colony for their weightiness and flavoursome meat. He estimated that “some farmers [had] flocks to the amount of six or seven thousand”. Cows here reportedly “gave richer milk than elsewhere, and in greater abundance” with butter “carried in great quantities to Cape Town, where it [was] always eagerly bought up”.Footnote ²⁸

Allowing for vegetative and seasonal differences, it has been estimated that one sheep or goat required roughly four acres of grazing land, whereas one horse or head of cattle required some eighty acres – a fact that likely contributed to the prevalence of small stock over cattle holding in the frontier districts.Footnote ²⁹ While sheep farming is relatively non-labour intensive compared to crop farming, this was less true in the nineteenth century. Shepherds played an invaluable role at a time before fencing was a mainstay and amidst the ever-present threat of predation or theft by Khoisan raiders. Daily kraaling Footnote ³⁰ meant that each flock of a few hundred heads required a dedicated shepherd.Footnote ³¹

Where land was to be had on the frontier, ownership was almost exclusively conferred in accordance with the loan farm system, which offered temporary land-use permits in exchange for annual rents.Footnote ³² There is little evidence to suggest that this system created inherent insecurity, since it conferred tenants with exclusive rights to their fixed plot of land. In addition to permits being indefinitely renewable, settlers were free to farm their land without any restrictions or fear of interference from governing authorities. Moreover, applications for licences and permits were seldom refused. Smith notes that “while the Company did not legally surrender its right to take back a loan farm, this was so seldom done that the Boers came to accept that the farms were their own, until they decided to leave; even the failure to pay recognition fees did not result in the revocation of a permit”.Footnote ³³

While the land could not technically be sold, settlers could transfer the improvements, resulting in a system of de facto permanent property rights.Footnote ³⁴ In contrast with the visible “affluence and prosperity” of many farmers in the southwestern Cape,Footnote ³⁵ a salient feature of life on the frontier was the high proportion of virtually impoverished farmers.Footnote ³⁶ That is not to say that a small “landed gentry” did not exist, but the majority of pastoral farmers of the frontier it seems, struggled to make ends meet.Footnote ³⁷

Labour demands on the frontier were always high and in times of acute labour shortage, “all who could be exploited as labourers, were exploited”.Footnote ³⁸ This was true for many colonial economies in Africa,Footnote ³⁹ where the incorporation of native and child labour, both free and unfree, kin and non-kin, was viewed as a necessary and acceptable means of meeting domestic production needs, as well as to supply the ever-expanding European market.Footnote ⁴⁰

There were three main sources of labour available to the frontier farmer: family, slave, and indigenous Khoisan. The importation of slaves to the colony began in 1658, but only flourished at the beginning of the eighteenth century. By 1793, there were 14,747 slaves in the colony compared to 13,830 Europeans.Footnote ⁴¹ Unlike the wine and wheat farmers of the regions closest to Cape Town, frontier stock farmers kept comparatively few slaves and while many carried with them the attitudes of a slave holding community, most could only afford to keep one or two.Footnote ⁴² The average number of slaves per household calculated from the Graaff-Reinet Opgaafrollen for the period 1798–1828 is 1.2 (see Figure 2). This closely matches the traveller John Barrow’s contemporary reckoning that in the district of Graaff-Reinet, in the last decade of the eighteenth century, there were no more than six or seven hundred slaves; approximately one to each family. Moreover, a detailed return for the field cornets office of Op Sneeuwberg (a sub-district of Graaff-Reinet) for 1808, suggests, too, that the distribution of slaves was highly uneven, and that most of the 197 slaves present in that year were concentrated in the hands of the wealthiest inhabitants.Footnote ⁴³ Before 1806, when the British banned the importation of slaves, farmers could have bought newly arrived slaves at the slave auctions in Cape Town. Others may have purchased slaves from other farmers the commercial centre of the southwestern Cape and occasionally non slave-owners may have hired slaves from the large slaveholders, as happened in the southwestern Cape.Footnote ⁴⁴ But Lichtenstein recalls a passage from the journal of a friend, who, during his visit to the Graaff-Reinet home of “one of the most substantial men the country”, described that the situation of the slaves was “truly lamentable, for he [the farmer] only brought such from Cape Town as were to be purchased at a very cheap rate”.Footnote ⁴⁵ Barrow noted, too, that Boers preferred the Khoe as labourers because slaves were too expensive.Footnote ⁴⁶ Indeed, according to the Opgaafrollen, indigenous Khoisan made up the majority of the agricultural labour force on the eastern frontier (see Figure 2).Footnote ⁴⁷ Note that the Khoisan we are able to capture in our data are limited to those working on European farms. The true population was likely far greater, as some Khoisan were employed by the colonial government, while others managed to survive as independent pastoralists.

Figure 2 Mean number of slaves and Khoe employed on settler farms annually, 1798–1828.

Lack of reliable information on the latter prevent us from making even an educated guess regarding the size of the total Khoisan population in the district. Some Khosian will have been residing in the district prior to the arrival of the Europeans, while some may have had originated from the northern and southwestern Cape and accompanied the Europeans as they moved into the interior, working for them as interpreters, carriers, and domestic servants. The Khoisan residing on the European farms consisted of captured runaways (who had joined the Xhosa fighting against further European expansion), apprentices, and “free” men, who, having lost their kraals were left with no other choice than to work for Europeans.Footnote ⁴⁸ The latter group was the most important source of non-family labour in the district, reportedly often living in dismal conditions on their employers’s farms. Traveller John Campell recalled in 1822 that the conditions for some Khoe labourers were often worse than that of the slaves.Footnote ⁴⁹ However, the daily lives and circumstances of these two groups cannot be thought of as distinct from one another. As Mason argues:

We are therefore careful not to treat slaves and Khoisan labour as two separate groups of free and non-free labour, but rather as a combined group, distinct from so-called family-labour.Footnote ⁵¹

THE CLOSING OF THE GRAAFF-REINET FRONTIER

The period under investigation in this paper was characterized by major political and institutional change. Initial European expansion into the eastern frontier was followed by decades of instability and hostility between the Europeans, the indigenous Khoisan, and the Xhosa tribes. The expansion of the European settlement on the eastern frontier made it increasingly difficult for the Khoisan to survive as independent pastoral farmers. As early as 1773, Governor Van Plettenberg wrote that in Graaff-Reinet “[…] there are no Hottentots except those who since several years had hired themselves to the colonists”.Footnote ⁵² This was likely an exaggeration; nonetheless, in 1775, in an attempt to further tighten European control over Khoisan labour, the VOC formally instituted the Inboekstelsel – a pass system that severely inhibited the movement of Khoe who did not carry papers showing that they were employed.

Some Khoisan groups resisted European expansion with success. In 1785, Khoisan had successfully regained control over the land in Sneeuburgen in the north and Camdeboo in the west of Graaf-Reinet, but their success was short-lived.Footnote ⁵³ Instability deepened due to political uncertainty following the decline of the Company and the first British occupation (1795–1802). Countless episodes of violence and conflict littering the final decade of the eighteenth century (indicative of the continued dissatisfaction amongst the Khoe with their former dispossession at the hands of the Boers)Footnote ⁵⁴ together with mounting dissatisfaction with their working conditions, culminated in the so-called Bushman Wars (1799–1802), which saw an alliance between Xhosa chiefs and Khoe rebels temporarily cripple the frontier economy and dispel many Boers from their farms.Footnote ⁵⁵

In the early nineteenth century, the English made further attempts to formalize indigenous subjugation through indenturing Khoe labour to European settler farmers by implementing the Caledon Code of 1809. It allowed European farmers to keep children of slave fathers and Khoe mothers on their farms until they reached the age of twenty-five.Footnote ⁵⁶ While the tension between the Europeans and the indigenous populations never completely eroded, the situation gradually stabilized after the second British occupation.Footnote ⁵⁷ By the end of our period of investigation, European settlers had regained control over most of the desirable land in the district, though some Khoisan were permitted grazing rights European controlled land.Footnote ⁵⁸

The British takeover of the Cape Colony also resulted in a major exogenous shock to the settler population, not least due to the arrival of some 4,000 British settlers to the eastern parts of the Colony, but perhaps more importantly with the ban of slave importation in 1808.Footnote ⁵⁹ While the use of slaves was not as widespread in the Graaff-Reinet district as it was in the southwestern Cape, there is evidence suggesting that frontier farmers were nevertheless affected by the increased prices of slaves and the resulting increase in wages for Khoisan labourers.Footnote ⁶⁰

These changes took place in parallel to the closing of the frontier, in both an economic and administrative sense. Administratively, the establishment of the Uitenhage district in 1804 effectively closed the southern border of Graaff-ReinetFootnote ⁶¹ leaving only the northern border open. More importantly, the availability of land was substantially reduced during the period of investigation. We lack systematically recorded data to quantify the closing, but all qualitative evidence suggests that, although land was initially cheap and plentiful, settlers, particularly those with expanding flocks, soon found sufficient grazing lands more difficult obtain. By 1812, only thirty-nine per cent of married men owned their own farm, and there were no independent Khoisan settlements left in the district.Footnote ⁶² This does not mean that there was a large group of landless Boers, as many occupied land illegally or rented land from relatives or friends.Footnote ⁶³ Still, the possibility of accessing land cheaply was becoming more limited and, in 1813, the loan farm system was replaced by the quitrent system. Coupled with ineffectual administration, the processing of new land claims became increasingly protracted.

In 1824, the district boundaries were officially extended beyond the Orange River to meet the “insatiable need of an ever-increasing population for more land”.Footnote ⁶⁴ Increased population pressure further limited access to grazing lands with sufficient water supply. In 1826, the Landrost of Graaff-Reinet, Stockenstrom concluded that, “[...] when we speak of occupation, there is not even a stagnant pool that keeps rain water for any length of time which is not regularly occupied, so that of course no spring remains vacant”.Footnote ⁶⁵

The frontier was never entirely closed. Instead, the increased population pressure and shrinking land availability led to the well documented Great Trek of 1834–1838, as trekboers crossed the Orange River and moved further into the northeast of present day South Africa.Footnote ⁶⁶

DATA AND METHODS

Until recently, most research on historical household formation in South Africa has been anecdotal, either coming from traveller accounts, or based on records that are subject to selection problems.Footnote ⁶⁷ New research offers a more comprehensive account of the settler fertility transition beginning in the second half of the nineteenth century.Footnote ⁶⁸ But very little is known about pre-transitional household composition in this society and geographically disaggregated micro-level data are needed to better understand the demand for children on frontier farms and how it differed across socio-economic strata.

We combine two newly transcribed datasets. The first is the Opgaafrollen: annual tax censuses collected between 1663 and 1834 by the Dutch East India administration, and after 1795, by the British colonial administration, of all free households of the Colony. Household-level information includes name and surname of the head of the household and spouse, the number of children present in the household, the number of slaves and indigenous Khoisan employed, and several agricultural inputs and outputs, including cattle, sheep, horses, wheat sown, wheat reaped, vines, and wine produced. All males over the age of sixteen were assessed for tax purposes. Unless they headed a household, females were not included. The full series of Opgaafrollen are in the process of being transcribed and the household contained therein, identified and linked across years using a probabilistic record linkage strategy to create an annual panel of household production across more than a century.Footnote ⁶⁹

We use a sub-sample of these censuses for the Graaff-Reinet district covering the period 1798–1828. The panel contains 35,199 observations over twenty-eight years,Footnote ⁷⁰ comprised of 10,162 unique households. But, since these data do not contain information on birth dates (crucial in controlling for, and distinguish between, age, period, and cohort effects) or death dates (crucial for controlling for lifecycle wealth effects), we supplement these data with individual-level demographic data for the household head from the South African Families Database (SAF). Obtained from the Genealogical Institute of South Africa (GISA), these data contain information on all settler families from 1652 to approximately 1830 as well as those of new progenitors of settler families up to 1867. The probabilistic record linkage strategy that was used to identify and match households over time to create the CGHP is applied to identify and match heads of households across these two sources with a match rate of between 40 and 60 percent annually.Footnote ⁷¹

Information on household assets and production in the Opgaafrollen was used to estimate household wealth.Footnote ⁷² We choose to estimate separately what we will henceforth refer to as “asset wealth” and “capital wealth”. We take the asset component of wealth as the total real value of livestock, by multiplying the number of small livestock (sheep, cattle and goats) owned in a given census year by the estimated value thereof using a series of livestock prices from probate inventories (MOOC 8 Series, TANAP 2012).Footnote ⁷³ In order to determine the capital component of wealth, we sum all productive assets owned by an individual in a given census year.Footnote ⁷⁴ Through the selection of variables, we have tried to identify those elements necessary to capture improved capacity to produce a surplus for the market. These include number of horse wagons, horses, oxen, ox wagons and carts.Footnote ⁷⁵

CONCLUDING DISCUSSION

Since the 1960s, studies have found that fertility declines in the nineteenth-century Western world cannot be fully understood without considering the population dynamics in the rural areas. Focusing on the rural US and Canada, scholars found that fertility levels systematically differed between newly established frontier regions and older ones. Fertility levels were significantly lower in the established and more densely populated areas compared to the less densely populated frontier regions. Although a lack of data prevents more precise empirical testing, these systematic differences in fertility levels have commonly been explained by the land–labour hypothesis. Fertility levels were higher in frontier regions either because the demand for child labour was larger and/or the cost of having children was lower.

In this paper, we contribute to the debate by analysing the relationship between land availability and children present in settler farming households in the Graaff-Reinet district at the eastern frontier of Cape Colony. Different from the frontier literature, we find for the Graaff-Reient district that the closing of the frontier was associated with an increased presence of children.

We identify two weaknesses in the frontier literature: the failure to account for wealth inequalities and the failure to distinguish between family and non-family labour. Rural populations will respond differently to shrinking land availability depending on their wealth because the latter determines the available adaption strategies. A rich farm household may respond to diminishing land availability by substituting labour for capital, while this option is simply not available to a poor household. Wealth inequality in Graaff-Reinet was considerable and the vast majority of frontier farmers struggled to subsist. So, while a small minority did indeed respond to shrinking land availability by developing more capital-intensive methods of production, the poor did not. We find that the poorer farmers responded by increasing their relative use of family labour. This strategy makes economic sense if we – different from the frontier literature – make a clear distinction between family and non-family labour. The employment of non-family labour is dependent on the marginal productivity of labour. This is not the case for family labour as shown by development economists, agrarian historians, and economic historians. On the contrary, in pre-industrial societies, farmers will respond to shrinking land availability by adding more labour to land despite decreasing marginal productivity of labour. This Boserupian path was what the majority of the Graaff-Reinet population followed. Lacking alternative means to using family labour, they intensified labour use in light of shrinking land availability, which created a demand for having more children.

This leads us to conclude that, for the poor, we see a negative correlation between land availability and labour demand. The reason why the Graaff-Reinet households experienced, on average, an increased presence of settler children in the midst of a closing frontier, is that a large majority of the population consisted of the poor. In that regard, our study does not necessarily contradict previous findings. It may very well be that farmers in nineteenth-century US and Canada were, on average, wealthy enough to respond to shrinking land availability by employing more capital intensive farming methods. The implications of our findings are, however, that the positive relationship between fertility and land availability may not be universal, as previously thought, but largely dependent on levels and distribution of wealth among the rural inhabitants.