Global and local forces in deindustrialization: the case of cotton cloth in East Africa’s Lower Shire Valley

ABSTRACT

Numerous scholars have suggested that nineteenth-century industrial decline in the global “periphery” was driven by externally wrought global forces that promoted cash-crop agriculture and dis-incentivized local industry, particularly strong global demand for tropical agricultural commodities and increasing competition with imports of cheap, factory-produced manufactures from industrializing regions. To what extent did global market forces affect production choices, and to what degree did local forces guide outcomes? The deindustrialization process is investigated through a case study of Malawi’s Lower Shire Valley, where the Mang’anja cloth industry declined – and cash-crop production began – in the second half of the nineteenth century. I demonstrate that changing production choices were not directly motivated by global market opportunities. Indeed, other cloth-producing sub-Saharan African regions faced nineteenth-century global forces but did not deindustrialize. Rather, economic change in the valley was stimulated by local factor-endowment shifts precipitated by both global and local forces. Labour declined sharply due to slave raiding and famine, while supplies of fertile land increased due to environmental change. Within this altered context, Mang’anja villagers responded by abandoning labour-intensive cloth production in favour of cash-crop cultivation. In more labour-abundant African regions, on the other hand, cloth production continued to thrive alongside cash-crop exports. The mechanisms behind deindustrialization can only be understood through careful local-level examination of the local contexts that influenced responses to broader global processes.

KEYWORDS:

• Deindustrialization
• cotton cloth
• East Africa
• global
• local

In the late 1850s, David Livingstone entered the Lower Shire Valley in southern modern-day Malawi (colonial Nyasaland) and marveled at the prodigious manufacture of machila cloth made from local cotton.¹ From at least the sixteenth century, the durable unbleached cloth was the valley’s primary regional export commodity, used to obtain iron agricultural implements from the nearby Shire Highlands and to trade with the more distant Lower Zambezi region, where hand-woven cloth was the “single most important trade item”.² During the 1860s, production and export of Mang’anja cloth rapidly halted and subsequently failed to regain footing, while similar labour-intensive cloth industries successfully continued in other parts of Africa and beyond well into the twentieth century.³ Many deindustrialization theories argue that it was primarily global market forces and foreign imports that precipitated industrial decline in places like the Lower Shire Valley. Why, then, did the valley’s cloth industry deteriorate while others persisted?⁴

I argue that the disappearance of production in the Lower Shire Valley was prompted by local factors that were influenced in part by global forces. This occurred over two broad contiguous periods, but was only cemented during the second. Production was initially disrupted in the 1860s when regional slave raiding, driven by global demand for slave-produced commodities, temporarily threw the valley into chaos and halted cotton cultivation and cloth production. At the same time, local drought conditions led to an unprecedentedly severe famine. The resulting devastation of the Mang’anja population produced lasting effects on the local labour supply, which affected production choices in the decades that followed.

The second phase of deindustrialization began in the mid-1870s when raiding subsided in the valley. Village life resumed, but organized cloth production did not. I argue that this sectoral shift was prompted not by global market forces or competition from foreign cloth imports but by an increase in productive land relative to diminished labour. In the mid-1870s, a long-term drop in Shire River dry-season levels exposed large tracts of exceptionally fertile riverbed in the valley. With a contracted population and no extant slave-labour institutions to shore up diminished local labour supplies, villagers responded to the new factor-endowment ratio by abandoning community-based labour-intensive industry in favour of higher-productivity, household-based cash-crop farming. The labour supply issue was solved, but the valley’s economic activity narrowed, creating a dependence on agricultural exports. The situation was markedly different in sub-Saharan African regions with larger populations and slave-labour institutions, like Ethiopia, the Benadir Coast, and northern Nigeria, where cloth production and cash-crop cultivation flourished simultaneously and enjoyed sectoral linkages.

This paper examines the mechanisms behind the Lower Shire Valley’s shift from industry to cash crops. The general claim I make is that a synthesis of global and local forces and a careful unpacking of their respective and interrelated influences is crucial for understanding deindustrialization processes. More specifically, this case study suggests that cloth production was more likely to continue in African regions with relatively large populations and coercive labour institutions, which provided ample labour and local markets for domestic cloth. Areas with more constricted labour sources, like the Lower Shire Valley, were more inclined to abandon cloth production in favour of less labour-intensive alternative production opportunities geared toward expansive global markets.

Global and local forces in deindustrialization

In the economic literature there are various perspectives on the causes of deindustrialization. Williamson argues that nineteenth-century global demand prompted an “industrialization-driven great divergence” when “peripheral” regions reallocated labour from industry to primary production as European manufacturing productivity and demand for inputs increased, creating strong terms of trade for primary products relative to imported manufactures.⁵ Dependency theorists have reasoned that large-scale imports of manufactured goods undermined existing local industries through increasing market competition and diversion of labour to non-productive activities, like ivory and slave hunting, to procure sought-after imports.⁶ Other theorists seeking external influences have pointed to the role of colonial metropolitan agendas in driving African production choices.⁷

Denemark and Thomas, on the other hand, have suggested a theoretical incorporation of local factors into historical studies of economic development, following a tack taken by Brenner, but they still maintain that global market forces generally prompted local-level change.⁸ In his work on the Lower Shire Valley, Mandala has highlighted the role of environmental forces in influencing local production choices. However, with respect to textile production, he focuses principally on “the impact of merchant capital on the cloth industry,” which, in his view, “was so devastating that some oral historians today deny that the Mang’anja ever made cloth from cotton.⁹ The Heckscher–Ohlin Theorem of international trade points more directly to local conditions as the major determinant of production and trade developments. The theorem proposes that resource allocation is governed by specific local factor endowments as a country or region will tend to produce and export goods – industrial, agricultural, or otherwise – that require the intensive use of the factor that is most abundant in that particular locale.¹⁰ Tosh and Austin have stressed the need to take into account location-specific resources – particularly labour and land – to understand production choices in Africa.¹¹ The relationship between Africans and their environment has received renewed attention, but with a caveat against environmental determinism, for the impact of environmental influences is contingent upon a vast array of possible human reactions.¹² I show that in the Lower Shire Valley, local environmental and demographic change helped precipitate socio-economic reorientation, but within the context of the local institutional framework and a changing global trading landscape.

In the Lower Shire Valley, production choices were impacted by outside forces, but in different ways than global-focused theories have suggested. The presence of imported cloth, for example, did not drive the decline of production. I have constructed new cloth import statistics using trade and shipping records of the primary nineteenth-century cloth exporters to East Africa (Bombay, the United Kingdom and the United States), which reveal that cloth imports into the region only took off in the last decade of the nineteenth century, well after the disappearance of Mang’anja cloth production by the early 1880s.¹³ Figure 1 shows a gradual increase of cloth imports into East Africa during the late pre-colonial period when foreign cloth was acquired in the East African interior primarily via ivory exports.¹⁴ Ivory, however, was not a traditional Mang’anja export good, nor were the terms of trade for ivory (relative to cloth) favourable in the interior.¹⁵ Further, much of the growth of cloth imports into East Africa during the last quarter of the nineteenth century was tied to an increase in exports of coastal products (i.e. cloves, copal, etc.), and imports were principally comprised of flimsy British and Indian cloth, consumed in large part by emerging coastal producer groups, rather than the durable unbleached American “merekani” cloth that was strongly favoured in the interior.¹⁶ Thus, as Figure 1 illustrates, villagers in Malawi had very limited access to foreign-made cloth until a dramatic upsurge of imports occurred around the start of the colonial period in the 1890s when increasing amounts of cloth were regularly purchased using growing cash-crop profits.

Figure 1. Cotton cloth imports per capita into East and West Africa, 1848–1946. Sources: Pre-colonial figures (1840–1899): US shipping records: MSS901, David Pingree Papers, Phillips Library, Peabody Essex Museum, Salem (PEM); MSS24, Emmerton Family Papers, PEM; MH23, Michael Shepard Papers, PEM; MH235, West Family Papers, PEM; Arrival and Departure of American Vessels, Jan 1, 1857 to June 29, 1894, Record Group 84, Records of Foreign Service Posts, Zanzibar, Vol. 84, The National Archives, College Park; Bombay annual trade data, 1848–1901: Report of the Commerce of Bombay; Annual Statement of the Trade and Navigation of the Presidency of Bombay; Jenkins, Report on the External Commerce; UK annual trade data, 1857–1901: Annual Statement of the Trade of the United Kingdom; Secretary of State, “Zanzibar,” 33; Pratt, “Zanzibar,” 841; Cave, Zanzibar, 5; colonial figures (1897–1946): British Central Africa/Nyasaland Blue Books, 1902–1938, Colonial Office (CO) 452/6–42, The National Archives, London (TNA); Tanganyika Blue Books, 1920–1946, CO 726/1–30, TNA; Uganda Blue Books, 1905–1916, CO 543/1–36, TNA; Statistical Abstracts for the Several British Oversea Dominions and Protectorates, 1909–1923, 1924–1933 and 1929–1938, The Hague, Central Bureau of Statistics (CBS); Bureau of Manufacturers, “German East Africa,” 776; Department of Commerce, “German East Africa,” 177; population data: Frankema–Jerven African Population Database 1850–1960, version 1.0.

Even if cloth imports had heavily penetrated the valley earlier on, the mere presence of foreign cloth did not automatically spell disaster for local industry. As Figure 1 shows, imports per capita were comparatively higher in western Africa, where cloth production flourished in many places, like Kano in northern Nigeria, into the twentieth century.¹⁷ In northeastern Africa, too, weavers in Mogadishu on the Benadir Coast and throughout Ethiopia retained their foothold in regional markets well into the twentieth century, sometimes by adapting products to compete with foreign varieties.¹⁸ In the interior-situated Lower Shire Valley, by contrast, cloth imports arrived more slowly, yet production ceased entirely during the last half of the nineteenth century, and the former cloth-producing Mang’anja turned to agricultural production for global export. What accounts for the very different trajectory of the Lower Shire Valley relative to places like Mogadishu, Ethiopia and Kano? All were traditional cloth-producing areas confronted with similar global forces. However, their local circumstances differed markedly, particularly with respect to their population levels and labour institutions, which dramatically impacted the nature of respective regional responses to global stimuli.

The following section describes the decline in the Mang’anja population as a result of mid-nineteenth-century slave raiding and famine. Contemporary discussion comes from European travellers whose abolitionist agendas no doubt coloured their writing. However, the regional population drain from slave raiding must have been great, as an official report from Zanzibar noted that nearly 80% of the slaves imported in 1860 came from the Lake Nyasa area.¹⁹ Further, slave trading was a common feature in Ethiopia, the Benadir Coast and northern Nigeria, but severe attenuation of local populations and industrial decline was not similarly reported there by European travellers. Taken with a grain of salt, these reports provide insight into the causes of Lower Shire Valley population decline and consequences for production choices.

Slave raiding and socio-economic disorder

In nineteenth-century Mang’anja villages, individuals from different households worked together to produce industrial goods, including cloth, with women and children often cleaning cotton and women sometimes spinning it into thread, while men regularly spun and invariably wove.²⁰ Subsistence agriculture, conversely, was organized on the household (banja) level and involved both mphala dryland and dambo wetland fields. Village-adjacent dryland fields were cultivated under the rain-fed munda system, while fertile wetland fields on the marshy banks of the Shire River required no rain under the drought-resistant dimba system. The wetland fields were annually refreshed with alluvial river deposits during the wet season, then exposed for cultivation during the dry season, which typically lasted five to six months (July–December) until the mid-1870s. These two ecosystems facilitated two different, sometimes overlapping agricultural seasons.²¹

However, the Lower Shire Valley temporarily became a “disaster economy” in the 1860s and 1870s when villages were regularly attacked by slave raiders seeking labour for Indian Ocean slave-plantation islands.²² Villagers sought refuge on small islands in the Shire River, leaving behind all “they possessed, except the little they could carry on their heads”.²³ Livingstone lamented, “[Villages] were all deserted: one where we […] two years before […] saw a number of men peacefully weaving […] was burnt”.²⁴ Disorder caused by raiding disrupted the relationship between Mang’anja households and village-communities, undermining work-group industry. Further, when villagers fled, they left behind their cotton plants, which grew exclusively in dryland fields since river wetland fields were, historically, not fully exposed for a long enough annual period to allow cotton plants to reach maturity (six months). More still, multi-year drought conditions led to a severe famine from 1862 to 1863, intensified by the disruption of cultivation.²⁵ Visitors reported, “The river-banks, once so populous [were] all silent […] an oppressive stillness reigning [and] the sight and smell of dead bodies was everywhere”.²⁶ Labour was focused entirely on producing what little food could be grown on river-island refuges. Any labour-intensive cotton and cloth production was halted, for cloth – unlike food – was not a survival necessity in the temperate valley.

While raiding and drought were only temporary disruptions, they produced lasting consequences. I argue that the resulting population drop profoundly influenced labour-allocation choices in the decades following the decline of slave raiding. In a single year, roughly 15,000 captives were taken from the area, while the 1862–1863 famine caused many deaths.²⁷ Contemporary estimates of the decline in the valley’s population range from 50% to an undoubtedly over-stated 90%.²⁸ Livingstone reported, “Labour had been  […]  swept away from the Great Shire Valley”.²⁹

Raiding likely affected the Mang’anja gender balance as raiders often targeted females for use as slave-wives.³⁰ Even if Mang’anja cloth production had immediately resumed in the post-slave-raid period, a relative decline in females might have created an input supply bottleneck given that women were reportedly involved in labor-intensive cotton cleaning and spinning processes.³¹ Perhaps more importantly, a gender imbalance may have placed more of the subsistence-agricultural labour burden on men, diminishing available time for male non-agricultural activities, including weaving.

The role of foreign imports revisited

Local disorder, brought on by external demand and local environmental forces accounted for the initial disruption of cloth manufacturing in the valley. But why did the industry not resume after slave raiding diminished by the late 1870s? Mandala points to the steady loss of traditional export bases for Mang’anja-produced cloth due to foreign competition.³² While imported cloth was not flooding the the valley prior to the colonial period, it had penetrated more heavily in other parts of southern east Africa, although still in relatively limited quantities, as a form of currency used to purchase captives in the Lower Zambezi region.³³ Eventually, competition with foreign cloth may have affected closer export markets, as Vaughan argues that emerging foreign-trade activity enabled “acquisition of exotic goods” in the Shire Highlands.³⁴

At face value, it seems plausible that the presence of foreign cloth in traditional Mang’anja export areas compromised the valley’s industry. But recall the cases of Mogadishu, Ethiopia and Kano, where weavers responded to competition not by closing shop but by adapting and expanding. Alpers argues that incorporation into “the world capitalist system of production and exchange” actually stimulated industry developments in spite of portents by contemporary “doomsayers” predicting failure in the face of competition.³⁵ Similar responses have been identified in resilient cloth industries in nineteenth- and twentieth-century Java and India.³⁶ Furthermore, Alpers has noted that Mang'anja machila cloth had long existed alongside and even competed with Indian cloth imports in the Zambezi region because of the domestic cloth's unique durability.³⁷ Further, a loss of export markets does not account for the valley’s decline in production for local consumption, especially since the substantial increase in the area’s cloth imports did not occur for several more decades (refer to Figure 1). Rather, local influences must be considered, particularly changes in the valley’s land-labour ratio. A decline in the local population not only reduced available industrial labour but also diminished the local market for Mang’anja-produced cloth. In far more densely populated northeastern and western Africa, larger markets demanded more product.³⁸ At nearly the same time, environmental changes in the Lower Shire Valley altered agricultural production possibilities and helped generate “alternative employment opportunities” in cash-crop production.³⁹

Land and labour in the Lower Shire Valley

Just prior to the decline of slave raiding, ecological changes began in the valley, which affected the Mang’anja dryland–wetland cultivation system once village life resumed. From the mid-1870s to the 1930s, dramatically diminished dry-season Shire River levels annually exposed increasingly larger tracts of exceptionally fertile marshland for cultivation. As of 1910, the dry-season depth was little more than one foot (0.3 m).⁴⁰ A colonial official reported, “Formerly the highway into Nyasaland, [the Shire] cannot be regarded as a navigable river”.⁴¹ This stands in stark contrast to Livingstone’s account from the early 1860s, which cites 20 ft. (3.65 m) as the lowest observed depth, corresponding to dry-season depths during the late 1930s when the river suddenly returned to its pre-1870s level.⁴²

Furthermore, the increasingly larger wetland fields were exposed for progressively longer annual periods, extending the wetland agricultural season until it ultimately reached 10 months – as opposed to 5 or 6 – by the early twentieth century.⁴³ However, this ecological shift was not associated with a decline in rainfall, which would have impaired dryland cultivation.⁴⁴ Villagers boosted agricultural capacity by developing a three-season system, including rainy-season cultivation of mphala dryland fields along with two consecutive dimba wetland cultivation periods during the river’s elongated dry season.⁴⁵ Export-oriented and subsistence-agricultural production came to dominate the entire Mang’anja year, leaving little space for non-agricultural activities. Further, economic activity came to be more family oriented rather than village oriented, for agriculture was pursued on the household level, while cloth production had been undertaken in village workgroups.⁴⁶

Local ecological changes clearly influenced production choices. However, taken alone, environmental change cannot adequately explain the abandonment of cloth production. In fact, geological evidence reveals that the Shire River had experienced many decades-long periods of diminished dry-season levels, even as recently as the early nineteenth century.⁴⁷ However, earlier ecological shocks had not generated a permanent redistribution of industrial labour hours to agriculture. This response was occasioned by the simultaneous decline in the Mang'anja population and the influence of the local institutional context. A change in the valley’s land-labour ratio altered “what was practicable from a purely agricultural point of view” and strongly influenced the choice to abandon industry in favour of export-oriented cultivation.⁴⁸ And for the Mang’anja, who knew from past experience that periods of river-level decline could last a lifetime, an economic transition centred around wetland exposure must have seemed a rational response to the altered circumstances.

Even before slave raiding reached a peak, Livingstone noted little spare labour in the area, where head porters were often unattainable “not because labour is dear, but because it is scarce”.⁴⁹ In decentralized Mang’anja villages labour scarcity was not shored up with slave-labour institutions, which had faded following the mid-eighteenth-century breakdown of centralized Lundu rule. Sena immigration began in the 1890s, but Mang’anja villagers did not typically enslave or employ immigrants, who settled in sparsely inhabited areas previously controlled by slave raiders.⁵⁰ Labour in the valley was thus confined to diminished village-community members, unlike in Mogadishu, Ethiopia and Kano, where slavery and other forms of servile labour were common and integrated into the textile industry. Ethiopia, for example, lost an estimated third of its population to severe famine and disease epidemics from 1888 to 1892, but this series of population-depleting events produced no fundamental reorientation of economic activity. In densely populated Ethiopia, population decline did not generate severe labour depletion, but areas that did experience shortages were sent labourers from newly conquered southern regions.⁵¹

The logic and impact of cash-crop production

As village life resumed in the Lower Shire Valley following the decline of raiding from the mid-1870s, Mang’anja households began introducing cash crops into their fields, particularly sesame oilseeds demanded in Europe to produce cooking oil.⁵² Mang’anja households were taking advantage of increasing agricultural production possibilities on expanding wetlands in an effort to maximize their sharply diminished labour resources by funnelling labour previously applied to labour-intensive industry toward agriculture.

An observer recalled how villagers “took [oilseeds] down, in their large dug-out canoes, to the Zambesi to sell to the Banian traders”.⁵³ Later, in the 1920s, the colonial-era railway better connected the valley to the world and “ensured that minimum prices there were always higher than those obtained in less fortunately situated […] areas further north”.⁵⁴ As villagers began to engage in export activity, Williamson’s theory of deindustrialization would suggest that the valley’s sectoral shift from industry to export-oriented agriculture was motivated by improved terms of trade as global demand for primary products increased.

However, oilseed terms of trade (relative to imported merekani cloth) had dropped in the years preceding the valley's turn to export-oriented agriculture (see Figure 2). Oilseed prices declined (see Figure 3), while imported cloth prices increased due to a global “cotton famine” precipitated by the American Civil War.⁵⁵ Although a subsequent decrease in cloth prices provided a modest boost for oilseed terms of trade in the second half of the 1870s, this upturn was soon met with a renewed pattern of decline in the 1880s as oilseed prices continued to fall. It was only during the 1890s, well after cloth production had already been largely abandoned, that oilseed terms of trade experienced a massive rise. Importantly, the terms of trade for oilseeds in the Lower Shire Valley were likely weaker than reflected in Figure 2 given that the calculation relies on coastal imported cloth prices. During the nineteenth century, prices for imported cloth generally remained much higher in the interior relative to the coast as a result of high transportation and transaction costs and a relative scarcity of imported cloth in interior markets that kept prices elevated.⁵⁶

Figure 2. Terms of trade for oilseeds, 1846–1910 (1902 = 100). Sources: Global oilseed prices: Sauerbeck “Prices” series, 1886–1917; cloth prices on the East African coast: Sheriff, Zanzibar, 253–6; MH23, MH235, MSS901, and MSS24 series (PEM); Statistical Abstract for the Several British Colonies, Possessions, and Protectorates in Each Year from 1891 to 1905, 110–11 (CBS); Statistical Abstracts for the Several British Oversea Dominions and Protectorates in Each Year from 1909 to 1923, 163–4 (CBS).

Figure 3. Oilseed price index, 1846–1910 (1902 = 100). Source: Sauerbeck “Prices” series, 1886–1917.

Why, then, did villagers turn to export-oriented sesame cultivation? Why not instead retool their cloth industry and adapt to compete with foreign imports in regional markets? Or, alternatively, disengage from export markets entirely as village life recommenced, resuming cloth and agricultural production, but only for subsistence? This paper suggests that in the face of altered factor endowments – particularly diminished labour – the transfer of labour from industry to agriculture was influenced by the relative labour demands of the two products at hand. Oilseeds are a single-stage product requiring limited labour investment, while cloth requires labour-intensive cotton cultivation and manufacturing (cleaning, spinning, weaving).⁵⁷ As Livingstone remarked before the disruption of production:

From the amount of native cotton cloth worn […] it is evident that a goodly number of busy hands and patient heads must be employed in the cultivation of cotton, and in the various slow processes through which it has to pass, before the web is finished.⁵⁸

Clothing the valley’s inhabitants required considerable labour. According to contemporary reports, a 2-man team worked four 4-hour days (32 man-hours) to weave a one-by-two yard cloth, or 16 hours to produce one square yard.⁵⁹ Roughly 112,500 yards were likely consumed annually in the valley based on early colonial-era consumption habits (5 yards per capita) and the population in 1895 (22,500).⁶⁰ Multiplying this by the hours required to weave one yard yields 1,800,000 annual hours. This excludes export-oriented cloth production and labour hours applied to “painfully slow” spinning.⁶¹ Deindustrialization freed this labour for agricultural activity. Indeed, a colonial-era observer noted, “There is no period of the year when the Lower River native is not occupied with […] his garden”.⁶² Further, labour reallocated from industry to agriculture was progressively more productive with the increasing availability of fertile marshland. Villagers could consume some (foreign) cloth – using cash-crop profits – without expending depleted labour on cloth-making, although imports remained low until cash-crop profits increased in the 1890s.⁶³

However, export-oriented agriculture could only arise if labour-scarce Mang’anja households could simultaneously achieve food security. Sesame was an ideal cash-crop choice, requiring relatively little labour and doubling as an emergency food source.⁶⁴ Tosh argues that cash-crop production could threaten food security in areas with short planting seasons, diverting resources from subsistence production.⁶⁵ But the extension of the valley’s wetland planting season mitigated this problem, while labour-saving intercropping techniques allowed villagers to cultivate oilseeds with a variety of subsistence crops in the same drought-resistant wetland fields.⁶⁶ The Mang’anja thus fully maximized their diminished labour supplies, producing export goods and ensuring greater food security than would have been possible had labour reverted back to its pre-1860s industry-agriculture division.

This transition affected gender and generational dynamics in Mang’anja households. With the disappearance of village-community industrial workgroups and the rise of household cash-cropping, spouses now shared in export-oriented output, working their wetland fields together. Agricultural tasks even became more gender-neutral, as male dryland bush-clearing responsibilities, for example, diminished in importance with the rising dominance of the wetland system. Spousal power relations consequently evened in monogamous households, as women claimed equal income control, regularly accompanying their husbands to market their product. However, controls over youth labour were extended, and non-agricultural economic activity was strongly discouraged. While this tack guaranteed labour, it restricted the range of possible Mang’anja economic activities and heightened cash-crop dependence.⁶⁷

The return of cotton

Oilseed exports and cloth imports surged in the mid-1890s, partly stimulated by global cloth prices, which in the case of merekani cloth fell by nearly 50% between 1880 and 1900, but also by emerging colonial policies.⁶⁸ In the early 1890s, a hut tax was implemented to raise revenue and generate wage labour for European plantations. The Mang’anja circumvented plantation labour by instead increasing household sesame cultivation to pay taxes and take advantage of trading opportunities.⁶⁹ A Nyasaland administrator complained that it was difficult to induce the cash-crop farmers to provide labour since they “find it so easy to obtain the small sum of money needed to pay hut tax”.⁷⁰

However, oilseed exports soon declined. Early-twentieth-century drought conditions forced villagers to focus on food security, and oilseeds were no longer “specially cultivated for export”, presumably retained for local consumption.⁷¹ A brief post-drought resurgence in oilseed exports was followed by a transition to increasingly more lucrative export-oriented cotton production in the 1920s, as the Mang’anja again reallocated available labour, this time from sesame to cotton cultivation (see Figure 4). Compared with stagnant, then falling oilseed prices, cotton prices rose precipitously (see Figure 5). Over time, as the valley’s economy became increasingly dependent on cash-crop exports, Mang’anja production choices had become more sensitive to global demand patterns.

Figure 4. Villager-grown sesame oilseed and lint cotton exports, 1897–1939. Sources: Sesame oilseed exports: British Central Africa/Nyasaland Blue Books (TNA); villager-grown lint cotton exports: Mandala, Work, 37 (Table 4.1).

Figure 5. Unit prices of sesame oilseeds and raw cotton, 1902–1938. Sources: British Central Africa/Nyasaland Blue Books, 1902–1938, TNA.

The initial disruption of cloth production had been linked with the sudden desertion of cotton fields, which, according to Livingstone, “every family of any importance own[ed] [and] carefully cultivated” before the slave raids.⁷² Cotton remained neglected for decades before making its dramatic comeback.⁷³ Whereas European settler-produced cotton failed in the Shire Highlands, abandoned in favour of tobacco, Mang'anja villagers and recent Sena immigrants succeeded in the valley. Cotton exporting dominated the valley economy until the late 1930s.⁷⁴ Cotton cultivation was encouraged by the British Cotton Growing Association, which distributed Egyptian and American Upland seeds. The colonial government, eager to secure cotton for British looms, also made efforts to guarantee prices at which “it pays to cultivate”.⁷⁵

However, the transition to export-oriented cotton cultivation was enabled by local environmental factors. By the early twentieth century, the extended dry season made it possible to cultivate cotton in fertile wetland (as opposed to dryland) fields prior to the annual floods.⁷⁶ Although cotton cultivation was more labour-intensive than sesame, yields of wetland-produced cotton were extraordinarily high: a one-acre dryland field yielded roughly 70 pounds of cotton, while one acre of wetlands yielded approximately 300 lbs. Furthermore, while the short native thonje-kaja species was traditionally grown in separate fields to avoid over-shading, taller American Upland and Egyptian cotton could be intercropped with food plants, dramatically diminishing weeding labour demands and food security limitations of cotton cultivation.⁷⁷ After a half-century hiatus, cotton had regained its position in the Mang’anja economy, but as a global export rather than a local industrial input.

A renaissance of local cloth?

In spite of the growing availability of cloth’s requisite raw material, Mang’anja villagers continued to focus their labour on agriculture, pursuing no renaissance of the local industry. According to Austin, cloth production traditionally faced an agricultural-season “labour bottleneck” that limited cotton supplies, as allocating more labour to cotton cultivation threatened food-crop output.⁷⁸ However, ecological changes and cultivation innovations in the valley had enhanced agricultural productivity, and food crops grew alongside unprecedentedly large cotton yields. Why did the reappearance of cotton in the Lower Shire Valley fail to stimulate yet another economic shift, with some labour reallocated back to industry to revive cloth production for home consumption and external trade?

To begin with, export-oriented cotton cultivation did not take off in the valley until nearly half a century after industrial collapse. There was no existing cloth industry to benefit from the resurgence of cotton cultivation. In much of western Africa, by contrast, cloth manufacturing had continued alongside the rise of cash-crop production, creating sectoral linkages, as cultivators provided raw materials, while global-export earnings stimulated demand for both imported and local cloth among prospering producers and traders.⁷⁹ In northeastern Africa, hide exports had generated increased wealth and demand, and while cloth imports were used to make trousers, local weavers continued to supply togas.⁸⁰ In the already-deindustrialized Lower Shire Valley, however, growing demand stimulated by increasing cash-crop profits was met exclusively by foreign cloth. In fact, the initial rise of cash-crop agriculture had essentially required the demise of industry in the context of scarce labour, thereby depriving the valley of the possibility of agricultural–industrial linkages enjoyed in more labour-rich regions.

The failure to resume cloth production and take advantage of potential sectoral linkages was influenced by increased opportunity costs of cloth production in the valley, particularly after environmental shifts had made possible profitable cotton exporting. An early-twentieth-century resumption of cloth production would have diminished export profits, with cotton diverted from global markets to local producers. As global demand pushed up cotton prices, while regional demand for Mang’anja cloth had long since evaporated, greater profits could be realized by allocating labour to increasingly productive cash-crop cultivation on fertile wetlands than back to labour-intensive industry.⁸¹ Furthermore, the staggered scheduling and increasing annual length of the valley’s dual-ecosystem agricultural system required male and female labour year-round, often in both dryland and wetland fields.⁸² There was scarce time for what had once been a lower-opportunity-cost industrial activity during the previously longer agricultural off-season.

Higher agricultural productivity levels relative to industry no doubt impacted early-twentieth-century production choices. Mang’anja cloth production had traditionally relied on hand-spinning and narrow ground looms.⁸³ More sophisticated labour-saving methods were not adopted before industrial decline for a number of possible reason. Firstly, time-consuming industrial labour had not impeded other vital activities during the longer agricultural off-season; secondly, cloth output had been limited by non-intercropped dryland cotton output, which competed with food crops; and thirdly, demand-stimulating cash-crop production had not yet developed.⁸⁴ The cloth industry had disappeared before the agricultural off-season shortened, before intercropped wetland-cultivated cotton overcame the old input bottleneck, and before cash-crop profits stimulated increased cloth demand, any of which could have inspired more efficient methods. In other parts of Africa, robust cloth industries served growing demand by achieving economies of scale through labour-saving innovations. These included, for example, large cloth-dying vats in northern Nigeria and spinning wheels and superior double-heddle looms in northeastern Africa.⁸⁵

Further, in these more resilient cloth-producing regions, slave-labour systems helped mitigate labour constraints, allowing industrial-input cultivation and local manufacturing to continue alongside food and export-crop production.⁸⁶ Internal use of slaves in western Africa had increased during the nineteenth century with the cessation of the trans-Atlantic slave trade.⁸⁷ By the mid-nineteenth century, plantation slave labour often supplied raw materials in places like the densely populated, centralized Sokoto Caliphate.⁸⁸ Similarly, on East Africa’s Benadir Coast, cloth production was supported by slave-produced cotton and dye products in the nineteenth century – later replaced by foreign-imported yarn – while rural plantations supplied food, freeing up labour hours for urban weavers.⁸⁹ In the Lower Shire Valley, by contrast, well-defined slave-labour institutions had disappeared before the nineteenth century.⁹⁰ In fact, as the valley’s inhabitants continued to apply as much household labour as possible to subsistence and cash-crop cultivation, colonial Nyasaland went through a consumer revolution, with demand for labour-saving finished items on the rise by the 1920s. Unfinished cloth imports into Nyasaland were progressively replaced with ready-made apparel, while in more diversified, labour-abundant economies, like Ethiopia, foreign and local cloth was often finished by professional tailors.⁹¹ Relatively high labour availability thus facilitated multi-sectoral economies in parts of northeastern and western Africa, while economic activities progressively narrowed in the labour-scarce Lower Shire Valley.

Evaluating global and local factors

The case of the Lower Shire Valley’s defunct cloth industry has highlighted the relationship between local and global factors in guiding development outcomes and shown that the role played by global forces was different and less decisive in the valley than deindustrialization theories suggest. When considered alongside other sub-Saharan African cases, this study illustrates how diverse local contexts within the “periphery” generated different responses to broad global processes. Only through comprehensive local-level studies can we begin to identify patterns and build more nuanced theories about the mechanisms, both global and local, that influenced economic change in Africa and beyond.

In general, the purportedly damaging role of foreign imports has been overemphasized in dependency theory arguments. Cloth imports only took off in the Lower Shire Valley decades after the process of deindustrialization was complete. Imported cloth may have circulated somewhat sooner in the Lower Zambezi region, a traditional Mang’anja cloth export base, but this was not a crucial mechanism in the deindustrialization of the valley. After all, cloth industries in western and northeastern Africa faced competition from foreign imports earlier and in greater quantities but competed very effectively with – and were even stimulated by – machine-made cloth. What of the suggested role of favourable agricultural terms of trade as a primary motivator for deindustrialization and cash-crop production theorized by Williamson? This is, again, an unconvincing assertion in the case of the Lower Shire Valley. Global terms of trade for oilseeds had actually declined substantially relative to cloth just before the valley abandoned cloth production in favour of export-oriented sesame cultivation in the second half of the nineteenth century, then showed significant volatility through the 1870s and 1880s. A substantial improvement in oilseed terms of trade occurred only after the domestic cloth industry had deteriorated.

Rather, the choice of Mang'anja villagers to permanently abandon cloth production can best be explained by shifting local conditions, particularly a sharp decline in labour immediately preceding an increase in fertile land. Within this altered local context, labour-intensive cloth production had simply become less feasible relative to agricultural production for subsistence and export. In the more densely populated regions of northeastern and western Africa, larger labour supplies and expansive internal markets allowed cloth production to thrive alongside cash-crop agriculture as rising incomes stimulated further demand for local and foreign cloth. The already large populations of Ethiopia, northern Nigeria and the Benadir Coast were further augmented by slaves, whose additional labour simultaneously provided industry-supporting inputs, food sources and export commodities.

The valley’s altered factor-endowment ratio was partly influenced by global forces, especially the growth in global demand for slave-produced commodities. But the ultimate impact was conditioned by the local context. In the valley, destructive and disruptive slave raiding interacted with multi-year drought conditions to produce a harsh famine that resulted in stark attenuation of the Mang’anja population. Northeastern Africa also experienced severe famine in the late-nineteenth century, but large populations and entrenched servile labour systems mitigated labour deficiencies in the period that followed. In fact, the centralized Ethiopian state sourced fresh labour from newly conquered regions. This was not possible in the decentralized, labour-scarce Lower Shire Valley of the mid-nineteenth century, where forced labour institutions had disappeared. When combined with rapidly intensified labor scarcity, the environmentally contingent emergence of rich, cultivable land tipped the valley’s factor-ratio balance in favour of export-oriented cultivation and deindustrialization. This illustrates how, as posited by Mandala, environmental conditions can influence social and economic outcomes within the parameters of the local institutional and historical context.

By the beginning of the colonial period, Mang’anja production choices became increasingly sensitive to global forces, as the valley’s economy had come to depend entirely on agricultural exporting. The adoption of oilseed production in the 1870s and 1880s had largely been a response to local circumstances, but the early-twentieth-century abandonment of oilseeds in favour of cotton exporting was in large part a response to global market opportunities. Even here, however, local environmental developments had made the transition possible. With respect to colonial forces, British institutional policies and agendas did not directly influence deindustrialization. However, certain colonial schemes did provide indirect incentives for villagers to continue focusing labour on household-based agricultural production for subsistence and export rather than working on European plantations or pursuing a reinvigoration of the cloth industry as raw cotton supplies increased in the twentieth century. However, the choice to forgo reindustrialization is also linked to local circumstances. Prior to deindustrialization, historically contingent local factors had likely dis-incentivized the adoption of more efficient cloth production techniques, which may have made a reverse reallocation of some labour from export-oriented agriculture to cloth production more feasible and profitable. More efficient methods had been developed to supply large populations in northern Nigeria, the Benadir Coast and Ethiopia, where cloth production thrived well into the twentieth century and benefitted from demand-enhancing sectoral linkages.

Household-based cash-crop agriculture remained the keystone of the Mang’anja economy until the valley’s factor-endowment ratio – and, thus, production possibilities – again changed. At the end of the 1930s, the cash-crop economy collapsed when the dambo wetlands became once again submerged as the Shire River suddenly returned to its pre-1870s annual dry-season level with the breakup of a large sandbar at the mouth of Lake Malawi. At nearly the same time, the Great Depression caused global cotton prices to plummet. Increased sensitivity of the valley’s economy to global market forces strongly intensified the effects of ecological change, and the impact was harsh and far-reaching. The now-undiversified valley economy was unable to absorb the local shock of diminished land and the global shock of falling prices, and many Mang’anja men became migrant labourers almost overnight. Formerly balanced gender power relations became skewed in favour of males as women lost their stake in raw cotton exporting and were forced to rely on male family members for economic assistance.⁹² Food security was regularly compromised as villagers now depended primarily on drought-sensitive drylands.⁹³ Thus, over the course of less than a century, the interaction of local and global forces had produced numerous substantial social and economic changes in the valley, including the decline of industry, the rise of export-oriented agriculture and global integration, and, eventually, the breakup of Mang’anja villages and households.

Acknowledgements

The author wishes to thank the editors and two anonymous reviewers of the Journal of Eastern African Studies along with Ewout Frankema, Elise van Nederveen Meerkerk, Kazuo Kobayashi, William Gervase Clarence-Smith, Gareth Austin, and Pieter Woltjer for their indispensable comments as this article developed.

Disclosure statement

No potential conflict of interest was reported by the author.

ORCID

Katharine Frederick http://orcid.org/0000-0001-8599-2624

Additional information

Funding

This research was supported by a Graduate Programme grant from the Netherlands Organisation for Scientific Research (NWO) awarded by the N. W. Posthumus Institute.

Notes

1. Also known as machira.

2. For production descriptions, see Livingstone, Narrative, 102, 123. For trade, see Pearson, Port Cities and Intruders, 122; Machado, “Networks,” 109; Isaacman, The Zambezi Prazos, 66, 73; and Alpers, Ivory, 55. Rowley, Story, 195; Mandala, Work, 48–9; Davison and Harries, “Cotton,” 187.

3. See Clarence-Smith, “Textile;” van Nederveen Meerkerk, “Java;” van der Eng, “Indonesia;” Roy, “Indian Weaving;” and Haynes, India.

4. In the context of this study, “industry” refers to handicraft production.

5. Williamson, Trade and Poverty, 27.

6. See Rodney, How Europe Underdeveloped Africa; Sheriff, Zanzibar; and Amin, “Underdevelopment.” For the impact of India’s cloth trade, see Palat and Wallerstein, “India,” 36–7. See also Kjekshus, 106, 109.

7. See Wrigley, Uganda; Freund and Shenton, “Vent-for-Surplus Theory.”

8. Denemark and Thomas, “Debate,” 49, 57–9 and Brenner, “Origins,” 91.

9. Mandala ventures that by the mid-nineteenth century, “Only those Mang’anja without easy access to foreign calico combined […] cotton and food cultivation on the mphala with dimba agriculture” and argues that “few branches [of the non-agricultural sector] survived the onslaught of foreign imports” with the cloth industry on the “verge of a total collapse” by the first years of the 1880s. More generally, he suggests that domestic production of pottery, salt, iron and cloth “declined as manufactured products penetrated rural markets” (on ecological factors: Mandala, Work, 8–11, 94–5, 270–72; for the impact of merchant capital and imports on domestic industry: Mandala, Work, pp. 293, 137 and 56, also 43–44, 92; Mandala, “Capitalism,” 147, 148; Mandala, “Cotton,” 30).

10. Krugman, Obstfeld, and Melitz, Economics, 80.

11. Tosh, “Lango,” 417 and Austin, “Resources,” 588.

12. Widgren, “Geography,” 3–4; Beinart, “Environmental History,” 269, 287; and Vansina, Paths, 255.

13. Livingstone, Narrative, 381 and Rowley, Story, 112. By the early 1880s, Morrison saw Mang’anja villagers using bark cloth if they could not obtain imported cotton cloth (cited in Mandala, Work, 92, 308n136). By the start of the colonial period (1891) officials commented on the absence of what was once “universal” cotton cultivation for local cloth production (Johnston, First Three Years and Duff, Nyasaland, 306).

14. The pre-colonial figures do not include smaller amounts of fancy cloth exported by Muscat and Germany and consumed primarily by coastal people and interior elites. Rigby, “Report,” 8 and Hardinge, Report for the Year 1896, 8.

15. Mandala, Work, 40 and Frederick, Limits.

16. Frederick, What Drove; Burton, “Lake Regions,” 429, 431; Portal, Report, 11; and Bureau of Manufactures, “German East Africa,” 777.

17. Imports into western Africa were probably still higher than depicted in Figure 1 as these regional statistics only include British cloth imports.

18. Alpers, “Benaadir,” 79–97.

19. Rigby, “Report,” 9.

20. Mandala, Work, 41. James Stewart alleged that Livingstone had overstated the valley’s cotton culture. Mandala notes that Livingstone had motives to present the valley’s cotton cultivation in a positive light but also points out that Stewart visited later, in 1862–63, when productivity-stalling famine and slave raiding had ravaged the region (Mandala, Work, 41–3, 293–4). John Kirk, a botanist accompanying the Livingstone expedition, gave a detailed description of regional cotton cultivation from 1858–60 that supports Livingstone’s favorable account. He reported that in the Lower Zambezi region, where already “the slave trade and war ha[d] combined to desolate this rich country,” only some wild cotton was found growing. But in the Lower Shire Valley, which was at that time less affected, he reported extensive cotton cultivation for the “manufacture of cloths [which] all engage in […] from the chief to the poor people” (John Kirk, “Report,” 25–30).

21. Mandala, “Capitalism,” 139–42.

22. For the “disaster economy” concept, see Johnson, “Cotton Imperialism,” 180. In the north of the valley, villagers sought protection from armed Makololo immigrants, who subsequently established authoritarian control. This article focuses on the independent south, which retained autonomy.

23. Livingstone, Narrative, 381.

24. Ibid. See also Rowley, Story, 112.

25. Mandala, Work, 7, 44, 76–8.

26. Livingstone, Narrative, 450; see also Foskett, Zambesi, 500 and Rowley, Story, 363–84.

27. Rigby, “Report,” 9.

28. See Rowley, Story, 384 and Anderson-Morshead, Mission, 39.

29. Livingstone, Narrative, 471.

30. Ibid., 202, 206, 407, 473, 593–4; Rowley, Story, 92–3 and McCracken, Malawi, 32.

31. According to Mandala, in the Lower Shire Valley, “Male and female labor prevailed in different stages of cloth-making. Women spun cotton, and men wove it into cloth” (Mandala, “Capitalism,” 141). Elsewhere, however, he reports that male labor dominated in all aspects of cloth production with occasional female assistance (Mandala, Work, 41; Livingstone, Narrative, 112).

32. Mandala, Work, 47–8.

33. Lovejoy, Transformations, 229.

34. Vaughan, “Food Production,” 355–7.

35. Alpers, “Benaadir,” 80, 87–8.

36. See van Nederveen Meerkerk, “Java;” van der Eng, “Indonesia;” Haynes, India; and Roy, “Indian Weaving.”

37. Alpers, Ivory, 25.

38. Singer, “Demographic,” 253–4.

39. Johnson, “Technology,” 267.

40. Nicholson, “Fluctuations,” 227 (Appendix A). Mandala, Work, 6-7.

41. Murray, Nyasaland, 67.

42. Livingstone, Narrative, 78 and Mandala, “Capitalism,” 154.

43. Rowley noted flooding from late October to early May in the 1860s, but 20 years later flood waters only persisted from late December to February. Sieger reported water-level decline from the mid-1870s, and Kerr witnessed increasing marshland exposure. See Rowley, Story, 62 and Mandala, Work, 7; Sieger, “Schwankungen Der Innerafrikanischen Seen;” Kerr, Interior; and Nicholson, “Fluctuations.”

44. Shire water levels are affected by rainfall patterns in southern Tanzania. Nicholson, “Fluctuations,” 218–22.

45. The once-per-annum labour required to clear freshly exposed marshlands now facilitated two planting cycles. For the three-season system, see Mandala, “Cotton,” 27 and Mandala, Work, 7, 58–9, 94–5.

46. Mandala, Work, 93.

47. Nicholson, “Climatology,” 258 and Nicholson, “Fluctuations,” 214–6.

48. Tosh, “Revolution,” 83.

49. Livingstone, Narrative, 472.

50. A limited system of servitude (ukapolo) did exist in the valley prior to the slave raids of the 1860s but largely disappeared in the post-slave-raid period. Akapolo laborers were traditionally not engaged in industrial branches of the economy (Mandala, Work, 32–6, 97). On the region of Sena settlement, see Mandala, Work, 95; Rowley, Story, 62–3.

51. Pankhurst and Johnson, “Famine,” 54–6. In fact, Ethiopia may have been experiencing over-population prior to the 1880s.

52. Sheriff, Zanzibar, 71, 134 and Mandala, Work, 93.

53. Murray, Nyasaland, 44.

54. McCracken, Malawi, 198 and Murray, Nyasaland, 71.

55. Farnie, Cotton, 135–70.

56. For example, in nineteenth-century Tanzania, this dynamic significantly depressed the terms of trade for ivory (relative to imported cloth) in the interior from which the valuable export came, while far more favorable terms of trade surged on the coast, primarily enjoyed by the coastal merchants who ultimately exported the ivory abroad. Frederick, Limits.

57. Tosh, “Lango,” 428 and Mandala, Work, 55. Dryland-produced cotton demanded considerable weeding labour since the short native thonje-kaja cotton could not be intercropped as it would be overshadowed by taller food crops. Harvest was also time-consuming because thonje-kaja adhered tightly to its seeds. Labour-intensity later declined when taller foreign cotton species were introduced as cash crops and intercropped in marshland fields (discussion below).

58. Livingstone, Narrative, 397.

59. Wallis, Expedition, 155.

60. See Figure 1; Mandala, Work, 95, 309.

61. Livingstone, Narrative, 112.

62. Port Herald District Annual Report, 1932 as quoted in Mandala, Work, 134.

63. Johnston, First Three Years.

64. Similarly, the Langi in Uganda initially produced cash-crop oilseeds partly because they could serve as a food source. Later, when production of non-edible cash crops took off in the Lower Shire Valley, villagers often neglected export-oriented crops to focus on food production. Tosh, “Revolution,” 85–6, 428; Nyasaland, 1907–8, 12; Nyasaland, 1908–9, 9; and Nyasaland, 1919–20, 7.

65. Tosh, “Revolution,” 85–6.

66. Intercropping significantly reduces labour inputs relative to sole cropping (i.e. growing different crops in separate fields) since weeding is performed in a single operation. Intercropped plants are also less vulnerable to disease due to greater ecological diversity, increasing overall yield potential. See Richards, Indigenous, 66–9. For intercropped subsistence crops planted in the valley, see Livingstone, Narrative, 111 and Murray, Nyasaland, 70.

67. Davison, “Women,” 410 and Mandala, “Cotton,” 30–1.

68. McCracken, Malawi, 88; see cloth-price sources for Figure 2.

69. Mandala, Work, 94, 111–32.

70. Sharpe, Report, 13–4.

71. Mandala, Work, 176 and British Central Africa Protectorate, 1906–7, 5.

72. Livingstone, Narrative, 111.

73. Johnston, First Three Years.

74. Green, Highlands, 11, 21 and McCracken, Malawi, 90, 91.

75. British Central Africa Protectorate, 1904–5, 10 and British Central Africa Protectorate, 1905–6, 13. Mandala, Work, 134–140.

76. Mandala, Work, 7.

77. Mandala, Work, 55, 135, 139.

78. Austin, “Resources,” 597–8, 603.

79. E.g. in Nigeria, Ghana and Senegal. See Austin, “Manufacturing,” 209–13 and Kriger, Cloth, 45–7.

80. Pankhurst, Ethiopia, 261.

81. As a counter-example, diminished agricultural opportunities in Java – caused by land constraints and declining agricultural prices – encouraged a 1930s resurgence of labour-intensive cloth-making. See van Nederveen Meerkerk, “Java,” 25–6.

82. Mandala, Work, 94–5 and Mandala, “Capitalism,” 149.

83. Livingstone, Narrative, 102, 123.

84. This reasoning is inspired by Austin, “Resources,” 603.

85. Johnson, “Technology,” 260; Austin, “Manufacturing,” 206–7; and Alpers, “Benaadir,” 81.

86. Austin, “Manufacturing,” 206.

87. Lovejoy, Transformations, 149.

88. Lovejoy, “Plantations,” 342–7, 356–9.

89. Casanelli, “Somaliland;” Alpers, “Benaadir,” 83, 85, 90; Clarence-Smith, “Textile,” 267; and Sheriff, Zanzibar, 71–2.

90. Mandala, Work, 97.

91. Nyasaland, 1927, 12 and Pankhurst, Ethiopia, 261–2.

92. Davison, “Women,” 411 and Duly, “Notes.”

93. Mandala, Work, 7, 177.