C. A comparison of the cost of free and slave labour: slavery as Hobson’s Choice

A direct comparison of the two kinds of labour in the same period seems to be impossible: precisely because slaves and wage labourers barely overlapped chronologically in the service of Asante masters/employers. It should be noted that this observation is itself evidence that free labour was not a real alternative to slaves as a source of regular labour while slavery was legal (and before the economic conditions conducive to it were changed by the adoption of cocoa cultivation, as we will see in Section 3).

But it is possible to develop a quantitative perspective from an oblique angle. First, European prospectors working in major Akan gold-producing areas around 1900 obtained an average yield of a shilling’s worth of gold (range 10d. - 1s.2d) per day.⁴⁰ Having compared this with other recorded returns, Garrard concluded that it ‘is probably reasonable to take’ this range as ‘the average daily earnings of an Akan gold-winner up to the end of the nineteenth century’.⁴¹ Broadly consistent with this is a statement (not cited by Garrard) in the colonial administrative report on Asante for 1898 that, while Asante men had largely abandoned ‘gold washing’ in favour of the wild rubber industry, ‘Gold washing is now carried on mostly by the women and children, who can make on average 9d. to 1s. a day’.⁴² On the basis of these observations we can assume that, around the turn of the century, the opportunity cost for a free able-bodied man in Asante of working for wages rather than working for himself in the dry season would have averaged at least a shilling a day. In 1911-13, when wage labour was beginning on Asante cocoa farms, the prevailing wage rate for (male) unskilled labour seems to have been between 1s. and 1s. 3d., of which the latter could include a 3d. subsistence allowance.⁴³ The cost of living for labourers does not seem to have changed much c.1898-c.1913, though this issue requires further research. So far price indices have been calculated only for various categories of imported goods, and from 1900. The most relevant of these, ‘clothing, leather and textiles’ showed a rise from 9 in 1900 to 10 in 1913.⁴⁴

What stands out is that there was a gross difference in cost between wage and slave labour. Indigenous gold production was concentrated largely in about three months of the dry season.⁴⁵ If we make the relatively conservative assumption that labourers did the equivalent of 20 days work a month for three months a year, average gold-season receipts at 1s. a day would average a total of £3 a head. Garrard assumed 75 days ‘at the most’ in three months,⁴⁶ which would imply average earnings of £3.75 from a relatively intensive season. The appendix assembles the available evidence on the prices of slaves within nineteenth-century Asante and in the markets from which, directly or indirectly, slaves were imported into Asante. The following table expresses those prices in terms of the number of days of gold mining required to pay for a slave.

Where slave prices were expressed in headloads of kola nuts, the implication is similar. In the Appendix, Table A.1 gives the following such prices: in the admittedly flooded markets of 1817, 1 load (Bowdich’s ‘basket’); in the years immediately preceding 1874, 3 - 5.8 loads (calculated from figures provided by Ramseyer and Kühne) or 2-3 (Brackenbury), or 4.5 - 5 for a man and 7 for a woman (according to Rattray’s informant). According to Lovejoy, ‘producers needed the harvest from only a few trees to obtain a load of 2,000 nuts. Individuals often accumulated as many as 10 loads.’⁴⁷ Ramseyer and Kühne commented that ‘an Ashanti . . . can easily collect at home’ enough nuts to ‘enable him to buy a slave in Salaga.’⁴⁸ Terray has likewise found that slaves were very cheap calculated in relation to the prices of oxen, salt and guns from various observations at Salaga in the last quarter of the nineteenth century.⁴⁹

To be sure, the intending purchaser would often have to use some of his (as it usually was) earnings from gold or kola production to pay for items other than slaves - though, importantly, little would normally go on food. For a young man working mainly for himself it could take a few years before he could afford a slave, as an example I have given elsewhere shows.⁵⁰ But someone able to concentrate the earnings of more than one miner or kola collector, such as a household head and/or an existing slave or pawn holder, could acquire slaves more frequently. Even if slaves could not be induced to work as hard as free people (a risky assumption),⁵¹ they could be expected to pay for themselves within a very few years if they did not run away.

At least in one period a higher rate was paid for female captives.⁵² Though this was not true later, it confirms that the prices paid for females included an element reflecting their value as potential wives and as potential mothers of free men’s heirs. What matters here is that the fact that prices for females were raised by their roles outside direct production reinforces the conclusion that the price of slave labour power was very low compared to the price necessary to hire free labour.

Evidence about what happened in the early colonial period, when the end of slavery was first anticipated and then gradually realised, confirms that there was/had been a decisive disparity between the costs of slave and free labour. In 1925 Rattray was informed by village elders in Wam (south of Berekum in western Asante) that ‘We have now stopped [gold] mining because it only pays when you have slaves’.⁵³ In a general economic and social survey in 1931, the Chief Census Officer stated that the shafts made by traditional mining methods ‘are now no longer worked, as with the disappearance of slave labour the results would not possibly be sufficient to cover the expense.’⁵⁴ Earlier, at the very beginning of the colonial occupation, a British traveller opined that ‘On the abolition of domestic slavery the kola industry in Ashanti will tend to die out’.⁵⁵ His prediction proved too pessimistic, but his implicit assumption that kola producers would not be able to afford to hire regular wage labour seems to have been justified. In the mid-1930s, for example, ‘In the principal kola areas in Ashanti it is in most cases the owner himself who picks and prepares the nuts for sale but occasionally he may let his trees to another who will do this work for one half of the crop.’⁵⁶ An overview of the economic significance of slavery was provided by the chief and elders of Adanse in 1906. In a petition for continued government toleration of slavery and pawning they observed that, as Asante chiefs and headmen, they depended on the labour of slaves and the descendants of slaves ‘as we have no money like Europeans to hire men to do necessaries for us’.⁵⁷

The implication of the evidence considered in this section is that - until cocoa cultivation was adopted early in the colonial period - compulsion was a necessary condition of labour being cheap enough for there to be a labour market at all. Without coercion the reservation price of labour consistently exceeded the maximum wage rate which potential employers could have afforded. Amid such physical and institutionalised abundance of land there was too little occupational specialisation to generate a market for goods sufficiently remunerative to make it profitable for a producer to employ a labourer on any terms that it would be profitable for the labourer to accept. It took force to reduce the labour-supply cost enough to make it possible for the supply and demand curves to intersect. Thus Domar’s model applies: ‘the wage of a hired man . . . will have to be at least equal to what he can make on his own farm; if he receives that much, no surplus (rent) will be left for his employer.’⁵⁸

That cocoa began to be adopted before slavery was made effectively illegal makes it interesting to examine the ending of Asante slavery as a possible case of ‘induced institutional innovation’. Can it be explained in ‘new institutionalist’ terms, as a response to a shift in relative factor prices which made it socially efficient for the property rights regime to be changed to accommodate the new economic circumstances?⁵⁹ The spread of cocoa-farming altered the relative scarcities of labour, land and capital: and this was reflected, implicitly and explicitly, in changes in relative factor prices. Did this shift contribute to the end of slavery in Asante? A ‘Third World’ parallel, to judge from Feeny’s account, would be nineteenth and early twentieth-century Thailand, where he sees a transition from rights in people to rights in land as land became less abundant.⁶⁰ Sub-Section 3A asks whether the concept of induced innovation can help to explain why, after waiting twelve years, the British finally introduced prohibition. In tackling that question evidence is provided about the impact of abolition on cocoa farming. This is much needed because previous publications on the history of cocoa-farming in Ghana generally have said little either way about the effect of the decline of property rights in people upon the growth of cocoa-farming.⁶¹

Turning from policy decisions to changes in practice, 3B examines whether the cocoa ‘revolution’ helped to turn mere prohibition into something approaching genuine abolition. This idea was put forward for the Gold Coast by the pioneer economic historian of British West Africa, McPhee, in 1926. He asserted that the abolition of slave dealing and the legal status of slavery there in 1874 had ‘negligible’ effect until changes including ‘the development of the cocoa industry’ dealt it a ‘staggering blow’ by, he implied, creating a ‘free labour market’ offering ‘alternative employment’ to former slaves.⁶² In the tradition of McPhee, but for West Africa generally, Hopkins argued in 1973 that export agriculture assisted newly emancipated slaves establish themselves as free farmers and migrant labourers. This in turn helped make it possible for slavery to be ended with relatively little social and economic disruption.⁶³ It was not until the 1980s that detailed work began to be published on the decline of slavery in the Gold Coast Colony, to the immediate south of Asante.⁶⁴ Such research has enormously enriched and, implicitly, qualified McPhee’s sketch, but without mentioning his hypothesis or considering the evidence on it in any detail. 3B examines the proposition for Asante, and goes on to ask whether the growth of cocoa-farming, and related economic changes, help explain the social unevenness of the decline of slavery.
3A. Induced institutional innovation? The decision to prohibit slavery and the impact of cocoa cultivation
This section examines the British decision effectively to ban slavery in relation to the hypothesis of induced institutional innovation. The major economic change of the period was the adoption and rapid spread of cocoa production. This progressively altered the factor ratios, and is thus very much the kind of economic change that, in terms of the hypothesis, might induce institutional change. We ask first how much factor ratios had shifted by 1908. We go on to examine the contribution of various forms of coerced labour to the cocoa take-off in Asante. We then consider whether the colonial government’s decision to forbid slavery, when it was eventually made, can be considered as in any sense induced by economic change within Asante.

By its nature the spread of cocoa cultivation was progressively increasing the ratio of fixed capital (mostly in the form of cocoa trees) to labour, and reducing the relative abundance of land in relation to labour. But how far had this shift gone by the critical decision-year of 1908? To answer this question we need to estimate how much land had been put under cocoa trees by then. The basic method is to start with the annual export figures (as noted earlier, we can take exports as equivalent to output). Ideally, we would select a year in which all cocoa beans sold were the produce of trees planted by the end of the major rains in 1908, which was, in effect, precisely when the prohibition on slavery was introduced. Yet it would also be a year in which all those trees had come into full bearing. Though no such perfect year exists, we can assume that almost all the cocoa sold in 1914 came from trees planted before the prohibition; but a high proportion of these trees would not have reached peak bearing. Thus to use the 1914 crop to estimate pre-prohibition planting would be somewhat to understate the latter. To use the 1919 crop, on the other hand, would be to overstate it. Most of the trees planted before prohibition would have reached or come close to peak bearing by the time they contributed to the bumper 1919 crop, but it would have been boosted by output from newer-bearing trees planted c.1909-14. Thus the true acreage planted by the time of the ban would be represented by an output figure somewhere between the 1914 (11,633 tons) and 1919 (31,944 tons) harvests, but probably closer to the former.⁶⁵

To obtain the acreage we divide the output by the average yield per acre. Figures for the latter, from African farms in Asante and/or the Gold Coast Colony, vary from 7 to 10 60-lb. ‘loads’. Given that much of the acreage was not yet in full bearing, it is appropriate to reject the highest of these figures. This was an estimate for Asante in 1932-3, by J. C. Muir, the head of the Ashanti Division of the Agricultural Department.⁶⁶ By that year a much higher proportion of the trees would have been at their peak. The lowest figure, which was Beckett’s finding for the village of Akokoaso in Akyem in the Gold Coast Colony in the early 1930s,⁶⁷ is probably too low for our purposes. For the evidence is that yields in the early decades of Asante cocoa farming tended to be higher, as Muir’s figure dramatized. The safest yield number for this period is perhaps the 9 loads per acre reported by Cardinall for 1931, apparently for colonial Ghana as a whole.⁶⁸ Assuming 540 lbs (0.241 tons) an acre, the area of bearing cocoa was 75 square miles in 1914 and 297 square miles in 1919. This amounted to 0.31 and 0.84 per cent, respectively, of the total surface area of Asante. Even on Beckett’s 420 lbs an acre, the proportion would have been only just over 1 per cent (1.07) in 1919, the first year in which such a proportion had been even approached. Therefore, though the territory of Asante included some savanna and other land unsuitable for cocoa, including towns, it is clear that land was still relatively abundant. Virtually all the population of Asante with control over their time had free or nearly free access to land, while labour was still relatively scarce. Thus, as yet, cocoa had not changed factor ratios enough to generate a shift in relative factor prices remotely sufficient to encourage property-owners to convert themselves from masters to employers.

Consistent with this is the extensive use of corvée in the early colonial period, at the initiative of chiefs and of the colonial government. For this can be seen precisely as a response to the fact that the conditions which had made coercion essential to the recruitment of extra-familial labour before 1896 still obtained after it, certainly beyond 1908.⁶⁹ Admittedly, the labour requisitioned by the government was underpaid rather than unpaid (except for the post-revolt punishment tasks). For instance, in 1898 the government paid chiefs 9 pence a day, including subsistence, per worker supplied for ‘a day’s march averaging 18 miles’.⁷⁰ It was the same government source which observed that ‘the women and children . . . can make on average 9d. to 1s. a day’ in gold washing, an activity which, as we have seen, had been largely given up by men because it was harder and less profitable than rubber production. So we can assume that, even aside from resentment at the colonial occupation, no voluntary labour would have been forthcoming in the dry season of 1898 at 9d. a day for such an arduous exertion as long-distance headloading. The British government thus got its labour at significantly below the market-clearing price - even for the dry season. It is harder still to envisage 9d. a day attracting labour services from Asante farmers during the rains, when planting and weeding needed to be done on their own farms to assure the adequacy of the next food harvest.

The chiefs, too, were obtaining labour at below any plausible free-market price when they called their subjects to make cocoa farms for the stool. While food crops would have been planted to shade the young plants, the rights to them would presumably belong to the chief. Let us consider Amansie district, to the south of Kumasi, a case which I have studied in detail - so far the only detailed study of the early years of cocoa production in Asante. One of the findings, from fieldwork plus some archival evidence, was that corvée was a major source of labour on early cocoa farms. This was so in the case of first cocoa farm in Amansie district, Kwame Dei’s at Asiwa.⁷¹ The labour of free subjects created the first stool cocoa farm in Manso Nkwanta, which was said to be as old as any cocoa farm in the town.⁷² In Esumeja, too, the early stool cocoa farm was made by communal labour.⁷³ Of the village of Tetekaso, in Amansie but under the Esasehene, it was said in court in 1957: ‘I remember about 1911, Agyie-Tiaa was enstooled chief of Tetekaso, and two big cocoa farms were made for the Tetekaso stool’.⁷⁴ Agyie-Tiaa however denied this, claiming ‘that the farms which I have are by my own labour, and nobody has made the farm[s] for me’.⁷⁵

There are other known cases of chiefs’ farms in Amansie district. It is highly likely that these too were made by corvée, though I cannot confirm this. In 1913 F. C. Fuller, the Chief Commissioner of Ashanti, noted that Chief Owiabu of Feyase had ‘made’ four cocoa farms. He ‘will relinquish the stool . . . and as he has disposed of 2 cocoa Plantations the remaining two will become Stool property against which Owiabu will have no claim’.⁷⁶ The Adumasahene established a stool farm as one of the first cohort of cocoa farms in the Ofoase-Kokoben area, soon after 1900.⁷⁷ At Asubina near Manso Nkwanta the chief’s farm was ‘about 8 years old’ in 1921.⁷⁸ A stool cocoa farm was made at Amoafo: exactly when is unclear, but probably before 1914.⁷⁹ As for Adanse district, in 1911 Adansehene Kobina Foli told a commission on land tenure ‘I have a very large’ cocoa farm.⁸⁰

What of the mass of private farmers? Perhaps the most revealing context in which to consider this is the fact that cocoa was not heavily planted in Asante until shortly after the Yaa Asantewaa War, the armed rising (mainly of Kumasi chiefs and their followers) against the British in 1900. An obvious reason for this is that diffusion of the exotic crop took time. Though it had been experimented with over quite a long period in the Gold Coast, it was only in 1893 that annual exports had reached 1 ton.⁸¹ Further, as noted earlier, it was not until 1903 that the railway reached Kumasi, decisively improving the profitability of the crop to Asante growers. But it may well be that an additional reason was an acute shortage of labour. This was not so much because of the British moves against fresh imports of slaves: the activities of Samori Toure’s army in the Asante hinterland had permitted large-scale imports in the immediately preceding several years.⁸² But the colonial government’s ‘constant’ demands for carriers (in the words of the Resident, quoted earlier) were clearly disruptive of Asante enterprise. This, in turn, helps explain why chiefs were so prominent as pioneer cocoa-farmers in Asante: they could draw on corvée themselves. But after the rising the Asante cocoa take-off gathered pace: initially perhaps especially in those chieftaincies, such as Bekwai, which had not revolted and therefore escaped punishment labour.

Slave (and pawn) labour was far more important than wage labour in the wave of planting, mainly in the southern and eastern areas, c.1900-1916, which established cocoa cultivation in Asante. Despite the shortage of new slaves and the flight of some old ones, it is reasonable to assume that slaves - including efie nipa, the children of slaves, as well as first-generation captives - still constituted a large proportion of the labour force. As Fuller was to note of Asante generally, ‘Although the purchase of slaves had ceased since the British occupation in 1896, all families of importance owned slaves, who, in reality, formed part of the family circles’. All members of the ‘family circle’ who could contribute did so.⁸³ Kwame Dei, who was the pioneer of cocoa-farming in at least his district of Asante, was remembered as having used slaves and pawns on his cocoa farm at Asiwa near Lake Bosumtwi.⁸⁴ In 1913 the provincial commissioner in Obuasi was faced with a ‘Curious claim by ex slaves to share in profits of a cocoa farm left by their deceased master’.⁸⁵ Such a claim can only have been based on the premise that they had laboured in making or operating it. Conversely - if Amansie district, which was prominent in the cocoa take-off, was representative of the cocoa expansion generally - wage labour on early cocoa farms, when it existed at all, was limited to some casual work. Regular hired labour (six-month, annual or sharecrop) does not seem to have existed before at least 1918. By then enough trees had been planted in the district to yield over 10,000 tons a year when they came into bearing, a milestone reached by 1921-2.⁸⁶ Overall, it is reasonable to conclude that the speed and scale of the early growth of cocoa cultivation owed much to inputs of coerced (slave, pawn and corvée) labour, supplementing self and free family labour, and probably co-operative work groups.

Further evidence on this comes from Fortes’s notes of an interview he conducted in Asokore, in Sekyere, in 1945.
Wealthy cocoa farmers. (They are not very willing to talk about other people[’]s wealth. I asked if they knew anyone who had become wealthy only from cocoa farming) At Nyamfa there was a man Kwame Dampa, who had very big cocoa farms wh[ich] he made in the early days. He had ‘[e]fie nnipa’ (descendants of slaves) to work his farms and so got very rich. But if you do not have many sons or money to employ labour you can’t make big cocoa farms.⁸⁷
In the first twenty years or so of Asante cocoa-farming, relatively few people could have afforded regular wage labour. But those who had the services of slaves (of whatever category) and/or of pawns could nevertheless expand their cultivation beyond the scale permitted by their own labour, plus the service of free family members.

There is no evidence of Asante farmers, having established their farms, choosing to employ labourers rather than to own them. Their problem was that the banning of slave trading made it difficult to acquire new slaves. The price of slaves had been so low compared to the cost of regular wage labour that it is hardly surprising that there is no evidence of emancipation at owners’ initiative; whereas, as we have seen, there are a number of documented cases of slaves running away. The growth of cocoa-farming was not enough to make the prohibition of slavery a pareto gain.⁸⁸ Overall, there is no evidence of any economic pressure for abolition from within Asante: and no sign of it in the behaviour of chiefs or their free subjects; nor from the local colonial officials.

Politically, it is very clear from the fairly voluminous and certainly protracted correspondence between the colonial administration in Asante and their superiors that the decision to apply the prohibition on slavery to Asante was anything but an initiative of the colonial ‘men on the spot’. On the contrary, it was imposed upon them from above, and attempts to water down or re-interpret the instructions were specifically and explicitly overruled by the governor.⁸⁹ Far from being induced by economic change, the ban was very much exogenous to Asante and even to its colonial administration.

The comparison with corvée is interesting. Briefly, there too, the political pressure for abolition ran ahead of the availability of alternative sources of affordable labour for the tasks concerned. In the case of forced labour for road maintenance, however, the pressure was weak enough for change to be evaded to a considerable extent, until the very era of decolonization.⁹⁰