THE FOOD ECONOMIES OF URBAN
The Case of Ghana
THOMAS T. POLEMAN
Reprinted from Food Research Institute Studies, Vol. II, No. 2, May, 1961
FOOD RESEARCH INSTITUTE
FOOD RESEARCH INSTITUTE
Established at Stanford University, Stanford, California, in 1')2],
jointly by Carnegie Corporation of New York and the Trustees o' the
Leland Stanford Junior University, for research in the product ion,
distribution, and consumption of food
MERRILL K. BENNETT
JOSEPH S. DAVIS
HELEN C. FARNSWOI rH
HOLBROOK WORKING :
Associate Director Emn ritus
VLADIMIR P. TIMOSIIENKO
WILLIAM O. JONES
ROGER W. GRAY
RICHARD J. HAMMOND
BRUCE F. JOHNSTON
S. DANIEL NEUMARK
E. LOUISE PEFFER
BORIS C. STERLING
VERNON D. WICKIZI R
ARNOLD B. LARSON
Acting Assistant Econo 'nit
CII RULESS O. MEIBUR
TI IOM.\S T. POLEM \ ,
ROSAMOND H. PEIRCE
P. STANLEY KING
Cartographer and Editorial Assistant
Transformation of the fragmented and heterogeneous societies
of tropical Africa into national states that are supported by modern market-
oriented economies will inevitably be accompanied by rapid increases in urban
populations and in the demands that cities make on the rural societies. In ap-
praising the impact of urbanization on African economic growth, knowledge of
the technical capabilities and economic responsiveness of African food farmers
is essential, and both have been the object of several recent studies. Equally im-
portant is better knowledge about the character of urban demand itself, about
the changes in demand that are likely to accompany rising incomes, and about
the capacity of marketing systems now supplying African cities. Thomas T.
Poleman's study is an attempt to provide this sort of information for the cities
When this study was originally undertaken it was hoped that it might em-
brace all those cities in Middle Africa for which budget and consumption studies
have been published. It soon became obvious, however, that best results could
be had by careful detailed study of all information available for a very few cities.
Quantitative information about most aspects of African economic life is notori-
ously defective; it is no less so for African cities. But it is not entirely lacking,
and many African governments are making serious efforts to repair the de-
ficiencies. Mr. Poleman demonstrates that when existing statistics are handled
with care and supplemented by nonquantitative sources they can be made to
yield a great deal of information about African economic organization. But to
obtain the greatest possible amount of juice from the statistical grape requires
Intensive consideration of expenditure studies conducted in the Ghana cities
raises important questions. It suggests that formulations of the relationship be-
tween income and consumption based on observation of western societies are
inadequate when applied to African societies either because the relationships
themselves differ or because pertinent variables are neglected.
We hope that Mr. Poleman's study will encourage others to conduct similar
intensive investigations of other African cities for which comparable data exist.
In particular, it is important to learn whether the consumption responses re-
ported in Ghana are found elsewhere and to identify more precisely the deter-
minants of these responses, if we are to say more about urban demands on the
developing African economies.
W. o. J.
THOMAS T. POLEMAN
THE FOOD ECONOMIES OF URBAN
THE CASE OF GHANA*
One of the most striking features in the recent course of Middle
African development has been the growth of urban centers. Encouraged by
great improvements in communications, enlarged investment, and the increas-
ingly commercialized character of African life, many cities here have doubled,
tripled, or even quadrupled in size since World War II. This is not to suggest
that Middle Africa is as yet anything like an urbanized region. At midcentury
this vast area was thought to contain only 108 cities with populations in excess
of 20,000, and these centers together accounted for only about five per cent of
the region's 122 million people (6, pp. 131-32, Map 41, Table D). But, as a com-
parison of the prewar and postwar population estimates given in Table 1 makes
clear, a strong movement of people to the cities is definitely under way, and it is
Urbanization within the African milieu poses a multitude of problems, not
the least of which relates to food supplies. African food farming is still an ex-
tremely primitive operation. Until very recently excluded from the assistance
programs of most governments, it has changed but slightly with time. The hoe
persists as the principal implement, and shifting cultivation is still almost ex-
clusively relied upon for fertility maintenance. To such a system, obviously, a
growing population of town dwellers presents a major challenge. Not only will
the needs of an increasing number of nonself-suppliers have to be met by a re-
The present study is one of several prepared under the sponsorship of the Human Environ-
ments in Middle Africa project of the National Academy of Sciences-National Research Council.
Following the definition agreed upon by the senior staff of that project, "Middle Africa" is taken here
to mean that portion of the continent which has as its southern limit the southern boundaries of
Tanganyika and the Congo (excluding Katanga) and which extends north to include Ethiopia, the
Sudan, and all of the former French West African territories save Mauritania. So defined, it refers
to that portion of tropical Africa which is still primarily agricultural, most of the countries or regions
with major mining or industrial sectors having been excluded.
The author's thanks are due Patricia L. Cedarleaf and P. Stanley King for preparation of the
charts and maps, and W. O. Jones, Helen C. Farnsworth, and Rosamond H. Peirce for the suggestions
which grew out of their critical leading of the first draft. Mr. Jones was also kind enough to undertake
the final editing and see the manuscript through the press. Finally, grateful acknowledgment is made
to Carnegie Corporation of New York for a grant of funds to the Food Research Institute which
made publication of this study possible. The Corporation is not, however, the publisher or pro-
prietor of this publication and is not to be understood as approving by virtue of its grant any of the
statements made or views expressed herein. For these the author is solely responsible.
122 THOMAS T. POLEMAN
TABLE 1.-MIDDLE AFRICA: PREWAR AND POSTWAR POPULATION
OF SELECTED CITIES*
City Territory Prewara F stara
Abidjan Ivory Coast .......... 10 (1931) 12,; (1955)
Addis-Ababa Ethiopia ............ 150 (1939) 400 (1952)
Bangui Ubangi-Shari ........ 14 (1937) 8'; (1957)
Brazzaville Middle Congo ....... 40 (1937) 9" (1957)
Dakar Senegal ............. 93 (1936) 23 (1955)
Dar-es-Salaam Tanganyika ......... 33 (1931) 9'1 (1952)
Freetown Sierra Leone ......... 55 (1931) 7' (1956)
Ibadan Nigeria ............. 387 (1931) 451 (1952)
Kano Nigeria ............. 89 (1931) 131 (1952)
Khartoum Sudan .............. 45 (1938) 95 (1956)
Lagos Nigeria ............. 126 (1931) 31! (1956)
Leopoldville Belgian Congo ....... 36 (1938) 301 (1955)
Monrovia Liberia .............. 10 (1938) 41 (1956)
Data for postwar years from United Nations (UN), Dept. Econ. and Soc. Af:., Demographic
Yearbook, 1957 (1957), pp. 150-51, except for Bangui for which estimate is from Fi bench Equatorial
Africa, High Commis., L'A.E.F. Economique et Sociale 1947-1958 (Brazzaville, 1 5)), p. 8. Pre-
war data from Walter Yust, ed., Encyclopaedia Btitannica World Atlas (prepared by G. D. Hudson,
Chicago, 1942), pp. 182-211, except for Abidjan which is from B. F. Johnston, 7 he Staple Food
Economies of Western Tropical Africa (Stanford, Calif., 1958), p. 5. As most Af ican population
estimates are notoriously unreliable, the foregoing statistics must be accepted only 'ith reservation.
a Figures in parentheses are the year of census or estimate.
duced number of growers, and the means adapted through which fc dstuffs may
be moved considerable distances at reasonable costs: at the same time per capital
demand will in all likelihood increase. One of the familiar features of economic
development is a change in dietary patterns (3, pp. 224-25). As income levels
in the cities rise, more and more people will be able to afford foodstt Ifs in greater
quantities and of a "preferred," more expensive type.
Whether this challenge can be met without sizable imports fr )m more ad-
vanced regions is a much debated question. Though several closely documented
studies have recently suggested that African agriculture is much more viable
than it was once thought to be (cf. 27, 28), the consensus remains r one too opti-
mistic. Indeed, the postwar period has witnessed a rash of almost pa' icky "Grow-
More-Food" campaigns and price-fixing schemes intended to stim late produc-
As an example of this general feeling of apprehension, one may cite the
reaction to a series of sharp increases in urban food prices which occurred about
a decade ago in Ghana,1 the country with which we shall be co icerned here.
Though at least partially cognizant of the special circumstances w)ich underlay
these rises, one party of visiting economists interpreted them as clear evidence
that "the production of food for market has not increased fast enough to meet
the enormous increases in demand associated with growing population, rising
1Though Ghana did not become an independent state and adopt that name until March 1957,
I shall not labor the point by calling it the Gold Coast in references to the period wh 'n it was actually
GHANA'S URBAN FOOD ECONOMY 123
CHART 1.-GHANA: INDEX OF RETAIL PRICES OF LOCALLY PRODUCED
FOODSTUFFS IN SEVEN CITIES, AND PRODUCER PRICE
OF COCOA, 1948-1958*
(Av. 1948 = 100; shillings per 60-lb. load; logarithmic vertical scale)
ITS ------ --- -- -- ---- ocoa--" 70
,. 150 \60 J
Cc2 I ~ Foodstuffs
125 1 50
100 I-I 40
1948 1950 1952 1954 1956 1958
Retail price index is a composite of indices for Accra, Kumasi, Sekondi-Takoradi, Tarkwa,
Tamale, Keta, and Ho weighted according to their populations in 1948. In 1954 a new index using
a revised market basket was begun with that year as the base period; here it has been related to the
older one using 1954 as the year of comparison. Quarterly data for 1949-51 from Dudley Seers and
C. R. Ross, Report on Financial and Physical Problems of Development in the Gold Coast (Accra,
1952), p. 23; for 1952 from Gold Coast, Min. Fin., Economic Survey, 1953 (1954), p. 19; for 1953-
58 from Gold Coast (or Ghana), Off. Govt. Stat., (Quarterly) Digest of Statistics (various issues).
Cocoa prices are fixed prices paid by the Cocoa Marketing Board for a 60-lb. load of Grade I and
II dry beans at collecting stations during the September-March primary crop buying season. 1947/48-
1954/55 prices from Gold Coast, Dept. Agr., Annual Report . (various issues); 1955/56-1958/59
prices from Ghana, Min. Fin., Economic Survey (various issues).
money incomes and the rapid growth in the number of people working in the
towns" (38, p. 23).
Food prices in Ghana are closely related to the price paid producers for cocoa.
This relationship, illustrated graphically in Chart 1, exists for several reasons,
the most important being the dominant role cocoa plays in the economy.2 Regu-
larly accounting for over two-thirds of the country's exports, it is by far the most
important source of Ghana's wealth. If cocoa prices are high there is prosperity,
if they are low, depression; and the prices of other items, foodstuffs included,
respond accordingly. Indeed, it is reported that in many parts of the country the
cocoa price is used directly as a yardstick, other prices being adjusted immediately
following announcement of the price to be paid by the Cocoa Marketing Board
(38, p. 23). Thus it was in 1948/49, albeit to an unprecedented extent. In that
year the Board announced that the price to producers would be increased from
2 Involved also is the fact that, unlike their Nigerian opposite numbers, a majority of Ghana's
cocoa farmers grow most of their own food (27, p. 16). In years when cocoa prices are particularly
favorable, however, they tend to slight their food plots, preferring instead to buy for cash-thereby
both heightening demand and reducing supplies. This happened in both 1948/49 and 1950/51 (38,
THOMAS T. POLEMAN
40 shillings per 60-pound load to 65 shillings. Food prices quickly fo lowed suit,
rising according to the newly devised urban food price index by 15 per cent
during the year. Although the following season witnessed a slight c ecline-the
cocoa price having been reduced to 45 shillings-the same chain of events was
repeated in 1950/51. Following an increase in the cocoa price to 70 sh killings (the
effects of which were reinforced by an attack of rust which severely d imaged the
maize crop) food prices rose again, this time climbing to a level alr ost double
that which prevailed in 1948.
Coming so quickly in succession, these two steep rises in the p ices urban
consumers were compelled to pay for their foodstuffs created a great deal of
official as well as private anxiety, particularly as they occurred at a time when
the authorities were already faced with considerable civil unrest it the cities.
Especially alarmed was the Department of Agriculture, which voic( d n its re-
port for 1949/50 the fear "that if something were not done the cou itry would
be faced with a chronic food shortage and greatly inflated prices" (15, 3. 3).
Such extreme fears would seem to have been at least temporarily allayed by
the situation since 1950/51. As Chart 1 indicates, urban food prices have been
reasonably stable (save for one general rise following an increase in the cocoa
price in 1955/56), while the Department of Agriculture has repeatedly expressed
its satisfaction with the over-all level of food production (cf. 11, p. 3). Nonethe-
less, gloomy inferences continue to be drawn from the experiences of 1948/49
and 1950/51. Writing in 1954, for instance, F. J. Pedler remarked cn the wide
seasonal variations in the prices individual foodstuffs commanded n Ghana's
cities in 1950/51 (35, p. 51), noting among others the following ( n shillings
Foodstuff Accra Kunasi Sekond -T; koradi
Max. Min. Av. Max. Min. Av. Max. \lir. Av.
Maize ................ 81 26 50 62 25 43 89 27 54
Rice .................. 89 72 82 84 72 80 95 76 86
Yams ................ 50 32 40 36 19 27 -
Manioc ............... 22 9 15 8 4 6 10 6 7
Cocoyams ............. 37 23 31 23 13 19 31 21 25
Such sharp variations, he concluded, could only be "an indication th it :oo little
food is produced to meet the needs of the people through the year. Tr: ders know
this, and they buy food at harvest time to hold it until prices rise. The v are often
blamed for high prices and scarcity, but their action is the result of shortage, not
the cause of it" (35, p. 51).
It is difficult either to defend or refute reasoning such as this witi certainty.
In many respects Africa is still the Dark Continent; the economist, Fparticularly,
is confronted with an extreme paucity of evidence on which to app y the tools
of his trade. About the price elasticity of food production, so vital t, t ie long-
run question of urban provisionment, for instance, next to nothing is known;
and so it is throughout most of the economic spectrum. Given such i situation,
manifestly, the first steps forward must hinge not on unsupported though in-
cisive opinion, but rather on exploratory descriptive outlines. It wa; with this
in mind that the present study was undertaken. How wide an area lo African
GHANA'S URBAN FOOD ECONOMY
cities presently draw on for their food supplies? What is the make-up of urban
diets? And what changes in consumption patterns occur seasonally and with
rises in income? Light must be shed on these and related questions before rea-
soned attempts are made to see into the future.
The choice of Ghana for detailed study warrants explanation. Inquiries into
the economics of Middle Africa have typically been of two general types: macro-
studies dealing with a particular economic problem on a continental basis, and
microstudies examining economic phenomena in so far as they bear on the social
structure of a particular tribe or people. The present inquiry was originally con-
ceived as being of the first type. A thorough review of the available evidence
soon suggested, however, that the time was not yet ripe. Although food-con-
sumption sample surveys are now available for a sizable number of urban cen-
ters,3 complementary evidence on the supply side remains excessively sketchy.
Apart from Ghana, the only country for which any reasonably complete picture
of supply characteristics can be built up is Nigeria, and here one is obliged to
depend chiefly on railroad statistics, which throw light only on long-distance
movements. Favoring Ghana, moreover, was the number of consumption sur-
veys available for its cities. Surveys of this type, and by African standards good
ones, have been carried out in all three of its principal cities, a situation equaled
elsewhere in Middle Africa only in Senegal and Uganda.
The study falls into three main sections. The food-consumption character-
istics of Ghana's urban dwellers form the subject matter of the second of these;
the supply side is taken up in the third. Preceding them is a brief description
of Ghana, its agricultural resources, and its chief cities.
Economically, Ghana ranks among the more advanced countries of Middle
Africa; in fact, it is probably the most advanced if territories with substantial
European settlement are excluded from consideration. Far and away the world's
largest producer and exporter of cocoa (a position held since 1913), it is also the
home of a major mining industry. Gold has been mined by European concerns
for over 80 years, while more recently the country has become one of the prin-
cipal suppliers of industrial diamonds and manganese. All this has made for a
relatively high level of investment, both private and public, and a fairly high
level of living. The stock of overhead capital facilities, such as railways and
roads, is among the most complete on the continent, while, according to esti-
mates published by the United Nations, gross output per capital (the equivalent
of $194 [U.S.] in 1957) is exceeded in subsaharan Africa only in the Union of
South Africa (42, p. 15).
This is not to say, of course, that Ghana has as yet passed beyond the first
stages on the path to sustained economic growth. Despite impressive gains in
the extractive and mercantile sectors, the economy remains predominantly agri-
cultural. In 1948, the most recent year for which complete census statistics are
available, almost three-quarters of the adult male population still depended di-
3 Namely: Abidjan, Accra, Dakar, Dar-es-Salaam, Enugu, Freetown, Ibadan, Jinja, Kaduna,
Kampala, Kumasi, Lagos, Leopoldville, Nairobi, Saint-Louis, Sekondi-Takoradi, Thies, and Zaria.
THOMAS T. POLEMAN
rectly and exclusively on farming for its livelihood (18, p. 370). Moreover, a
substantial fraction of the peasantry is still made up of subsistence growers par-
ticipating little, if at all, in the national economy. Development and underde-
velopment are relative things. Although in the forefront as far as tropical Africa
is concerned, Ghana's economy is nonetheless primitive indeed.
Like Caesar's Gaul, Ghana is divided into three parts-three major and
sharply contrasting natural regions which reflect in turn difference; in vegeta-
tion and in climate (Map 1). They are the Northern Savanna, the Fcrest, and
the Coastal Savanna. Of these, the largest by far is the Northern Sa-.anna, com-
prising about two-thirds of the country's area. Rainfall here is rela ively low-
averaging almost everywhere less than 50 inches-and is restricted tc one clearly
defined rainy season (March-October). During the remaining fou- months of
the year the hot, desiccating harmattan winds from the northeast prevail and
the region turns exceedingly dry. As a consequence of this brief but intense
aridity, the natural cover is typically a more or less continuous carpet of short
grasses with occasional stunted trees intermixed.
The Northern Savanna is the least advanced section of Ghana. Ilt is also the
least densely populated; with something like 1.8 million inhabitant: (according
to provisional findings of the 1960 census), the zone contains only lightly over
a quarter of the country's 6.7 million people (13, p. 1). This low density pri-
marily mirrors a poverty of soil resources. Most of the area falls within the
Voltaian Basin, a rather featureless depression underlaid by porous sandstone.
The soils derived from this material are generally shallow and infertile and have
little moisture-retaining capacity. During the harmattan season the) tend to dry
out rapidly and form a hardpan, whereas during the rainy period they are subject
to heavy surface runoff. It is in the area of these soils that populat on densities
are the lowest, averaging for the most part less than 15 persons per square mile.
Only in the peripheral plateau sections of the Northern Savanna, \ here soils of
granitic origin prevail, are densities comparable to those found to the south. The
extreme example is the northern portion of Mamprusi District. Here are con-
centrated almost half the inhabitants of the Northern Region,5 tEe density in
places exceeding 200 per square mile.
SIn writing this section I have relied heavily on the two recent geographies ol Ghana: W. J.
Varley and H. P. White, The Geography of Ghana (London, 1958); and E. A. Bciteng, A Geog-
raphy of Ghana (Cambridge, 1959).
So many changes have taken place in the last several years with respect to the names and
boundaries of Ghana's administrative regions that it is necessary that the various on, s recognized in
this article be clearly identified. Under British rule the Gold Coast was originally nrade up of three
divisions: the Gold Coast Colony, Ashanti, and the Northern Territories. Following the first World
War Togoland was added as a League of Nations mandate, the southern portion bei ig administered
through the Colony, the northern section through the Northern Territories. Except for the forma-
tion of Trans-Volta/Togoland in 1952, this structure remained unchanged until :he country re-
ceived its independence in March 1957. Under the new constitution five regions v ere recognized:
the Northern Region (the Northern Territories plus the northern half of Togolanc Ashanti (re-
taining its former boundaries); the Western Region (the western half of the Coloi v): the Eastern
Region (the eastern portion of the Colony save that part which lay east of the 7olla); and the
Trans-Volta/Togoland Region (southern Togoland plus the section of the Colony ea t of the Volta).
The limits of these five regions were as shown in Map 1.
This structure lasted only two years. By the end of 1958 the constitution of 19'57 had been
heavily amended and the Government's "federalist" opponents almost completely si ienced. In April
1959 the northern and western portions of Ashanti, their former stronghold, weie detached (to-
GHANA'S URBAN FOOD ECONOMY
MAP 1.-GHANA: PRINCIPAL FEATURES*
.. .1 N .O E R.N...
S A V-A N
Oe dt 2.000 p.opl.
--LnLt of ~b un
VoLtnian basin II .o ,, x
Urban popl to
0 25 50 T
PRicip.1 IWt-cla ..o.d-
Rainfall, natural regions, and communications adapted from E. A. Boatcng, A Geography of
Ghana (Cambridge, 1959), pp. 33, 34, 63, 119; population distribution after W. J. Varley and H. P.
White, The Geography of Ghana (London, 1958), p. 274.
THOMAS T. POLEMAN
Agriculture in the Northern Savanna is characterized by the rearing of live-
stock and the cultivation of those annual crops which can either mature before
the onset of the dry season or are relatively resistant to it. In the zon 's northern
portions this almost exclusively implies the millets and sorghums. I' is difficult
to overemphasize the importance of these grains here; a detailed surv :y of farms
in northern Mamprusi revealed that over 90 per cent of the cultivated land was
planted to them (30, p. 22). As one moves south, and rainfall beco nes higher
and more evenly distributed, other crops begin to appear, notably y; ms, maize,
and peanuts. South of the Volta the predominance of the millets and sorghums
gives way altogether, their place being taken by yams, maize, and, to i lesser but
increasing extent, manioc. Yam production is especially intense alo ig the bor-
ders of the Forest zone, this more densely covered transition belt of er ng near
ideal conditions for the crop.
With few exceptions, farming in the Northern Savanna is a subsi tence oper-
ation; and where movements of foodstuffs take place they are chiefly to satisfy
local needs. The only exports of consequence to the more populou a id pros-
perous southern parts of the country are livestock and yams. Cattle :ome prin-
cipally from the northern portions of the zone (which adjoin the rich cattle area
of Upper Volta), while the southern portions provide most of the 'arns. Par-
ticularly around Wenchi, Ejura, and Kete-Krachi the production cf yams for
sale is pushed vigorously. As we shall see later, a large fraction of the y ms con-
sumed in the cities of the south comes from these regions.
Succeeding the Northern Savanna to the south is the second of G1 ana's three
major natural regions, the Forest zone. This is the principal source cf the coun-
try's wealth. Here is grown the cocoa-about one-third of the world s total out-
put-which so dominates the Ghanaian economy. And here too are Iound most
of the mineral resources: gold, around Tarkwa and Obuasi; diam nds in the
vicinity of Akwatia; and manganese, again in the neighborhood (f Tarkwa.
Over half of the population-about 3.6 million people in 1960-is cc ncentrated
in this zone (13, p. 1). Unlike the Northern Savanna, this region i; relatively
well provided with both rail and road facilities.
The Forest is a zone of high rainfall. Annual precipitation averages between
50 and 70 inches over most of the area, with portions of the southwest receiving
more than 80 inches. The seasonal distribution is that of the intertr )pical con-
vergence zone: two rainy periods (March-July and September-Nove nber) sep-
arated by brief dry seasons. In its topographic aspect the area is cor paratively
high and broken, its northern boundary being a line of hills (the Mlmpong
scarp in Ashanti, the Akwapim-Togo ranges to the east) rising abou: 1.500 feet
above the Voltaian Basin. This surface configuration has an importar t influence
on the region's climate. Acting as a barrier against the harmattan w nds which
gether with a small segment of the Northern Region) and formed into the Brong-A hafo Region.
Further changes were made effective July 1960: the Western Region was divided int > the Central
and Western Regions: the Northern Region into the Northern and Upper Regions; and Tr: ns-Volta/
Togoland was renamed the Volta Region.
In this study the various regions are referred to as they were constituted in 1957 ar : 1958. I do
this mainly because the data collected in the 1957/58 Produce Movement Census (on slhii:h I draw
heavily in the final part of the paper) had this structure as their basis. To do other 'c would re-
sult in a superabundance of footnotes and no small confusion for the reader.
GHANA'S URBAN FOOD ECONOMY
sweep down from the north, it sharply limits the intensity of the dry seasons and
has favored the growth of storied, semideciduous forest. (Man's action has, how-
ever, reduced much of this cover to low forest and secondary bush.)
As in the Savanna to the north, the production of food crops in the Forest is
characterized by shifting cultivation. Clearing and burning prepare the land for
cropping, and a given plot is exploited on a temporary basis. After only one,
two, or, at the most, three years, a combination of weed intrusion and fertility
exhaustion reduces yields to an uneconomic level and forces abandonment; a
new plot is then cleared, while the previous one is returned to "bush fallow."
But beyond adherence to this basic system there is little similarity in the agri-
culture of the two zones. In the Northern Savanna production is closely geared
to the seasons, with planting taking place at the beginning of the rains each
spring and harvesting being done in November and December. In the Forest,
on the other hand, year-round cultivation is practicable. Accordingly, a wide
variety of perennials as well as annuals can be grown: cocoa, plantains, bananas,
kolas, oil palms, manioc, cocoyams, maize, rice, and yams, to mention the most
In contrast to the Northern Savanna, where individual or small clusters of
dwellings are often widely dispersed, most of the Forest people live in compara-
tively large villages, with their compounds grouped closely together. Generally
surrounding the village is a more or less continuous belt of small vegetable gar-
dens, beyond which lie, helter-skelter, the farm clearings proper. Except for
cocoa and rice, which are commonly grown alone, mixed agriculture prevails
here, several crops that reach maturity at different times of the year being planted
in the same plot. With a new clearing the cycle begins in late March or early
April. In a typical sequence maize and manioc are then planted, to be followed
a month or so later by plantains and cocoyams. The brief dry period centered
in August sees the early maize collected, and often a second crop planted. Octo-
ber-December is the main harvest season. Plantains and cocoyams, however,
are gathered almost continuously, as is manioc, which can be harvested as re-
quired. As a result food is plentiful throughout the year; so plentiful in fact
that manioc and cocoyams, which are generally considered "inferior" foods, are
frequently treated as reserve crops and often only partially harvested (43, pp.
Some impression of the relative importance of the various food crops grown
in the Forest is to be had from the area estimates published for 1950 by the De-
partment of Agriculture (11, p. 32). In preparing these estimates both political
and natural boundaries were respected, except that in Ashanti the forested and
savanna portions were not distinguished; the Forest, accordingly, is best repre-
sented by that part which falls within the former Gold Coast Colony (the East-
ern and Western Regions plus the southern quarter of Trans-Volta/Togoland).
Here plantains are eminently the most important crop in terms of acreage, occu-
pying 43 per cent of the area reported in the major staples. Next come cocoyams
with 27 per cent, followed by maize and manioc, accounting for 16 and 10 per
cent respectively. Yams and rice together were thought to take up no more than
4 per cent of the total. The Forest is by no means a homogeneous area, however,
and though this general ranking is probably valid for the zone taken as a whole,
THOMAS T. POLEMAN
there are marked differences in local cropping. Rice, for instance, is a key crop
in the very moist southwestern corner of the zone, as it is in the forested area
which extends across the Volta into southern Togoland; indeed in tl e latter area
the 1950 estimates have it occupying a quarter of the land in staple ocds.
The importance of food-crop production for sale in outside r au kets also
varies considerably within the zone. Though commercial participation by food
farmers in the Forest is almost everywhere greater than in the Northe rn Savanna,
it is greatest in the closely settled portions near the larger towns. These areas
commonly enjoy a distinct competitive advantage, notwithstanding the fact (to
be brought out later) that the larger centers draw their food supply s from con-
siderable distances. Not only are they nearer the final markets; the, also enjoy
superior road facilities. Particularly in the densely peopled southea ti n corner
of the Forest between the Pra and Volta rivers (especially in tie Swedru-
Nsawam-Koforidua sector, where the effects of Swollen Shoot diseas- have been
most heavily felt by cocoa farmers) and within a radius of about -;0 miles of
Kumasi, production of food crops for the market accounts for a subst nl ial share
of total output. In contrast, many of the more distant sections of I'.shanti and
the Western Region contain extensive tracts of excellent farm lane which are
exploited entirely on a subsistence basis.
The third and smallest of Ghana's natural regions is the Coast; I Savanna.
Reflecting the dry conditions which obtain along the West African c :a't east of
Cape Three Points, this zone, with its scrubby vegetation, presents a sharp con-
trast to the adjacent forest. West of Accra it is confined to a narrow strip never
more than 10 to 15 miles wide; but to the east it broadens out to emb ace a com-
paratively wide plain. With a population of about 1.3 million, it .s the most
densely settled of the three regions. Including as it does both Accra alnd Sekondi-
Takoradi, the country's chief ports and first- and third-ranking citic it is also
the most urbanized. According to preliminary returns from the 1 60 census,
these centers, together with Cape Coast, Winneba, and Keta, contain well over
a third of the inhabitants of the region (13, p. 1).
The Coastal Savanna's only important resource is its location. S raiegically
situated between the Forest and outside markets, it has historically oft( red attrac-
tive opportunities for trade and settlement. Agriculturally, its capa )ilties are
quite limited. Rainfall is the lowest in the country, averaging betw< en 30 and
40 inches. Nonetheless, several factors combine to make production somewhat
more flexible than in the Northern Savanna. Precipitation, in the fir;t place, is
more evenly distributed over the year, following the bi-modal pattern found in
the Forest, while seasonal aridity is less intense, the full effect of the -armattan
Two crops, maize and manioc, completely dominate agriculture in t ie Coastal
Savanna, occupying, according to the estimates of the Department o: Agricul-
ture, 47 and 43 per cent, respectively, of the land in staple foods ( 1, p. 32).
Though both crops do well here (they are about the only staple crop; that do),
manioc is the more widely dispersed. Tolerant of both poor soils and ( brought, it
is almost the sole crop of the dry, overcropped plains around Accra. Ma ze. on the
other hand, has slightly higher moisture requirements and is chiefly found in the
more humid portions behind Keta and along the escarpments which Ilorder the
GHANA'S URBAN FOOD ECONOMY
plain on the north and which coincide with the beginning of the Forest. In these
regions two crops a year are possible. In contrast to the Forest, where the tsetse fly
limits the livestock population to sheep, goats, and fowl, cattle are also raised in
the Coastal Savanna. They are kept more for prestige purposes than for sale,
however, and slaughterings are limited (44, p. 77).
As would be expected in view of the semiurban and mercantile character of
the Coastal Savanna, food crop production here has become increasingly com-
mercialized. Along the roads leading into the principal towns especially, market
cropping of both vegetables and staple foods is a major industry. The zone is,
however, capable of meeting only a fraction of the needs of these centers; and,
as we shall see below, they rely heavily on the Forest and beyond for their food
As in Middle Africa generally, urbanization in Ghana has been proceeding
at a rate far faster than over-all population growth. How much faster is, unfor-
tunately, not entirely clear. As with many less-developed countries-African ones
particularly-the population data for Ghana leave much to be desired. This is
not to say that there have been no censuses; there have. But the results have
been so inconsistent as to seriously limit their usefulness as indicators of historical
Consider Chart 2. In it are plotted various estimates of the country's total
population and the size of its principal cities. Connected by the solid lines are
the figures reported in the last three censuses: that of 1931, "the first in which the
entire population was individually counted" (18, p. 7); that of 1948; and that of
1960, for which only provisional and incomplete returns are yet available. Also
shown are the latest estimates which were made prior to the 1960 count-esti-
mates which, though for the cities were based on water-consumption data, closely
followed the trend suggested by the 1931 and 1948 figures. The discrepancies
between the 1959 estimate and the 1960 count are huge. In mid-1959 the country's
population was officially reckoned at 4.9 million. The 1960 census returned a
provisional figure of 6.7 million. In 1958/59 it was considered that the Accra
and Sekondi-Takoradi water systems supplied populations of 208,000 and 68,000,
respectively. The 1960 enumeration suggests that 388,000 and 121,000 would be
better estimates. For Kumasi the comparison is less easy, because unlike Accra
and Sekondi-Takoradi, whose municipal areas are almost entirely urban in char-
acter, the domain of the Kumasi Municipal Council includes both the town
proper and a semirural suburban area. Nevertheless, a striking difference in
the estimates may be perceived. The 1958/59 water figure gave the town 82,000
persons, as opposed to 221,000 for the municipal area in the 1960 census. It is
generally agreed that the town's population accounts for well over two-thirds
of the municipal total.
Manifestly the three censuses cannot be used together: either the 1960 figures
are gross exaggerations, or the 1931 and 1948 counts understated reality by a con-
siderable margin. In all probability the latter is the case, though it would be
foolhardy to suppose that the level of accuracy attained in 1960 was anything like
that achieved in more advanced countries. But the 1960 enumeration, preceded
THOMAS T. POLEMAN
CHART 2.--GHANA: VARIOUS ESTIMATES OF TOTAL AND URBAN
(Thousand persons: logarithmic vertical scale)
1931 1948 1955 1960
Data for 1931 and 1948 from Gold Coast, Off. Govt. Stat., Census of Popldation, 19,48, Report
and Tables (1950). Data for 1960 are provisional and are from Ghana, Off. Govt. S it., Quarterly
Digest of Statistics, June 1960, p. 1. Estimate of total population, mid-1959, from ibid. March 1960,
p. 1. Estimates for urban areas, 1958/59, from Ghana, Min. Fin., Economic Survey, 1959 (1960),
p. 47. Implicit budget survey estimates were obtained by multiplying by four the numl er of persons
enumerated during the "population enquiry" phases of the surveys, as these purported o cover close
to 25 per cent of the populations. Multiplicands are from Gold Coast, Off. Govt. Stat. (Si it. ind Econ.
Papers), 1953 Accra Survey of Household Budgets (No. 2, December 1953), p. 3; Ktin. i'i Survey of
Population and Household Budgets, 1955 (No. 5, March 1956), p. 18; and Sekondi-Ta, oradi Survey
of Population and Household Budgets, 1955 (No. 4, March 1956), p. 16.
a Communities reported to have 5,000 or more inhabitants in 1948.
by a "census education" campaign of almost six-months duration (whi :h reached
a climax with countrywide festivities on "census night")6 and carrie(l out with
the assistance of United Nations' experts, was almost certainly the mo t accurate
count thus far taken (14, pp. 15-16). Both the 1931 and 1948 census es, on the
other hand, have long been suspect. Indeed, the officer in charge o the 1931
6 An exceptionally clever device for allaying suspicions, by the way, and one which ctlects credit
on the census authorities.
GHANA'S URBAN FOOD ECONOMY
enumeration called attention to numerous specific inaccuracies and stated that a
majority of his subordinates "were of the opinion that the figures collected by
them are lower than they should have been by from 2 per cent to 10 per cent"
(8, p. 123).
There are grounds for believing the 1948 census was similarly incomplete,
this despite the fact that the count was technically much better organized. Early
1948, when the census was taken, was a time of considerable civil unrest through-
out the country. Cocoa growers were bitterly opposed to the government's pro-
gram of compulsory cutting of trees affected by Swollen Shoot; and in the cities
the discontent which was to culminate in the bloody riots of February and March
was rife (37, pp. 15-16). This situation undoubtedly reinforced the African's
traditional suspicion of the census taker (based on the not unfounded fear that
his visit would be followed by one from the tax collector), and makes it highly
likely that a sizable proportion of the population was at pains to avoid enumer-
As indicators of the course of urban growth in Ghana, then, the pre-1960
figures have only limited value. They are not totally useless, however. While
apparently inaccurate in so far as absolute magnitudes are concerned, they prob-
ably reflect relative movements accurately enough. In any event they are all we
have, so use them we must.
Again consider Chart 2. It will be seen that between 1931 and 1948 the popu-
lation as a whole was thought to have increased from 3.2 million to 4.1 million,
or by 30 per cent. (Such an increase would reflect an average annual growth
rate of 1.56 per cent-not an unreasonable figure.) During the same interval,
in contrast, the number of people thought to be living in the country's 34 urban
centers rose by 68 per cent, from 314,000 to 527,000. But if the rate of over-all
urban expansion was over double that of the total population, most of it was in
the three principal cities. Accra, Kumasi, and Sekondi-Takoradi, which are to-
gether thought to contain about half of Ghana's urban dwellers, were each re-
ported to have experienced a population increase of over 90 per cent during the
period. The 1960 figures suggest that the rate since has been even faster.
Accra, the capital and largest city, owes its growth and pre-eminence to sev-
eral factors. Like most of the other coastal towns, it had its beginning in the
seventeenth century as a settlement surrounding one of the numerous European
trading forts. These coastal communities grew and prospered directly with the
ease and extent of their contacts with the interior. Thus, though Accra's future
was assured by its choice in 1877 as the colony's capital (under the mistaken no-
tion that it would prove a more healthful site than Cape Coast), its real rise to
pre-eminence dates from the construction of the eastern rail line. Begun in 1905
and completed, after innumerable delays, to Kumasi in 1923, this line tapped the
newly developing cocoa areas of the southeastern Forest. As its ocean terminus,
Accra flourished; quickly becoming the chief exporter of cocoa, it dominated
the trade until Takoradi harbor was completed in 1928. The 1931 census gave
its population at almost 70,000, nearly four times the 1911 figure.
7 Supporting this thesis are the findings of a social survey conducted in one of the older districts
of Cape Coast in 1953. One of the few such surveys to be based on an actual count instead of a
sampling of persons carried on the tax rolls, it revealed the population to be over double that reported
in the 1948 census (36, p. 37).
THOMAS T. POLEMAN
Accra's growth during the last 30 years reflects more than anyt irng else the
over-all development of the country's economy. Though surpassed by Takoradi
as the foremost port, it remains firmly established as Ghana's conlmercial and
administrative center. As the political capital, too, it has profited immensely
from the recent expansion of government's importance. The municipal area
has had to be enlarged repeatedly; and in 1960 its population was rut at 388,000
(1, pp. 16-31; 13, p. 1).
Kumasi, Ghana's second city and chief inland metropolis, is tile capital of
the Ashanti people-a fact which has contributed in no small part io its present
position. Unlike the other peoples of Ghana, the Ashanti have lot g enjoyed a
substantial measure of political unity; and as the seat of their kinj:, Kumasi is
today, as it was at least 200 years ago, the focus of their cultural mnd political
lives. More recently, however, its importance derives mainly frorr it; location
at the hub of inland communications. Reached by the western rail oad in 1903,
it rapidly developed into a major collecting center for cocoa and the staging
point for trade between northern Ghana and the Forest-positior s that were
further strengthened 20 years later with the completion of the eastern line. By
1931 the city and its suburbs were thought to include over 40,00C persons, as
opposed to only 6,000 twenty-five years earlier. Since then, aided greatly by the
westward migration of cocoa production and by construction of ; n extensive
network of feeder roads, Kumasi has boomed. The 1960 census returned a pro-
visional figure of 221,000 for the population of its municipal area (10 pp. 160-61;
33, pp. 4-7; 13, p. 1).
If railroads and cocoa largely explain the initial rise of Accra a ad Kumasi,
the combination was railroads and gold in the twin cities of Sekon i-Takoradi,
Ghana's third ranking urban center. Founded like Accra, in th,: shelter of
European trading "factories," the towns' site enjoyed no special adva ntages over
its many competitors until 1898, when Sekondi was selected as the t :rminus for
the western railway. Originally built to serve the gold fields aroui d Tarkwa,
this line was soon extended to Kumasi and later, by a branch, to tl e diamond
area around Akwatia. Drawing on such an extensive hinterland-rich in timber
and cocoa as well as minerals-the town could not but thrive. Nonetheless, it
did not really hit its stride until after the completion in 1928 of Takol adi harbor.
As the country's first, and until 1961 only deep-water facility (all the other ports
except the new one at Tema are open roadsteads, requiring the use o' inefficient
and costly surfboats), it immediately supplanted Accra as the chief po -t, drawing
traffic even from east of the Volta. The 1948 census put the population at 44,000,
an increase of 95 per cent over the 1931 figure. The provisional estim .te for 1960
is 121,000 (7, pp. 2-3; 13, p. 1).
As would be expected in view of the exceedingly rapid growth Df Ghana's
three major cities, a high proportion of their citizenry is composed of 'migrants"
(i.e., persons born elsewhere). This is particularly true of Sekondi-Ta woradi and
Kumasi, where, as Table 2 indicates, no less than 78 and 70 per cent, respectively,
of the persons enumerated in 1948 were born somewhere else. In Accr; the figure
was somewhat lower, 46 per cent, but the extent of that city's growth cluing the
last 12 years makes it not improbable that even there native-born 1 ers'ons are
now in a minority.
GHANA'S URBAN FOOD ECONOMY
This influx of newcomers has carried before it much of the cities' regional
flavor. As Table 2 makes clear, the pull of Ghana's towns has made itself felt
over a considerable portion of West Africa. The data suggest that fully 40 per
cent of the migrants to Accra, for instance, come either from the relatively dis-
tant Northern Region or from other countries (chiefly Nigeria and the former
French territories to the north of Ghana). Of the 62 tribal divisions of Ghana
recognized in the 1948 census, all but two were represented in Accra; and the
same count indicated that the indigenous Ga people accounted for only 52 per
cent of the inhabitants (1, pp. 30-31). If similar breakdowns were available for
Kumasi and Sekondi-Takoradi, they would almost certainly show that in these
centers too the predominance of local tribes had become equally precarious.
As is true throughout Middle Africa, the stream of migrants into Ghana's
cities is made up of many more men than women, with the consequence that
there is a sharp imbalance in the sex composition of their populations. In Kumasi,
for every 100 females in the 16- to 44-year age bracket enumerated in 1948, 136
males were counted. The disparity was even greater in Accra and Sekondi-
Takoradi. There the census recorded, respectively, 140 and 152 working-age
men for every 100 women (18, pp. 45, 51). Many of these men are unmarried;
others have left their families behind. On arriving in town they tend either to
rent living space from an established household (illegally if the dwelling is in
a government-housing estate) or to cluster together in dormitory-type buildings,
four or five or even more to a room (40, passim). The social instability which
results is, of course, considerable; and it is aggravated by the fact that much
migration is of a temporary character. Many of the migrants, especially the
married men, remain in town only a year or so-until they can accumulate a
certain amount of money or purchase certain goods-after which they return to
their homes. An impression of the prevalence of these "target workers" may be
had from Table 2. Roughly 60 per cent of the male newcomers enumerated in
each of the three cities in 1948 stated that they had lived there for less than
five years, and a quarter of them indicated that they had been there for less
than one year.
The picture of Ghana's cities that emerges, then, is one of extremely rapid
growth and of a heterogeneous citizenry which has yet to take root-of people
whose ties to their place of residence are as often as not unreinforced by those
of kinship or home. What bearing has this on the dietary characteristics of the
communities? Prima facie, it suggests that one would probably search in vain
to identify dietary preferences typical of all the people. It also suggests that
among the high proportion of single people ready-made or easily prepared foods
would find an uncommonly receptive market.
Our insight into the dietary behavior of Ghana's urban dwellers rests mainly
on the findings of three surveys of household budgets carried out some years ago
in Accra, Kumasi, and Sekondi-Takoradi (20, 21, 19). These surveys contain a
8 A fourth survey has been conducted in the Volta River town of Akuse. Because Akuse, with a
population according to the 1948 census of less than 5,000, hardly ranks as a true urban center, this
survey is not considered here.
TABLE 2.-AcCRA, KUMASI, AND SEKONDI-TAKORADI: DISTRIBUTION OF POPULATION BY BIRTHPLACE AND DURATION OF RESIDENCE, 1948*
(Per 1,000 of total population)
Birthplace of migrants Duration of residence of migrants
in All ern Togo- Foreign Less than 1-4 5 years
City and Sex Total city migrants Colonya Ashanti Terr. land countries 1 year years or more
Accra .............. Total 1000 538 462 205 25 61 49 122 118 156 188
Males 1000 463 537 207 29 88 52 161 149 166 222
Females 1000 627 373 204 19 29 45 76 82 144 147
Kumasi ............ Total 1000 297 703 189 271 134 22 87 177 228 298
Males 1000 277 723 165 251 172 23 112 198 224 301
Females 1000 321 679 217 293 90 20 59 152 232 295
Sekondi-Takoradi .... Total 1000 219 781 498 31 69 25 158 174 277 330
Males 1000 197 803 456 34 97 26 190 186 272 345
Females 1000 246 754 552 28 33 22 119 158 285 311
Based on data in Gold Coast, Off. Govt. Stat., Census of Population, 1948, Report and Tables (1950), pp. 16, 362, 366.
a The Eastern and Western Regions plus the southern quarter of Trans-Volta/Togoland.
b Excluding suburbs.
GHANA'S URBAN FOOD ECONOMY
wealth of evidence, most of it covering matters which previously had been the
subject of conjecture alone. They are, of course, not without their shortcomings.
As with similar ground-breaking investigations undertaken elsewhere in Middle
Africa, their usefulness is somewhat limited by the fact that they relate only to
short periods of time and to restricted segments of the population. These and
other deficiencies dictate that the surveys be interpreted with care and the data
not pressed too rigorously. But they are not such as to invalidate the basic
worth of the more general findings.
The Budget Surveys
Ghana's three urban budget surveys were conducted in 1953 (Accra) and in
1955 (Kumasi and Sekondi-Takoradi). All were undertaken by the Office of the
Government Statistician, and all had as their raison d'etre that agency's desire to
relate the weighting systems used in its indices of local prices to actual expendi-
ture patterns. Although no two of the surveys were exactly alike, the same
general procedures were followed in all.
The surveys consisted of two parts: a broad "population enquiry" and the
detailed budget survey proper. The function of the first part, though instructive
in itself, "was to provide a frame from which to select households for budget
recording and to show precisely to what section of the community the budgets
related" (21, p. 1). Toward this end it was based on a rather large sample, an
enumeration of 25 per cent of the African population in each municipality (or
town area in Kumasi) having been the objective. The sampling unit was the
house; the municipal property tax lists the basis for selection. Every fourth
address was visited, and each identifiable household-defined as "a group of
persons living and feeding together, i.e., a group forming a definite economic
unit" (21, pp. 1-2)-was questioned as to family size and structure, occupation
and income, and the number of rooms occupied. The only intentional omissions
were army barracks, government bungalows, and buildings classified as non-
That anything even approaching a 25 per cent sample was actually achieved
in any of the population inquiries is highly improbable. At the time it was
thought that coverage was reasonably complete; now the evidence points clearly
to the contrary. In all, 58,033 persons were enumerated-29,123 in Accra, 16,663
in Kumasi, and 12,247 in Sekondi-Takoradi (20, p. 3; 21, p. 16; 19, p. 18). If
these figures are multiplied by four (as has been done in Chart 2) the results fall
far short of the total populations suggested by the last two censuses; this remains
true even if allowance is made for the small number of individuals (notably
whites, transients, and members of the military) intentionally overlooked.
One can only speculate as to the causes of this understatement. It is not un-
likely, however, that use of the tax books as the basis for selection was a con-
tributing factor. As the 1960 census revealed, the three municipal councils, at
the time of the surveys, had no idea of how many people they were administer-
9 Accordingly, I intentionally pass over without comment many of the minor inconsistencies which
will be readily apparent to the reader.
THOMAS T. POLEMAN
ing; and though they were probably not equally ignorant of th( number of
dwellings within their jurisdiction, it is quite probable that difficulties were
encountered in drawing up the tax lists. We are informed that i Accra, for
instance, to find a particular house is often a major undertaking, th ,re being no
map which shows all street names. Furthermore, in some wards here are no
identifying house numbers, while in wards where there are numbers these are
not consecutive, but refer to an arbitrary order in a sanitation district: (4, p. 171).
Under such conditions it would be surprising indeed to find -hat the tax
lists were complete.
Also presumably involved were some of the same factors which ]: d to under-
enumeration during the 1948 census: political tension-especialli evident in
Kumasi at the time of its survey (19, p. 2)-and the African's long-standing
suspicion of strangers, particularly those acting in some official cap icily. Not a
few houses in urban Ghana, especially those occupied by migrants;, contain
upwards of 100 persons, and, as more than one government researcher has
found, "it is common for the occupants to deny the existence of a person who
is actually in the crowd of occupants expressing the denial" (4, p. 171). That
many men either were successful in doing just this during the pc pulation in-
quiries or were otherwise missed by the enumerators appears likely This con-
clusion is suggested by comparing the adult sex ratios found in the samples
with those reported in the 1948 census (which, though it was also incomplete,
was to all appearances a more reliable count). Such a comparison between the
1948 census figure and the population inquiry figures, multiplied by four, is
made in the following tabulation (in thousand persons) (20, p. 55; 21, p. 26;
19, p. 30; 18, pp. 45, 51):
Sex Accra Kumasi Takoradi
Adult males .................. 49.5 20.9 17.7
Adult females" ................ 37.3 14.9 11.2
Population Inquiries, 1953-55
Adult malesb .................. 27.6 18.7 16.2
Adult female ................. 31.3 19.1 13.1
a Persons 16 years and older.
b Persons 15 years and older, except in Accra where 19 and older.
The differences are startling. Whereas the census indicates a strong male
predominance in all three cities, the population inquiries imply tlat in Accra
and Kumasi women are actually in the majority and that in Sekordi-Takoradi
they account for fully 45 per cent of the adult population. Manifestl,, you cannot
have it both ways; and it is worth noting that to hold in favor of th: population
inquiries would involve some rather tenuous hypotheses. One we uld have to
assume, in the first place, that women were grossly underreported in 1948. There
is, in fact, evidence that some women did hide from the enumerators 'cf. 1, p. 38),
but that they would do so in great numbers during the census and then not do
so a few years later seems improbable. Furthermore, one would h ive to argue
that men, who are away at work all day, find it less easy to avoid numerationn
GHANA'S URBAN FOOD ECONOMY
during a house-to-house survey than do women. And finally, one would have
to contend that for some unknown reason the sex ratio in Ghana's cities differs
sharply from that prevailing elsewhere in urban Middle Africa.
All this rather strongly suggests that the "frames" provided by the population
inquiries were biased toward overenumerating of women. Nonetheless, any
atypicalities in the budget samples introduced at this stage were decidedly of a
secondary nature. For in selecting the households to be used for detailed budget
analysis no attempts were made to draw on cross sections of the populations.
Instead the samples were intentionally constructed so as to relate only to clearly
defined portions of the three communities.
The bases on which the three budget samples were selected are shown in
outline form in Table 3. It will be seen that whether or not a given household
was considered for one of the samples depended on both its size and composi-
tion as well as on the source and level of its income. The exact criteria differed
somewhat between the surveys. Most significant of these differences pertained
to income source. Households without wage earners were omitted from both
the Accra and Sekondi-Takoradi surveys. In deciding on this restriction the
authorities were guided by the fact that the surveys' chief end-products were to
be weighting systems for use in preparing indices of local prices. Households
depending mainly on wage incomes, it was reasoned, would be the ones primarily
affected by price changes (the incomes of traders and the like being more re-
sponsive to the level of prices) : if so, indices mirroring their expenditures would
be the most instructive. This restriction, however, was not applied in Kumasi.
The economic structure of Kumasi differs markedly from that of the other two
cities. Both Accra and Sekondi-Takoradi contain appreciable numbers of wage
earners: notably employees of commercial firms, the Government, and the port
in the former; and of the port and the railroad in the latter. Kumasi, on the
other hand, is still chiefly a trading center, and most of the workers are self-
employed. Of the households enumerated during the population inquiry, only
34 per cent received income from wages (19, p. 4). To have limited the budget
survey to just these households would clearly have defeated its purpose.
Otherwise, the factors governing the coverage of the samples were much the
same in all three surveys. Unfortunately, households with either very high or
very low incomes were not considered, all households in Kumasi and Sekondi-
Takoradi with monthly wage incomes outside the range f5 to 15 having been
disqualified, as were those in Accra whose heads earned less than 50 or more
than 180 yearly ( 4 4s. to 15 monthly). Still others were ruled out on "prac-
tical considerations": e.g., single persons and households lacking an adult female
were omitted on the ground that they would obtain much of their food at snack
bars or from cateresses, and households with nine or more members or with more
than a few sources of income were excluded to avoid unduly complicated budgets.
After these and a few other minor exclusions, there remained homogeneous
groups of households which could be described rather closely:
Accra.-Wage-earning households of 2 to 8 persons (excluding those of
one adult and three or more children), with no more than two income
earners, and with wage-earning heads in the income range of 50 to 180
per year (20, p. 4).
TABLE 3.-ACCRA, KUMASI, AND SEKONDI-TAKORADI: SIELECTiON OF BtlDGET SL-RVEY SAMPLES, 1953 AND 1955*
Accra Kumasi Sekondi-Takoradi
Households in Population Enquiry.. 4,898
Households excluded (in sequence):
1. Nonwage-earning heads ...... 2,566
2. Annual income of head under
50 or over /180 ........... 735
3. More than one income earner
besides head ................ 192
4. Single persons ............... 260
5. Nine persons or more ........ 109
6. One adult and more than two
children .................... 24
Total households excluded ........
Households meeting Burget Survey
as percentage of all households....
Budgets recorded ................. 500
Total useable budgets .............
Households in Population Enquiry.. 4,337
Households excluded (in sequence):
1. No regular income, or four or
more incomes ............... 323
2. Single persons ............... 770
3. Nine persons or more ......... 198
4. Monthly wage income under
100s. or over 299s. ............ 306
5. Without adult female ........ 206
6. Living in outskirts ........... 55
Total households excluded .........
Households meeting Budget Survey
as percentage of all households....
By 90 teachers .................
By 20 staff members ............
Total usable budgets .............
Ilouseholds in Population Enquiry.. 3,420
Households excluded (in sequence):
1. Without wage earners ........ 1,063
2. Monthly wage income under
100s. or over 299. ............ 606
3. Single persons ............... 555
4. Nine persons or more ........ 36
5. Without adult female ........ 120
6. Three or more nonwage in-
com es .....................
Total households excluded .........
Households meeting Budget Survey
as percentage of all households ....
By 90 teachers .................
By 16 staff members ............
Total usable budgets ............
Data from Gold Cuoat, Off. G,,ot. Stat. (Stat. and Econ. Papcrs), 193I' Accra Surrey of Household B udgets (No. 2, December 1953), 1p.. 5, 9Y Kninast Snrvcy
of Population and Household Budgets, 1955 (No. 3, March 1I'J0), pp. 7-8; and 5cikondi-Takoradi Survey of Popiultion and l IIouc ld dets, 1955 (No. A
March 1956), pp. 6-7.
GHANA'S URBAN FOOD ECONOMY
Kumasi.-Households of 2 to 8 persons (at least one of whom was an adult
female), with 1 to 3 incomes from all sources, but with any wage income
in the range 5 to 15 a month, excluding those living in the outskirts
(19, p. 7).
Sel(ondi-Tal/oradi.-Wage-earning households of 2 to 8 persons (at least
one of whom was an adult female), with no more than two sources of
nonwage income, and with wage incomes in the range 5 to 15 a month
From among these households the budget samples were chosen at random.
The effect of these selection procedures on the applicability and relative cover-
age of the samples defies precise appraisal. Table 3 shows the proportion of the
households enumerated in each population inquiry which met the budget survey
criteria. It will be seen that over-all coverage was greatest in Kumasi, where 57
per cent of the households qualified for the sample, and smallest in Sekondi-
Takoradi and Accra, where only 30 and 21 per cent qualified. This difference
mainly resulted from the exclusion from the survey in the latter two communities
of households without wage earners. In Accra 57 per cent of the households
received income from wages and the sample related to 36 per cent of them; in
Sekondi-Takoradi the figures were 69 per cent and 43 per cent, respectively
(19, p. 4).
Comparisons such as these are not particularly instructive, however, and
beyond them there is precious little to go by. This because, though coverage
per se was closely defined, next to nothing exists to which it can confidently be
related. In so far as household size and composition is concerned, the 1948 census
sheds no light, while the findings of the population inquiries were almost cer-
tainly biased in favor of women. That this bias was carried over to the samples
may safely be assumed, and also that it was magnified by the exclusion of single
persons. But apart from this, and the fact that the samples clearly covered house-
holds of larger than average size, further refinement is impossible.
The picture with respect to income coverage is similarly obscure. What the
distribution of incomes is among Ghana's urban dwellers is just about anybody's
guess. True, the households visited during the population inquiries were ques-
tioned as to income, but because of the difficulty of assessing nonwage revenue
during a single interview, the scope was restricted to income from wages. The
findings in all three cases were much the same. In both Kumasi and Sekondi-
Takoradi it was observed that fewer than one per cent of the wage-earning house-
holds had monthly wage incomes of less than /5 and that roughly a quarter of
them received more than 15 (21, p. 21; 19, p. 24); while in Accra it was found
that "the range of 50-180 per annum covered almost the entire lower income
range and included nearly 70 per cent of all families with wage-earning heads"
(20, p. 4). But if this suggests that the surveys related to households in the middle-
and lower-income brackets, there is no evidence that they included any falling
below the so-called "poverty line." This would imply that they did not have
enough to eat; and it is brought out below that the average apparent caloric
intake per capital reported for the Kumasi and Sekondi-Takoradi samples (quan-
tity data were not collected in Accra) was quite adequate and varied little from
THOMAS T. POLEMAN
one expenditure class to the next. What effect (if any) the inclusion of house-
holds with nonwage incomes or multiple sources of revenue had on this rough
picture is entirely a matter for conjecture; it can only be inferred .hat it was
reasonably uniform. Per capital monthly expenditure differed little among the
three samples: 3 14s. in Accra, 3 8s. in Kumasi, and 3 3s. in Sek >ndi-Tako-
radi (21, p. 11).
The budget surveys were carried out about a month after the population
inquiries and in each instance covered a period of 30 days (during May-June
in Accra, March-April in Kumasi, and February-March in Sekondi-Takoradi).
The briefness of this period is to be regretted; as elsewhere in Middle Africa
there are important seasonal changes in the dietaries of the urban ;hanaians.
It was felt, however, that this shortcoming would be more than cc mpensated
for by the larger number of budgets which a single month's record would permit.
In all, 1,569 useable budgets were obtained-453 of them in Accra (where 47
had to be discarded), 570 in Kumasi, and 546 in Sekondi-Takoradi. Records of
both income and expenditure were collected for each household. On the ex-
penditure side1" provision was made for data of two types-a daily accounting
of all domestic outlays (excluding those of a purely transfer nature), and an
estimate of the preceding twelve months' disbursements for clothing nd durable
goods. Because items of a durable character are purchased irreg ularly, the
findings were used in conjunction in arriving at a figure for "normal" monthly
expenditure. This was defined as "one-twelfth of the annual expenditure on
clothing and durables plus the survey period expenditure on other domestic
items" (21, p. 10). This definition is worth bearing in mind; on it; terms the
analysis of the budgets is grounded.
The day-to-day data collecting was done both by permanent men bers of the
Government Statistician's staff and school teachers employed on a par -time basis.
During the Accra survey the former acted as supervisors, while 100 t 'achers did
the actual recording. Each was assigned responsibility for five househo ld;. These
were visited every evening and their expenditures entered, item Ly item, in
specially printed booklets. In the other two cities a slightly different procedure
was followed. Again teachers-90 were employed in both instance --handled
five budgets apiece, but in addition to their supervisory duties each of the per-
manent interviewers (20 in Kumasi, 16 in Sekondi-Takoradi) wa; rlade re-
sponsible for six households. These households were selected as a subs; mAle from
the main group and were dealt with in somewhat greater detail; ir their case
quantities as well as values were ascertained. For items not purchased by definite
units, quantities were determined by weighing. The permanent i:iterviewers
were equipped with scales and arranged to meet the housewives daily) after mar-
10 The income side of the budgets has little relevance to the present inquiry. N i come per
household can he estimated for neither the Kumasi nor the Sekondi-Takoradi samples. This because,
though both wage and nonwage income was recorded, no account was kept in the n( nwige ledger
of changes in stocks and credit transactions. In Accra, on the other hand, each house ld was asked
to estimate its net income. But, either because of an unwillingness or inability to st. te this figure
accurately, these estimates were so unlikely (suggesting dis-saving was the rule in the igler income
brackets, saving in the lower) that for Accra too the budgets were analyzed on the bas s of expendi-
ture rather than income class.
It is not possible, therefore, to infer anything about the precise extent of saving : nd dis-saving
during the survey periods. Probably it was small.
GHANA'S URBAN FOOD ECONOMY
keting. These quantity figures provided an exceptionally detailed picture of diet
composition; accordingly, the Kumasi and the Sekondi-Takoradi surveys are
somewhat the more valuable.
The over-all expenditure patterns revealed by the surveys were remarkably
similar (Table 4). In all three cities food was by far the largest item of household
expenditure, accounting for an average of 57 to 58 per cent of all domestic out-
lays. Next came clothing, taking up between 12 and 15 per cent of expenditures,
followed by rent and taxes (5 to 8 per cent), fuel and light (4 to 6 per cent), and
drink and tobacco (4 to 6 per cent). Services, household durables, and miscel-
laneous items together accounted for no more than 10 to 14 per cent.
That food should bulk so large in the domestic expenditures of Ghana's urban
dwellers is not surprising; although an average per household figure of 57 to 58
per cent is considerably higher than one would expect to encounter in all but the
lowest-income classes in advanced regions, it is wholly consistent with the find-
ings of similar investigations undertaken in other low-income countries.11 What
is surprising (to one accustomed to analyzing Western budgets, at least) is the
constancy of the figure throughout virtually the entire expenditure (income)
range. This is brought out clearly in Chart 3. In both Accra and Kumasi the
proportion of total expenditures allocated to food declined but slightly with an
increase in total domestic outlays. The extreme observations are worth noting.
In the lowest-expenditure class observed in Accra food absorbed 63.3 per cent
of all outlays, whereas in the highest class it accounted for 58.3 per cent; in
Kumasi the percentages were 56.2 and 52.7, respectively-all this among samples
which embraced over a fivefold difference in total expenditures between the
extreme classes. Only in the Sekondi-Takoradi sample was there a truly pro-
nounced drop, with the proportion of expenditures allocated to food declining
from 63.8 per cent in the lowest class to 42.5 per cent in the highest. But a word
of caution about Sekondi-Takoradi. Over a third of the decline there was ac-
counted for by the highest-expenditure class. This class included only 10 house-
I1 Some findings of a cross section of recent surveys (23, Tables All and Bl1):
Number Per Per cent "Income"
of capital spent elasticity
house- expendi- for of food ex-
Country Year Coverage holds turea food penditureO
Egypt ........... 1955 3 villages 493 70 70.7 .92
Cevlon .........1952-53 All population 1,085 85 63.1 .82
Ghana .......... 1953 Accra 453 124 58.0 .98
Portugal .........1950-51 Porto 667 160 62.8 .72
Puerto Rico ...... 1952 Urban wage earners 786 218 51.4 .80
Italy ............1953-54 Nonfarm population 1,599 285 57.8 .66
Cuba ........... 1951 Havana 881 549 43.8 .73
Canada ......... 1955 5 large cities 787 1,298 26.8 .40
United States ..... 1955 Urban population 2,241 1,504 32.5 .44
a Annual average per capital living expenditures in U.S. dollars, converted at official exchange
rates, in 1955 prices.
b Per cent of total expenditures.
c Double-log coefficient of expenditure elasticity, the influence of household size having been
THOMAS T. POLEMAN
holds, and we are advised that the pattern of their expenditures w: s distorted
by some "irregular" payments for services and rent (21, p. 13). The i-ference
is that in Sekondi-Takoradi as well, food items commanded a rath.-r constant
proportion of total outlays per household.
TABLE 4.-GHANA, URBAN HOUSEHOLD BUDGET SURVEY SAMPLES,
DOMESTIC EXPENDITURE PATTERNS*
1953 .se 1955:
Expenditures by Uses (per cent)
Food .............. ...... 58.0 57.3
Clothing ................. 12.1 13.8
Rent and taxes ............ 5.4 7.5
Fuel and light ............ 4.7 6.2
Drink and tobacco ......... 6.1 4.2
Services .................. 5.8 5.8
Durable goods ............ 3.6 2.1
M isccllaneous ............. 4.3 3.1
T otal ................ 100.0 100.0
Number of households ..... 453
Average size (persons) ..... 4.24
Average monthly cxpcndi-
tures ( .., ) ......... 314
Per capital monthly expendi-
tures (.. '.. ) ......... 74
(Fe .-Mar. 1955)
DI)ata from Gold Coast, Off. Govt. Stat. (Stat. and Econ. Papers), Selkondi-Tako adi Survey of
Population and Household Budgets, 1955 (No. 4, March 1956), pp. 11-12.
The explanation for this result is to be sought in terms of several f. ctors, some
clear-cut, some obscure. Chief among the latter, to be sure, is the que tion of the
samples themselves and, more specifically, of the income sectors to ,vhich they
pertained. This, it has been noted, is a point of singular elusivene our only
firm bench mark being that data for wage incomes suggest that cc ve age was
restricted to households in the lower- and middle-income ranges. But what about
nonwage receipts? Were these sufficiently large to elevate some of the house-
holds to the upper-income echelons, and to provide a reasonable cross section of
incomes? Certainly expenditures in the highest classes were not pI.ltry. In
Accra, for instance, the average monthly outlay in the highest-expenliture class
was 675 shillings, pointing to annual receipts, if savings are ruled out, of the
order of /405. This sum compares not unfavorably with the 19- 4 ;cale for
Grade-5 civil servants, the highest grade in the Government to which an African
could then normally aspire.'1 I suspect, nonetheless, that even these households
fell not in the upper, but the middle-income reaches, for household income in
12 In that year the lowest starting wage for Grade-5 civil servants was 220, wit] 580 as the
possible ultimate salary (1, p. 67).
GHANA'S URBAN FOOD ECONOMY
CHART 3.-GHANA, URBAN HOUSEHOLD BUDGET SURVEY SAMPLES, 1953 AND 1955:
PERCENTAGE OF TOTAL DOMESTIC EXPENDITURES SPENT ON FOOD, PER CAPITAL
MONTHLY FOOD OUTLAYS, AND HOUSEHOLD SIZE, BY EXPENDITURE CLASS*
0 100 200 300 400 500
TOTAL EXPENDITURES (SHILLINGS PER MONTH)
Based on data in Gold Coast, Off. Govt. Stat. (Stat. and Econ. Papers) 1953 Accra Survey of
Household Budgets (No. 2, December 1953), pp. 10-12; Kumasi Survey of Population and Household
Budgets, 1955 (No. 5, March 1956), pp. 36-37; and Sekondi-Talkoradi Survey of Population and
Household Budgets, 1955 (No. 4, March 1956), pp. 32-33. Positioning of data for extreme expendi-
ture classes reflects actual average expenditures; for other classes the midpoint. Figures do not include
expenditures on drink.
ACCRA Expenditure class limits
% spent on food hown 'by vertical lines. 7
_ ..* *.4....' .
_0 .... ..... ....
.o-"'' d Household sie
o0 .. ".. '| 3
'Per capital food outlay (463 house
0 200 30 I 5 0
500 600 700
0 100 200 300 400
THOMAS T. POLEMAN
urban Ghana is a close function not only of the level of individual ir comes, but
of household size as well. Accepting this interpretation as correct, it n ay well be
that had truly high-income households been included in the samples, tlhe expected
decline in the proportion of outlays allocated to food would have manifested
itself more strongly.
Turning now to a less speculative consideration, there is the relationship just
mentioned-that between income and household size. This is illustra ed graphi-
cally in Chart 3, except that total expenditure rather than income is shown.
Households in the lowest-expenditure class in all three samples had in average
of roughly three persons apiece, whereas those in the highest classes w ere almost
twice as large. Few households in urban Ghana rely on just one inco 'e earner.
Not only will a man customarily work, but his wife as well; of the wcmen 15
years and older enumerated in the Kumasi population inquiry, 77 per cent were
gainfully occupied-chiefly as "mammy traders"-as were 64 per ce it af their
number in Sekondi-Takoradi (21, p. 30; 19, p. 34). But this is not all. The Afri-
can concept of a household is such a broad one that it may encom ass many
adults, all of whom contribute to total receipts. As income rises, then, so often
does the number of persons who must be housed, clothed, and fed. T u:; it was
in the samples. Though the range of total expenditures among the extreme
classes covered varied between four- and sixfold, the differences on a per capital
basis were much less pronounced. Here are the figures (in shillings p 'r month:
20, pp. 10-12; 21, pp. 32-33; 19, pp. 36-37):
Expenditure class Accra Kumasi Takoradi
Total Expenditures per Household
Lowest expenditure class............... 119 124 :135
Highest expenditure class.............. 675 628 521
Total Expenditures per Capita
Lowest expenditure class.............. 37 44 47
Highest expenditure class.............. 111 103 88
But the tendency for an increase in incomes to be accompanied b) a rise in
the number of mouths to be fed only partially explains why food it:ms com-
manded so constant a proportion of total expenditures. Also involved, as would
be expected and as Chart 3 reveals, is that as income goes up so do per capital
outlays for food. This was most strikingly evidenced in the data for Accra and
Kumasi. There, per capital monthly food expenditures rose from abolt 25 shil-
lings in the lowest-expenditure classes (where total domestic outlays mounted
to about 40 shillings per person) to an average of about 60 shillings in tl e highest
classes (total per capital expenditures of over 100 shillings), and even the a showed
no signs of having approached a ceiling. These figures suggest that th( demand
for food is highly income elastic; indeed a double-log regression fitt,'d to the
Accra data gives a per capital constant expenditure elasticity coefficier t of .98,
and for Kumasi .86. A somewhat different situation is reflected in the data for
Sekondi-Takoradi. Though there again per capital food expenditures increased
with income, they did so rather less sharply than in the other two san ples; the
double-log method gives an elasticity coefficient of .55. Closer examination of
the data indicates this disparity chiefly mirrors a rather low income elasticity of
GHANA'S URBAN FOOD ECONOMY
demand in Sekondi-Takoradi for fish products. Before discussing this point,
however, it is desirable first to examine the surveys' findings with respect to the
broad characteristics of the diets of Ghana's city dwellers.
In their essential features, the diets of urban Ghanaians closely resemble those
of other Middle African peoples (28, p. 3). As is true over virtually all of Middle
Africa, most meals are built around a mass of stiff porridge or doughy paste made
from one of a small number of the starchy-staple foods-the cereals, the starchy
fruits, roots, and tubers-bits of which are dipped into a sauce or thin stew before
being eaten. The porridge or paste will invariably supply the bulk of the meal's
calories, whereas the sauce, in addition to adding flavor, will contribute most of
the protein, fat, vitamins, and mineral content. Sauces may be made from a wide
variety of edibles. In the forested regions of Ghana, palm nuts and palm oil are
standard ingredients, as are, to mention but a few, dried fish, beef, snails, peanuts,
peppers, tomatoes, and onions.
Some idea of the relative contributions of the various food groups to the diet
is to be had from Table 5, a rather unwieldy (but instructively detailed) tabula-
tion of per capital purchases among the households in Kumasi and Sekondi-
Takoradi from which quantity data were obtained. Its more striking features
are summarized below in percentage terms:
Food group Calories Protein Calories Protein
Starchy staples .............. 70.2 32.4 63.5 30.7
M eat ....................... 7.2 25.8 4.8 14.9
Fish ...................... 3.3 23.9 5.6 36.5
N uts, fats, and oils........... 11.4 6.4 21.1 13.0
Vegetables and pulses......... 3.6 8.3 1.4 2.6
Drink .................. .. .8 .2 .4 .2
M miscellaneous ............... 3.5 3.0 3.2 2.1
Total .................. 100.0 100.0 100.0 100.0
Among other things, the data suggest that roughly two-thirds of the calories in
the typical diet in Sekondi-Takoradi are derived from the starchy staples, while
in Kumasi even more-70 per cent-come from this source.
Diets dominated by the starchy staples are characteristic not only of Ghanaians
and other Middle Africans, but of poorer peoples the world over.l" This for a
13 Some examples, as of 1934-38, from other low-income regions (3, pp. 214-15):
Percentage of total
calories derived from
Country starchy staples
China ........................................................... 77
Philippines ....................................................... 73
India .... ....................................................... 69
M exico ................................................ ....... .. 55
Brazil ..................................... ..................... 50
By way of contrast, it is estimated that the starchy staples contributed only 32 per cent of the calories
consumed in the United States during the same interval.
148 THOMAS T. POLEMAN
TABLE 5.-KUMASI AND SEKONDI-TAKORADI BUDGET SURVEY SAMPLES 1955:
AVERAGE DAILY PURCHASES OF FOODSTUFFS PER CAPITAL, HOUSEHOLD [IS
INTERVIEWED BY PERMANENT STAFF*
Manioc and manioc products
Fresh roots .............
Gari (m eal) ............
Kokonte (dried roots) ...
O theira ................
Y am s ...................
Maize and maize products'
K enkyc ................
Shelled maize ...........
Cocoyam s ................
B reada ..................
Total starchy staples....
Fresh red meat ..........
Canned red meat ........
Fresh fish ..............
D ried fish/ .............
Canned fish ............
Palm nuts ..............
O their" ................
Fats and oils
Palm oil" .............
Other vegetable oils' .....
Butter and margarine ....
Tom atoes ...............
O kra ..................
Households) (96 Housel olds)
Protein Quantity Protein
Calories (grams) (grams) Calori :s (grams)
GHANA'S URBAN FOOD ECONOMY 149
TABLE 5.--KUMASI AND SEKONDI-TAKORADI BUDGET SURVEY SAMPLES, 1955:
AVERAGE DAILY PURCHASES OF FOODSTUFFS PER CAPITAL, HOUSEHOLDS
INTERVIEWED BY PERMANENT STAFF-(Concluded)*
(120 Households) (96 Households)
Quantity Protein Quantity Protein
Foodstuff (grams) Calories (grams) (grams) Calories (grams)
Pulsesi .................. 13.3 45 3.1 2.6 9 .6
Beer ................... 7.5 4 .1 5.5 3 .1
Palm wine ............. 2.4 8 8.7 3
Soft drinks ............. 1.2 1 4.7 2
Imported spirits ......... .6 2 .2 1 -
Milk and milk products.. 14.9 29 1.4 14.0 26 1.3
Fruit" .................. 13.8 6 .1 1.9 1
Sugar ................. 11.3 31 14.5 35
Salt .................... 18.3 18.0 -
Eggs ................... 1.1 2 .1 .3
O their ................. .9 3.4 1
Total foodstuffs" ....... 1,570.0 1,943 54.3 1,367.6 1,984 60.9
Based on data in Gold Coast, Off. Govt. Stat. (Stat. and Econ. Papers), Kumasi Survey of
Population and Household Budgets, 1955 (No. 5, March 1956), pp. 48-49, and Selondi-Taloradi
Survey of Population and Household Budgets, 1955 (No. 4, March 1956), pp. 44-45. Caloric
and protein values computed by using conversion factors given in FAO, Food Composition Tables-
Minerals and Vitanmns-For International Use (Nutr. Studies 11, Rome, 1954), pp. 10-27; B. S.
Platt, Tables of Representative Values of Foods Commonly Used in Tropical Countries (Gr. Brit.,
Medical Research Council Special Report 253, London, 1945), pp. 10-34; and Lucius Nicholls, Tropi-
cal Nutrition and Dietetics (3d ed., London, 1951), pp. 424-26, except in the case of palm nuts, for
which values were derived by assuming 25 per cent waste and an oil content of the remainder of 50
per cent. Where quantities were given in terms of bottles rather than weights, the following bottle
sizes were assume: vegetable oils-half an Imperial gallon; beer, palm wine, and soft drinks-12
fluid ounces; imported spirits-one-fifth gallon.
a Dough and krako.
b Chiefly maize, but apparently also includes millets and sorghums.
c Calories and protein computed assuming kenkey (fermented dough balls) is 55 per cent water,
dough 45 per cent water, as reported in Faye W. Grant, "Nutrition and Health of Gold Coast Chil-
dren," I. Am. Diet. Assn., July 1955, pp. 690-91.
d Includes biscuits and togbei (dough cakes).
e Bananas are not technically starchy staples, but as they are often prepared in the same way as
plantains, they are included here as such.
f Includes smoked and fried fish.
9 Coconuts, peanuts, tiger nuts, and kola nuts.
h Both palm-nut and palm-kernel oil.
1 Coconut oil, peanut oil, and shea butter.
i Chiefly peas and beans.
Ic Oranges, limes, guavas, papayas, avocados, mangoes, and pineapples.
SIncludes sugar cane.
m Coffee, tea, etc., and unspecified foodstuffs.
n Figures do not include allowances for meals taken out or snacks purchased away from home
during the day.
very simple reason-their cheapness, whether expressed in terms of market price
or production cost. As a general rule, less land and less labor expense are required
to produce a thousand calories of energy value in the form of starchy staples than
in the form of other foodstuffs. In southern Ghana, as Table 6 brings out, they
are rivaled in this regard only by the oil palm. In Sekondi-Takoradi and Kumasi,
150 THOMAS T. POLEMAN
TABLE 6.-KUMASI (MARCH-APRIL 1955) AND SEKONDI-TAKORADI (FEBRI:ARY-MARCH
1955): AVERAGE RETAIL PRICES OF SELECTED FOODSTUFFS*
Pence per thousand calories Pence per kilogram
Sekondi- S kondi-
Foodstuff Kumasi Takoradi Kumasi Takoradi
Manioc and manioc products
Kokonte .............. 1.37 2.26 4.63 7.66
Fresh roots ............ 2.25 2.29 2.45 2.50
Gari ................. 2.47 2.71 8.35 9.17
Maize and maize products
Shelled maize ......... 1.82 2.55 6.48 9.07
Kenkey .............. 4.82 4.48 7.86 7.31
Plantains (fresh) .......... 2.56 3.63 1.92 2.72
Palm nuts ................ 2.90 1.34 9.60 4.45
Cocoyams ................ 3.16 4.22 2.72 3.63
Rice (raw) ............... 4.37 4.44 15.61 15.86
Yams .................... 4.97 6.42 4.47 5.78
Bread .................... 9.26 9.83 23.35 24.78
Smoked fish .............. 26.19 24.11 53.42 49.19
Red meat (cheaper cuts).... 28.94 22.29 46.31 35.67
Eggplant ................. 39.70 50.45 7.94 10.09
Tomatoes ................ 109.79 165.00 20.86 2;1.35
Based on data in Gold Coast, Off. Govt. Stat. (Stat. and Econ. Papers), Kumasi 5 ,frvy of Popu-
lation and Household Budgets, 1955 (No. 5, March 1956), p. 48; Sekondi-Takoradi S irv'y of Popu-
lation and Household Budgets, 1955 (No. 4, March 1956), p. 44; caloric values calcul: ted using con-
version factors given in the sources cited in the footnote to Table 5.
during the periods covered by the surveys, such other foodstuffs as m( at, fish, and
vegetables cost consumers between 10 and 50 times as much per calorie as did
the principle starchy staples.
But if the diets of urban Ghanaians are those of poor people, it do( s not neces-
sarily follow that they are poor diets. In Kumasi (see Table 5), recorded pur-
chases per person amounted to the equivalent of 1,943 calories and ;4 grams of
protein daily, while for Sekondi-Takoradi the figures were 1,984 and 61. These
totals, it should be noted, almost certainly understate actual consun option: esti-
mated caloric and protein content were computed using the most conservative
conversion factors, and meals and snacks taken away from home ale excluded.
They may understate actual intake by as much as from two to fo ir hundred
calories. But even at face value, they clearly point to a dietary wh ch, at least
in so far as protein and calories are concerned, is more than adequate e. If, using
the Oxford Scale, they are converted from a per capital to an adull-equivalent
basis, the following picture emerges:14
Average daily intake per adult c uivalent
City Calories Protein ,gtams)
Kumasi .......................... 2,841 7).4
Sekondi-Takoradi ................. 2,841 87.3
14 On the Oxford Scale of adult equivalents, the household head counts as one un.t, other adults
and children of 14 and over as .7 unit, and children under 14 as .5 unit (41, p. 7). H :re, because of
the age breakdown available, children under 15 have been taken as .5 unit.
GHANA'S URBAN FOOD ECONOMY
Using the subsequently reduced 1948 recommendations of the Food and Nutri-
tion Board as the basis for his calculations, Dr. Lucius Nicholls has suggested
the following daily allowances for a 121-pound man living in the tropics (31,
Nature of work Calories Protein (grams)
Sedentary ........................ 2,100 60
Active ........................ 2,500 65
Very active ....................... 3,000 70
Though most meals follow day after day on the same broad plan, and are
built around a limited number of staple foods, variety in the diet is not totally
wanting. Not only does the sauce offer a great range for variation; the staple
itself may be prepared in a number of ways. Although plantains, manioc, yams,
and cocoyams are often merely peeled, boiled, and mashed before being served,
the most common procedure is for the staple to be pounded into a flour or meal
prior to cooking. With the cereals this involves either a grinding stone or a
mortar and pestle; with the roots and tubers, slicing, drying, and sometimes cook-
ing are also needed. Minor variations at this stage may result in totally different
products. Manioc flour, for instance, is made by pounding roots which have been
split and sun dried (kokonte), while the preparation of manioc meal (gari),
altogether another thing, differs mainly in that, before drying, the roots are first
grated and pressed (28, pp. 112-13). Once the flour or meal has been readied,
many alternatives offer themselves. Simply by varying the amount of water that
is added, for example, thin gruels as well as thicker porridges or pastes may be
concocted. A Ghanaian cook book offers a comparatively wide range of dishes
which may be prepared from a restricted number of staple ingredients.16
The brevity of the list of staple foods on which urban Ghanaians rely for the
bulk of their calories is readily evident in Table 5. In Sekondi-Takoradi, just
four crops-manioc, maize, plantains, and rice-supplied over 90 per cent of
the calories derived from the starchy staples, while manioc and maize alone
contributed almost 70 per cent. Much the same situation obtained in Kumasi,
though as would be expected in view of its setting, the relative importance of
the several crops was somewhat different. There, manioc and plantains were
the chief calorie sources, together contributing two-thirds of those supplied by
the staple foods, while yams, maize, rice, and cocoyams accounted for the bulk
of the remainder. These figures must, of course, be taken with certain reserva-
tions; they relate to periods of only 30 days duration, and, as is shown later on,
the relative contributions of the several staples change seasonally with local price
and supply conditions. They are, however, indicative of a situation which pre-
vails throughout the year.
15 As this study went to press, an unpublished manuscript by Miss Pauline Whitby of the Medical
Research Institute, Accra, was brought to my attention. In this manuscript, entitled "Calculations of
Food Consumption and Nutrient Intake in Five Areas of Ghana," Miss Whitby offers some interesting
estimates of the "normal" per capital caloric requirements of the populations of Kumasi and Sekondi-
Takoradi. Using the standards set forth in the 1957 FAO Report on Caloric Requirements, these she
put at 2,050 and 2,230 calories per day. Given the extent to which the survey data understate actual
consumption, Miss Whitby's estimates offer further confirmation of the over-all adequacy of the caloric
16 For detailed descriptions of many common Ghanaian dishes and the methods of their prepara-
tion, see 24, pp. 687-88 and 17.
THOMAS T. POLEMAN
But restricted though the list of staples normally purchased by Gh inm's urban
dwellers may be, there is reason to believe that, because of their gene ally higher
incomes, they consume a slightly greater variety than the rural popul ition. This
conclusion is suggested by Table 7, in which the findings of the K urasi and
TABLE 7.-GHANA: RELATIVE CONTRIBUTIONS OF THE VARIOUS STARCHY STI. PL S TO THE
DIET IN KUMASI (1955), SEKONDI-TAKORADI (1955), SOUTH-EAST AKIM -B ULAKWA
(1952/53), AND THE ODA-SWEDRU-ASAMANKESE REGION (1955 *
Urban Rur I
Sekondi- South-East (iOa-Swedru-
Item Kumasi Takoradi Akim Abuakwa \samankese
Per Cent of Total Calories from Starchy Staples
Manioc and manioc products. 37.5 46.3 50.4 41.5
Fresh roots .............. 17.8 36.2 .... 40.8
Processed items .......... 19.7 10.1 .... .7
Plantains and bananas ...... 29.2 13.9 17.7 29.6
Y am s ..................... 9.0 3.9 2.2 2.9
Maize and maize products ... 7.8 22.1 5.6 7.7
Rice ...................... 7.4 8.8 1.2
Cocoyams ................. 7.2 1.2 24.1 15.7
Bread ..................... 1.9 3.8 -1.4
Total ................. 100.0 100.0 100.0 100.0
Total Calories from Starchy Staples
Per person per day .......... 1,364 1,260 1,435 1,312
Number of households ...... 120 96 832 1,080
Period covered .............. 30 days 30 days 12 months 3 months
(Mar.- (Feb.- (June- (Oct.-
Apr.) Mar.) May) Dec.)
Based on data in Table 5; and Gold Coast, Off. Govt. Stat. (Stat. and Econ. Pa )ers), Agricul-
tural Statistical Survey of South-East Al(im Abual(wa, 1952-53 (No. 1, September 195), p. 10; and
Survey of Population and Budgets of Cocoa Producing Families in the Oda-Swedru-Asc, na.akese Area,
1955-1956 (No. 6, July 1958), pp. 91, 98; caloric values calculated using conversion factors given
in the sources cited in the footnote to Table 5.
Sekondi-Takoradi surveys are compared with those of similar investigations
conducted about the same time in two rural areas-South-East Akin Abuakwa,
extending northward from Nsawam for about 25 miles; and the Clda-Swedru-
Asamankese region, similarly situated with respect to the town ,f Swedru.
Though both of these areas are located in the Forest, Ghana's riche i .nd most
diversified agricultural region, in neither do we find quite the variel v of staples
observed in the cities, chiefly because neither rice nor bread was cInsumed to
any appreciable extent. Both these items, it should be noted, are co pnaratively
expensive and (though rice is grown in some parts of the Forest) irr ported.
That the surveys cited in Table 7 should point up manioc, a root crop, as the
pre-eminent staple in urban as well as rural diets in southern Ghana warrants
GHANA'S URBAN FOOD ECONOMY
a word of explanation. As a general rule, the starchy roots and tubers grown in
Middle Africa do not lend themselves readily to commercialization. The cereals,
concentrated and relatively imperishable, can be transported from place to place
at little cost. Most of the roots and tubers cannot be. Though cheaply produced
in humid regions, because of their bulky and semi-perishable nature, they tend
quickly to lose their price advantage if moved any great distance. Thus, while
cocoyams figured importantly in the diets of peasants in the South-East Akim
Abuakwa and Oda-Swedru-Asamankese regions, they were too costly to do so
in Kumasi and Sekondi-Takoradi.
Manioc, however, is a singularly versatile commodity. Unlike the other roots
and tubers, it need not move to market in fresh form. Quite commonly, especially
if the distance to market is great, it is processed into kokonte or gari before leaving
the farm. Both of these products compare favorably with the cereals in price and
caloric value per pound, and in addition have good storage qualities. They are,
moreover, not without appeal to bachelors and other busy urban consumers be-
cause both products are virtually ready to eat; with gari, one need merely add
water. Still other features commend manioc to a prime role in urban provision-
ment. The least demanding of the staples with respect to soil requirements, it
may be grown on land too poor to support cropping of any other kind; accord-
ingly it is admirably suited to the overworked land surrounding the cities. Then
too, it is a reliable year-round source of calories. Because its roots may be left in
the ground for as many as 24 months after reaching maturity they need be har-
vested only when required (28, passim).
The contribution the various foodstuffs make to the diet is, of course, only
weakly related to the position they occupy in the food budget. Thus, while
manioc and the other starchy staples easily make up the most important group
of foods consumed in Ghana's cities in terms of calories, their cheapness makes
their significance from the point of view of expenditures appreciably less (20,
pp. 22-23; 21, pp. 44-45; 19, pp. 48-49). This is clear from the following tabula-
Percentage of expenditures
on food and drink
Food group Accra Kumasi Takoradi
Meat and fish ........................... 34 42 36
Starchy staples ................ ......... 35 28 32
Vegetables and pulses ...................... 9 12 12
N uts, fats, and oils .......................... 5 7 8
Drink ......................... .......... 6 4 5
M miscellaneous .............................. 11 7 7
Total ................................. 100 100 100
In terms of percentage of expenditures on food and drink, meat and fish tend
to stand in the premier position; during the survey periods these two items to-
gether accounted for between 34 and 42 per cent of all outlays for food and drink.
True, expenditures on the starchy staples were almost as great; still their sec-
ondary role is clear. Compared with these two groups, outlays for other food-
stuffs were minor. The various nuts, fats, and oils, the second-ranking source
THOMAS T. POLEMAN
of calories, commanded but 5 to 8 per cent of expenditures, and t e vegetables
and pulses only slightly more. Drink and miscellaneous items accot nted for the
small remainder. These figures, of course, are averages for all income groups;
surprisingly enough, however, the general magnitudes to which they point
persist with remarkable constancy throughout the entire observed range of in-
Dietary Changes with Income
That the broad pattern of food expenditures changes little with income is
one of the most striking revelations of the three surveys. So com nc n among
budget investigations conducted in Western countries is the finding that a rise
in income is accompanied by a decline in relative expenditures on the cheaper
foodstuffs (particularly the starchy staples) and by an increase in those for the
more expensive commodities (most notably meat products), that the relationship
has come to be treated as virtually axiomatic. Not so in urban Ghan a, or at least
not so over the range of incomes for which we have data. As Chart I brings out,
the proportion of expenditures given over to the various commodity groups was
for all practical purposes completely independent of total food outlas. Whether
in Accra and Kumasi, where per capital food expenditures rose sharply with
income, or in Sekondi-Takoradi, where they did not, the findings wc re the same:
in the highest-expenditure classes as well as the lowest, the starchy staples, meat
and fish, and the residual group of other foods each accounted for a:0out a third
of the total food budget.
It follows from this that the income elasticities of demand .Lu. I.d by the
surveys were much the same for all of the major food groups (20, p. 4; 21, p. 38;
19, p. 42) :
"Income" elasticity of Imanida
Food group Accra Kumasi 'Takoradi
Principal starchy staples .................. .89 .82 .77
Principal meat products .................. .92 .72 .80
Principal fish products ................... .82 .68 .35
a On a per capital basis. Computed by the double-log method, total expenditures rather than in-
come having been used as the regressor.
In fact, the foregoing tabulation points to a demand for meat and f sh which is
no more income elastic (indeed perhaps a bit less) than the deman( for starchy
staples. This is exceptionally noteworthy. It is a widely held supj position that
most Middle Africans have a strong craving for animal proteins w;iich low in-
come prevents them from satisfying. That the immediate extent o7 this desire
may well be overestimated is implicit in our findings.
It also follows, given the over-all rigidity in the pattern of food e qpenditures,
that most of the dietary changes which accompany rises in income are the re-
placement of cheaper and less preferred products by more highly re artded ones
in the same commodity grouping. This is borne out by the data presented in
Charts 5A and B, though it must be noted that with respect to the st: rchy staples
the survey's findings on this point are rather less conclusive than might have
been expected. (The heterogeneous character of Ghana's urban popt nation prob-
GHANA'S URBAN FOOD ECONOMY 1I
CHART 4.-GHANA, URBAN HOUSEHOLD BUDGET SURVEY SAMPLES, 1953 AND 1955:
PERCENTAGE BREAKDOWN OF FOOD EXPENDITURES, AND THE PER CAPITAL
MONTHLY FOOD OUTLAYS, BY EXPENDITURE CLASS*
ACCRA (453 households)
Expenditure class limits
shown by vertical ficks.
100 200 300 400 500 600
100 200 300 400 500 600
TOTAL EXPENDITURES (SHILLINGS PER MONTH)
Based on data in Gold Coast, Off. Govt. Stat. (Stat. and Econ. Papers), 1953 Accra Survey of
Household Budgets (No. 2, December 1953), pp. 10-14; Kumasi Survey of Population and House-
hold Budgets, 1955 (No. 5, March 1956), pp. 36, 37, 42; Sel(ondi-Takoradi Survey of Population and
Household Budgets, 1955 (No. 4, March 1956), pp. 32, 33, 38. Positioning of data for extreme ex-
penditure classes reflects actual average expenditures; for other classes the midpoint. Figures do not
include expenditures on drink.
aA residual figure. In addition to such truly unspecified items as snacks purchased away from
home, it includes expenditures for the less important starchy staples, meat and fish items, vegetables,
and nuts, fats, and oils. (For Accra, of course, it includes all outlays for items in the latter two
groupings.) It is best thought of as being prorated among the major commodity groups.
b Packaged imported items. It includes such things as sugar and canned meat and milk, but not
wheat flour and the like.
THOMAS T. POLEMAN
CHART 5A.-GHANA, URBAN HOUSEHOLD BUDGET SURVEY SAMPLES, 1 53 AND
1955: EXPENDITURES FOR SELECTED STARCHY STAPLES AS A PERCENTAGE
OF TOTAL OUTLAYS FOR THESE ITEMS, BY EXPENDITURE CLASS*
0 100 200 300 400 500
TOTAL EXPENDITURES (SHILLINGS PER MONTH)
* See source note on opposite page.
a Kenkey and dough.
b Gari and kokonte.
GHANA'S URBAN FOOD ECONOMY
CHART 5B.-GHANA, URBAN HOUSEHOLD BUDGET SURVEY SAMPLES, 1953 AND
1955: EXPENDITURES FOR SELECTED MEAT AND FISH PRODUCTS AS A PERCENT-
AGE OF TOTAL OUTLAYS FOR THESE ITEMS, BY EXPENDITURE CLASS*
60 1 1
Smoked and dried fish
40 1 .1. .. Fresh meat, cheaper cuts
20 Fresh meat, dearer cutf
0 100 200 300 400 500 600 700
TOTAL EXPENDITURES (SHILLINGS PER MONTH)
Based on data in Gold Coast, Off. Govt. Stat. (Stat. and Econ. Papers), 1953 Accra Survey of
Household Budgets (No. 2, December 1953), p. 14; Kumasi Survey of Population and Household
Budgets, 1955 (No. 5, March 1956), p. 42; Sekondi-Takoradi Survey of Population and Household
Budgets, 1955 (No. 4, March 1956), p. 38. Positioning of data for extreme expenditure classes re-
flects actual average expenditures; for other classes the mid-point.
THOMAS T. POLEMAN
ably accounts for this result. With traditional food preferences differing from
one tribal group to the next, it may well be that some of the less signific nmt move-
ments tended to cancel one another.) Thus, at first glance it is difficult to per-
ceive in Chart 5A any clear-cut correlations between income levels ar d relative
expenditures on the individual staple foods. Indeed, closer examination fails to
reveal any significant tendencies whatsoever for yams and plantains, bot h thought
to be highly esteemed by most Ghanaians; while relative outlays for m iize prod-
ucts increased with income in Kumasi and Sekondi-Takoradi, bui declined
sharply in Accra. So too with rice, another comparatively dear comm adity con-
sidered to enjoy a preferred position among urban consumers. Only in Accra did
relative outlays for rice go up with income; in Kumasi no trend is evident, while
in Sekondi-Takoradi expenditures actually fell off.
Still, two consistent tendencies are discernible. In all three sample!, expendi-
tures for bread rose significantly with income. The trend was partici larly pro-
nounced in Kumasi and Sekondi-Takoradi; there, income elasticities o 2 03 and
1.74 are suggested, as against 1.01 for Accra.17 This preference for bread, by
far the most expensive starchy staple in terms of cost per 1,000 calories, almost
certainly mirrors the convenience it offers to urban consumers. Chiefly purchasedd
for meals to be eaten on the job at lunch time, it is easier to carry, simpler to eat,
and probably better liked than any of the prepared indigenous foods. Then too,
considerable prestige is attached to eating bread; it has become the 'modern"
thing to do, and Ghanaians are no more immune to the challenge cf keeping
up with the Joneses than anyone else (26, p. 14).
The only other consistent correlation relates to the dry manioc prod ucts, gari
and kokonte, and is an inverse one. Outlays for these products (in Sekondi-
Takoradi the data are for kokonte alone) declined sharply as inconr e rose, to
the end that in the highest-expenditure classes they commanded onl a trivial
portion of the food budget. This apparent aversion to gari and kokonte (despite
their ease of preparation) is most important, for, as noted above, manioc is prob-
ably the one foodstuff best suited to play a key role in the provisic nment of
Ghana's cities. Accordingly, it is almost with a sense of relief that o ie fails to
find evidence of a similar lack of preference for fresh manioc roots. I fact, the
Accra and Kumasi data suggest a tendency for fresh roots to be substituted for
the dry products as income rises.
The extent of the dietary adjustments brought on by these two i Icome-ex-
penditure relationships is illustrated in Chart 6 (which relates to the s ibsamples
in Kumasi and Sekondi-Takoradi from which quantity data were obtained).
Changes in the quantities of bread purchased were negligible. In Sekondi-
Takoradi, where the trend is clearest, apparent per capital consumption rose only
slightly-from 25 calories per day in the lowest-expenditure class to :iut about
double that amount in the highest. This, of course, reflects the costliness of bread;
rather substantial changes in expenditures are needed to appreciably alter the
quantity consumed. The opposite is true for gari and kokonte. A nong the
17 These elasticity figures have only limited significance. Much of the bread consume :d in Ghana
never reaches the home. but is purchased for snacks or for lunches on the job, and escal ed inclusion
in the bread item in the surveys. A comparison of flour import data with the bread qur nity figures
reported for Kumasi and Sekondi-Takoradi reveals a sharp discrepancy (26, pp. 10-11),
CHART 6.-KUMASI AND SEKONDI-TAKORADI BUDGET SURVEY SAMPLES, 1955, HOUSEHOLDS INTERVIEWED BY PERMANENT STAFF: PER
CAPITAL DAILY PURCHASES OF SELECTED STARCHY STAPLES AND MEAT AND FISH PRODUCTS, BY EXPENDITURE CLASS*
MEAT AND FISH
<200 200- 250- 300+ AVE. <200 200- 250- 300+ AVE.
250 300 250 300
EXPENDITURE CLASS (SHILLINGS PER MONTH)
Based on data in Gold Coast, Off. Govt. Stat. (Stat. and Econ. Papers), Kumnasi Survey of Population and Household Budgets, 1955 (No. 5, March
1956), p. 44; Sekondi-Takoradi Survey of Population and Household Budgets, 1955 (No. 4, March 1956), p. 40; caloric values calculated using conversion
factors given in the sources cited in the footnote to Table 5. Because data relate to principal foodstuffs only, average figures do not add to totals given in
a Kenkey and dough. b Gari and kokonte. c Kokonte.
THOMAS T. POLEMAN
cheapest sources of calories available in the city markets, a modest slift in ex-
penditures for them results in large quantity differences. We may concl ude, then,
that the only truly significant adjustment in the staple dietary of Ghana's urban
dwellers that can consistently be expected to accompany a rise in incorne is for
consumption of the dry manioc products to decline.
Chart 6 is revealing in another respect: it sheds light on a quest on which
has provoked a good deal of discussion among economists interested n Ghana.
In a series of articles published a few years ago in the journal of the Economic
Society of Ghana, Miss Polly Hill remarked on some of the findings of th, urban
budget surveys which she found puzzling (25). Chief among these w. s the sug-
gestion, evident at first glance in the data for Kumasi and Akuse,'8 hat rising
incomes were associated with a marked increase in the quantity of st iple foods
consumed. Thus, in Kumasi, she noted that households which spe it in total
400 shillings or more per month on all their domestic requirements, purchased
around 1,350 grams of the starchy staples daily per person, as oppose I t about
1,000 grams in households with total expenditures below 250 shillings. This ap-
parent tendency to eat "more in bulk the richer one gets" she termed tl e 'cassava
law""9 (25, Pt. 2, p. 6).
Unable to believe the "law" was actually true-for it implied that in even the
highest-expenditure classes people behaved as if they were getting ins alicient to
eat-Miss Hill attempted to offer an explanation. This she succeeded in doing,
quite convincingly, in terms of the African "extended family" system n and the
almost obligatory hospitality it entails. Where the traditional custor s are still
preserved, the canons of hospitality dictate that a man feed any of his o his wife's
myriad relatives who happen to visit him, no matter how long the) choose to
stay, and generally too those individuals who chance to call near meal ime. Rea-
soning that this "family parasitism" applied particularly in urban area;, t) which
a steady stream of single men come to seek their fortunes, Miss Hil suggested
that the apparent rise in staple-food consumption could most logic illy be ac-
counted for by increasing kinship obligations; as a man's income went up, more
and more relatives would find it convenient to call-a better meal would be
expected. The implications of this were far reaching. In Miss Hill's words, "if
it [the 'cassava law'] is to be explained in terms of gifts and hospitality, 1.hen the
fears of the economists and sociologists are indeed well founded-our urban ..
society is such that only the exceptional man, the man who is prepare I to ignore
his social obligations and civilities to his fellows, can better himself 1iy his own
efforts" (25, Pt. 3, p.7).
It would seem, however, that on this point at least, economists and sociolo-
gists can rest easy. Though admirably reasoned, the analysis contain: one error,
and it is a fatal one. For while it is undeniably true that the total quantity of
staple foods purchased per person increased with income in Kum; si (and in
Akuse), apparent per capital consumption expressed in terms of calories exhib-
ited no such tendency. In fact, a glance at Chart 6 reveals that, a ter having
increased somewhat in the lowest-expenditure classes, apparent per c: pita intake
of calories derived from the starchy staples actually declined in the highest two.
18 For reasons noted above the Akuse survey is not considered in this study.
19 In the former British West African Territories, manioc is generally called cassa a.
GHANA'S URBAN FOOD ECONOMY
(Worth noting is the general constancy of the calorie figure throughout the
entire range of incomes sampled in Sekondi-Takoradi). There is, of course, no
conflict. All that happened was that as income rose, the dry manioc products
were increasingly supplanted in the diet by the bulkier staples (chiefly plantains
and yams),20 thereby on a weight basis creating the impression of greater con-
sumption. Exit, then, the "cassava law" and all its discomforting corollaries: they
rest on an illusion.
Miss Hill also expressed puzzlement over the trends in meat and fish con-
sumption evidenced in the surveys. In this connection, it can only be reiterated
that the data fail to bear out the supposition that a rise in income is accompanied
by a sharp upturn in consumption of animal proteins. In Kumasi a rising trend
was evident, but it could hardly be considered the "strong consistent upward"
movement for which Miss Hill was looking; while in Sekondi-Takoradi appar-
ent intake in terms of grams was actually less in the highest-expenditure class
than in the lowest. This last is indeed surprising and, other than pointing out
that quantity data by expenditure class are available only for selected principal
foodstuffs and that in Sekondi-Takoradi particularly, Chart 6 does not include
figures for many of the more expensive products, I will make no attempt to
account for it. A thesis by Miss M. Crossley is perhaps worth noting, however (9,
p. 15). Observing that apparent meat and fish consumption tended to be highest
among households with incomes in the middle range of those sampled, she has
suggested that this might reflect some sort of "natural expression of the need"
for certain persons to eat more of the protein-rich foods than others-her not
unreasonable assumption having been that people with incomes in the middle
range were chiefly manual laborers, as opposed to clerical employees in the higher
But apart from this seemingly anomalous behavior in Sekondi-Takoradi,
there appears little basis for puzzlement. The data merely emphasize what has
already been stressed-that throughout the entire range of incomes observed,
the over-all dietary pattern changed little, meat and fish being no exceptions.
The consumer, according to his own particular likes and dislikes, improves his
diet as his income rises, but apparently feels no compulsion to alter its general
composition. Whether this behavior carries over into the highest-income brackets
is, of course, quite another matter. It may, or it may not; we simply do not
know.2' And, until additional and more representative investigations are carried
out, it would be a mistake to speculate.
The system through which foodstuffs flow into Ghana's cities is at the same
time both simple and involved: simple because most commodities reach the con-
sumer in exactly the same form as they leave the farm; involved because a given
item commonly passes through several markets and through the hands of many
intermediaries before arriving at it final destination.
20 The tendency for gari and kokonte consumption to decline with income is also evident in the
data for Akuse. See 28, p. 201.
21 It is interesting to note, however, that in their recent study of the elite in Nigeria, Hugh and
Mabel Smythe found that "food has little relevance to social status." "The elite," they go on to say,
"may be able to afford a finer cuisine than the rest of the population, but they usually adhere to the
standard dietary pattern" (39, p. 145).
THOMAS T. POLEMAN
The Market Chain
The shop and the shopkeeper, as known in Western societies, )lay only a
minor role in the economic life of Ghana. Instead, internal trade i: conducted
almost entirely through local markets. In the larger towns and citie: these mar-
kets are vast, relatively elaborate affairs, attended by many thousands >f persons"
and extending over several blocks. In the countryside they are far less pretentious.
The typical rural market occupies no more than an acre or two in the middle of
a village or in a clearing between villages, and is equipped with few, amenities
save some flimsy lean-tos. In Ghana, where markets are indigenous and highly
developed, most people live within a few miles of several such si:es. and on
market days they flock there to visit, court, and enjoy themselves a; well as to
These small rural markets serve as places for exchange both between local
farmers and between farmers and traders who purchase for resale in the towns
and cities. As feeder markets they are related to one another and to tie country's
consuming centers through what have been called "market circles." The name
reflects the fact that most farm people think of themselves not as being served
by just one market, but as being near the hub of four or more, each of which is
held on a different day. These circles generally overlap, so that m st markets
belong to several. The result is that the countryside is crisscrossed )y a sort of
chain mail of circles, making it possible for goods to move appreciab e distances,
while allowing those who trade in them to remain comparatively clc se to home.
Usually within each circle or series of circles there is at least one m; ior market.
These larger country markets, now invariably reached by a road of some type,
form the funnels of the exchange system, channeling processed goods to their
satellite markets and acting as collecting centers for foodstuffs destined for the
cities. Such a bulking point is the market at Asesewa, located sore 60 miles
north of Accra in the heart of an important maize-producing region. Here, on
Monday and Fridays, a vast multitude assembles, including an average of about
4,000 persons-local growers as well as intermediaries who have bought in out-
lying markets-selling only foodstuffs. Trading with them are bulk buyers who
come with empty trucks from all parts of the country (but especially Accra) and
who ship directly to the cities. It is estimated that a minimum of 150 trucks visit
Asesewa on market days and that (because those traveling short dist inces make
several trips each day) the number of daily departures averages about 80( (29, pp.
Once the trucks reach the cities, the loads which have been so laboriously
assembled are quickly broken down. Generally this bulk-breaking; process is
carried out by a further series of intermediaries, each one dealing in : smaller lots,
until the retail stall-keeper or hawker will offer for sale no more than a few piles
of the commodities in which she trades. The upshot is that by the time an item
is purchased by the urban consumer it has passed through at least th-ee markets
and been handled by perhaps a half-dozen intermediaries; and it is by no means
unusual for double those numbers to have been involved.
This trade in foodstuffs is managed entirely by Africans. In Ghana, as in
22 Of sellers alone, Accra's 11 markets count well over 25,000 (34, p. 7).
GHANA'S URBAN FOOD ECONOMY
most other countries of western tropical Africa, an energetic group of traders
has grown up, its members handling all stages of food collection and distribu-
tion. This class has long existed; but, encouraged by the trend toward urbani-
zation and the concomitant expansion in the road network and cash cropping,
it has grown greatly in size and affluence since the war. Today, the speeding
trucks of African entrepreneurs are a characteristic feature of the Ghanaian scene,
each one representing a major investment of hard-won savings, and each one
attesting to the enterprise, if not the skill, of a new commercial elite.
Surprisingly enough to the outsider, the greater part of the food trade is con-
ducted by women. With few exceptions-livestock products are the most signifi-
cant-they dominate all of its aspects. It is the farmer's wife, not the farmer, who
walks up to five or ten miles each day to market a headload; it is the "mammy
trader" who assembles these small lots and sends them to town; it is she who
prevails in the great city markets; and more often than not it is to one of her more
prosperous sisters that the fancifully decorated truck-"mammy wagon" is not
a misnomer-will belong. How women came to control this crucial sector of
the economy is not altogether certain, though it doubtless in part reflects a legacy
from a time (not so many years ago) when only women could venture far from
their villages with impunity. But be that as it may, it is the women who now
dominate the market place. "The market mammies occupy a very important
position in this semi-capitalistic society of ours," President Nkrumah has been
quoted as saying (35, p. 139). If anything, he is guilty of an understatement.
This indigenous system of urban provisionment is a source of considerable
pride to Ghanaians, and justifiably so. Nonetheless, it has not been free of criti-
cism (cf. 35, p. 53; 7, pp. 22-23). Not a few observers have contended that be-
cause the system lacks the centralized direction and the greater capitalization
which the European firms and marketing boards dealing in export crops can
bring to bear, it has been rather less effective than these entities in calling forth
production. Also, because of the lengthy chain of intermediaries involved, some
have argued that the system fosters unduly large differentials between producer
and consumer prices.
As to the second of these criticisms, it may be noted that the number of its
supporters has fallen off sharply within the last decade. For while the system
is undeniably primitive from a technological point of view, as P. T. Bauer has
emphasized at length (2), its heavy reliance on labor inputs is admirably suited
to the present state of Ghana's economy. "So far from being wasteful," in his
words, "it is highly economic in saving and salvaging those resources which are
particularly scarce in West Africa (above all, real capital) by using those re-
sources . [i.e., labor] for which there is very little demand . ." (2, p. 22).
Increasingly it has come to be accepted that the large number and long line of
intermediaries must find ample justification in the economies to be obtained
from bulking and later breaking tens of thousands of tiny lots; otherwise they
would have long since been bypassed.
More serious is the contention that the system is cumbersome and sluggish,
that it fails to elicit from food producers the same response as do the more
centralized and sophisticated buying organizations which purchase export crops.
Here, regrettably, most of the argument on either side has lacked solid support
THOMAS T. POLEMAN
in empirical fact-and for good reason. Unlike the export trade, their e has been
for food movements an extreme paucity of data. Few are the sources tc be tapped:
hardly any of the thousands of traders keep records-even those who can read
and write-and because most movements are internal, customs figures are of little
value. Moreover, it has not been until the last few years that the trad: has posed
any serious problems for government administrators; few of them, accordingly,
have felt any particular compulsion to encourage study of it.
Fortunately, a truly major step was recently taken toward correctir g this state
of affairs. This was a year-long census of produce movements conducted in 1957
and 1958 by the Office of the Government Statistician. Thanks to th s census it
is now for the first time possible to build up a reasonably complete pii ture of the
sources and quantities of the foodstuffs which flow into one of Ghan 's cities. It
is also possible to infer something about the efficiency of the system n through
which they move. Strongly suggested is that the system permits both flexibility
and regional specialization among supply areas, with the outflows 'rom these
areas changing over smoothly and efficiently from one season to the r ex:.
The 1957/58 Produce Movement Census
Despite the fact that the census of 1957/58 yielded the first really serviceable
results, the collection of food-movement statistics in Ghana has, by Middle Afri-
can standards, a comparatively long history. Since as early as 1929 re words have
been kept of shipments passing over some of the Volta River ferries. I addition,
on several occasions during the early 1950's both the Office of the G ,vernment
Statistician and the Economic Research Division of the University C Olege con-
ducted brief traffic checks in the neighborhood of some of the lai ger towns
The data obtained, however, proved none too instructive: those from the ferry
crossings mirrored only a tiny fraction of the total volume of foodstu Ts moving
toward the cities; those from the temporary surveys related to peri d;d of not
more than two-weeks duration. This situation the Government Statistician set
out to rectify when, at the outset of the 1956/57 fiscal year, his Office w is assigned
exclusive responsibility for assembling statistics on produce flows. Beginning on
October 1, 1957 the number of checking stations was greatly increase, and con-
tinuing through September 30, 1958 rather more information was collected at
each than had been ever before.
Like the various short-term surveys which preceded it, the 1957/58 :ensus was
basically a record of the quantity of foodstuffs carried by truck." In all, 20 check
points were involved, a majority of them centered around Accra in the south-
eastern corner of the country (Map 2). Recording was done day and night, and
at most of the stations all produce carriers were stopped. Traffic wa; so heavy
at seven of the check points, however, that only sampling was feasible; at the
six stations on the outskirts of Accra and Nsawam every tenth truck was halted,
as was every third one passing the Asankare station. Each truck st ,pped was
23 The findings of some of these short-term surveys are discussed in 45 and 32.
4 Rail movements were ignored because of their trivial quantity. During the 1)53/54 fiscal
year, rail shipments of specified livestock and food items (ex-cocoa) amounted to onl. 3 077 long
tons (22, p. 55).
MAP 2.-GHANA, PRODUCE MOVEMENT CENSUS, 1957/58: ANNUAL FLOW OF STAPLE FOODS
PASSING THE SEVERAL CHECK POINTs*
NO T H E N R EG I N
y ^^A .5 H A N T I
Axi G1 TaoradO -
Major Noad 4s a
Ot Ke Accra
a THOt/5AND METRIC TONS d t 1tD
MIL-sTr Ij --s---
Ai THOAND kETR TraONd
Basedi on unpublished returns from the 1957/58 Produce Movement Census. Where data re-
ported other than by weight, tonnages were estimated using the following conversion factors: maize-
bag: 220 lbs.; rice-Gbag: 240 bs.; millets-Nbag: 204 bs.; sorghum-Abag: 200 bs.; manioc products-
bag: 150 lbs.; manioc roots-bag: 200 lbs.; cocoyams-bag: 150 lbs.; and yams-tuber: 7 lbs. Factors
from Ghana, Off. Govt. Stat., "Distribution Statistics-Produce Movements. . Conversion Fac-
tors" (n.d., mimeo.).
Movements of less than 400 metric tons not shown.
tors" (n~d., ininien.).
THOMAS T. POLEMAN
subjected to a thorough examination: origin and destination wern ascertained,
as was the size and composition of the cargo. This information was then entered
into specially prepared forms, and at the end of each month summarized and
forwarded to the central office in Accra.
Unfortunately, at this point an otherwise well executed prof ram bogged
down. Owing to an acute shortage of trained personnel, the cent -al office was
unable to give the incoming returns anything like the attention tl ey deserved;
unappraised, they were merely placed in files, and even now official analysis is
thought to be some time away (5). The Government Statistician has, however,
made the returns available to private scholars. Thanks to the courtesy of the
Government Statistician and of my colleague, Mr. Marvin P. Mira, le, I am able
to set forth here some of the findings.2" These relate to the starch" staples-the
foods which, we have seen, comprise the basis of the Ghanaian diet.
Map 2 summarizes the total flow of staple foods recorded durir g the census.
The over-all view it provides is revealing in several respects. Strikingly brought
out, in the first place, is the trivial extent to which the consuming :enters of the
south draw on the northern half of the country for their staple iood supplies.
Apart from yams-the movements of which from the Kete Krachi area to Accra
were recorded at the Otisu station-the region exported nothing of consequence.
Indeed, the data for the Bamboi, Buipe, and Yeji ferries indicate he Northern
Region was actually a net importer of staples during the year. This is unusual,
however (16, p. 23). The 1958 season was an uncommonly dry one in the north,
and during August and September the government was obliged :o send there
several thousand tons of maize and manioc products.
The map also clearly points up the urban areas on which the census sheds
light. It will be seen that there were no check points in the vicinity cf Kumasi;
for this city, accordingly, little can be inferred. Much the same applies to Sekondi-
Takoradi: the two stations in its environs were located rather too far (about 20
miles) from the urbanized area to record the total volume of moven tents destined
thereto. For Accra, on the other hand, the check points were admiL ably situated.
Four main roads link Accra to its hinterland, and on each of these a station was
set up near the point where it crossed the municipal boundary. As food produc-
tion within the city limits is inconsequential and may safely be ign red, the data
collected at these four points, plus that assembled at outlying static ns, permit us
to construct a rather detailed outline of the origins and quantities of the staple
foods consumed in Accra.
The Staple Food Supply of Accra
In all, the data indicate, a total of 144,000 metric tons of staple f( odstuffs were
imported into Accra during the census year. This total is altogethe- a reasonable
one, and as such reflects favorably on the accuracy of the census. It represents,
as Table 8 shows, the equivalent of 282 billion calories a year, o assuming a
1957/58 population for Accra of about 370,000 persons, something like 2,088
25 Because of the unique value of the census, the Food Research Institute has p spared a 20-page
mimeographed summary of the basic statistics copied by Mr. Miracle. This includes a monthly, com-
modity-by-commodity statement of all starchy staples passing each of the 20 check points. Requests
for copies should be addressed to the Institute.
GHANA'S URBAN FOOD ECONOMY 167
TABLE 8.-GHANA, PRODUCE MOVEMENT CENSUS, 1957/58: NET ANNUAL INFLOW
OF PRINCIPAL STAPLE FOODS INTO ACCRA*
Net Annual Inflow
Foodstuff (Thousand metric tons) (Billion calories)
Plantains .............................. 41.6 31.2
Maize" ................................ 32.5 115.7
M anioc products ....................... 26.8 90.8
Fresh manioc ................ ....... 22.0 24.0
Bananas ............................... 5.5 3.7
Y am s ................................. 4 .5 4 .0
Cocoyams ............................. 2.2 1.9
Rice .................................. .6 2.0
O others ............... .... ......... 8.3 8.7
T otal ............... .......... 144.0 282.0
Based on unpublished returns from the 1957/58 Produce Movement Census; caloric values
calculated using conversion factors given in FAO, Food Composition Tables-Minerals and Vitamins-
For International Use (Nutr. Studies 11, Rome, 1954), pp. 10-12, 18.
a Includes small amounts of sorghum.
b Millets, "other cereals," and "other roots and tubers."
calories per person per day. While this last figure at first glance suggests a level
of consumption considerably greater than that pointed to by the budget survey
data for Kumasi and Sekondi-Takoradi (Table 5), most of the difference is
surely illusory. Budget surveys by their nature invariably understate actual food
availability and, as noted earlier, it is not unlikely that those for Kumasi and
Sekondi-Takoradi did so by as much as from two to four hundred calories.
Furthermore, it cannot be assumed that the full amount of the tonnage arriving
in Accra was actually consumed. Some loss inevitably occurred, and though
we have no basis for estimating how great this might have been, 10 per cent
might not be an unreasonable guess.
As to the composition of this inflow, Table 8 offers further evidence of the
limited number of staple foodstuffs on which urban Ghanaians depend for the
bulk of their calories. Of the 395 metric tons of staple food products which
arrived on the average in Accra each day, fully 85 per cent was made up of only
four items: plantains (29 per cent; 114 tons), maize (22 per cent; 89 tons), proc-
essed manioc (19 per cent; 74 tons), and fresh manioc roots (15 per cent; 60
tons). Together these four items accounted for 93 per cent of the calories im-
ported in the form of starchy staples-though by this criterion their individual
importance was somewhat different. Because of their high water content, the
combined caloric contribution of plantains and fresh manioc was but one-fifth
of the total, whereas maize alone accounted for 41 per cent and the dry manioc
products 32 per cent.
The areas from which Accra drew its supplies of these four main staples
during the census year are shown in Map 3. This map, of course, is somewhat
generalized, and as such should not be interpreted too literally. Since origin was
recorded by administrative region rather than exact location, the dotted lines
are meant only to be suggestive (although in positioning them I have made full
MAP 3.-GHANA, PRODUCE MOVEMENT CENSUS, 1957/58: DIRECTION AND ANNUAL MAGNITUDE OF INFLOW
OF PRINCIPAL STAPLE FOODS INTO ACCRA*
S .- ; PLANTAINS
Net annual inf ow:
/ % \
31 billion catorL e
"FRESH MAN IOC
+ Nat annual inflow:
22,0ooo0 etric tons
N' ^ "-^^^^ L71IiI >^ 8 24 billion calories
I .6 0V1 '" ~o -
,Net annual inflow:
SA 5 33,000 _etric ton"
116 billion catories
116 bkllwss c8IoriOJ
S- "-,' PRODUCTS
/ "' " r' \
.Net annual inflow:
n". I^ ,' "7
'(-- '^ ^^ Net annu~la inflow:
271000 metric tons
-" n ^ 91 billion calories
* Based on unpublished returns from the 1957/58 Produce Movement Census.
GHANA'S URBAN FOOD ECONOMY
use of available qualitative evidence). Otherwise, however, the findings of the
census are faithfully portrayed. Where discrepancies occurred between check
points (as happened only twice), the differences are indicated by a question mark.
Two things stand out in this map. One is the clear indication it gives of the
distance over which the pull of the Accra market asserts itself. The area from
which the city draws the foodstuffs extends far beyond its immediate surround-
ings in the Coastal Savanna. Indeed, the role this region plays is decidedly a
secondary one, with the Forest and beyond supplying the bulk of Accra's require-
ments. About 55 per cent of the tonnage of the four main staples which arrived
in the city during the census year came from places over 50 miles away, and
some 30 per cent traveled 100 miles or more.
The other striking feature of Map 3 is the impression it conveys of the extent
to which local specialization in commercial food cropping takes place. Apart
from the Forest area to the northwest of the city, which supplies both fresh
manioc and plantains, none of the regions provisioning Accra ships appreciable
quantities of more than one major staple food item. Thus, while plantains and
fresh manioc roots move in principally from the northwest, most of the city's
maize comes from the north and northeast,26 and the bulk of its processed manioc
is imported from Togo to the east. That processed manioc should be brought in
from Togo rather than from the same region as fresh roots warrants a word of
explanation. The dry manioc products have achieved widespread acceptance
among Ghanaians only within the past two or three decades. In Togo (and
Dahomey), on the other hand, they have long been important staples; there,
gari and kokonte processing has developed into a major industry, whereas in
Ghana it is still in its infancy, with the quality of the output often decidedly
inferior (28, pp. 77-80).
Togo also supplies maize to the Accra market; just how much, unfortunately,
it is not possible to say. On this point, as the map shows, the evidence provided
by the census seems conflicting. Specifically, the Tefle Road check point recorded
far more maize arriving in the city from Togo than was noted at either the Tefle
ferry or Afloa, the other two stations on the main road to Togo. This may or
may not be valid; arguments can be offered both pro and con. The frontier be-
tween Ghana and Togo is long and poorly policed, and though the road through
Afloa offers the quickest route to Accra, it is not unlikely that other routes are
employed near the border to avoid payment of duties. Then too, shipment by
water is not out of the question. On the other hand, however, the accuracy of
the tally at Tefle Road cannot be considered above suspicion-if only because
it was the sole check point that yielded data which cannot readily be reconciled
with those of the others. As at the other three stations on the outskirts of Accra,
checking at Tefle Road was on a sample basis, with one-tenth of the passing
traffic having been stopped. The checkers were instructed to stop not any one
of ten vehicles, but every tenth one. Conceivably, this order was improperly exe-
cuted and a biased sample obtained. Much of the traffic from Togo reaches Accra
late in the day when other traffic is light. Perhaps the checkers found it con-
26 In line with the Department of Agriculture's belief (12, p. 90), I have indicated the Asesewa
area as the chief source of maize in the Eastern Region.
THOMAS T. POLEMAN
venient to stop many trucks at that time to make up for omissions during the
busier hours. This possibility notwithstanding, however, the price dat, fcr maize
shown in Chart 7 suggest that the Tefle Road data are essentially corr :ct. It will
be seen that maize prices in Accra were at a minimum during the October-
January period, the same months when most of the maize was record -d at Tefle
Road. Were the bulk of this inflow from Togo illusory, it would be difficult to
account for any decline in price; the opposite should have obtained.
The procedure followed in this study in computing aggregate inflow esti-
mates for Accra has been to accept the questioned Tefle Road figure s for both
maize and processed manioc as they stand. Until official analysis )f the full
findings of the census is undertaken, we have no firm grounds for ding other-
wise. It should be noted, however, that should the official appraisal reveal the
Tefle Road figures to be in error by the amounts indicated in the ma it would
have the effect of reducing the inflow totals given in Table 8 rather d as':ically-
by about 70 billion calories, or something like 500 calories per person per day.
The various commodity flows illustrated in Map 3 are, of course, totals for
the census year. As would be expected in view of the seasonal charact :r of much
of Ghana's agriculture, the individual magnitude of these flows varied from one
time of the year to another. This is borne out both by the small ch.Lrts within
the map and by Chart 7, which summarizes on a monthly basis the iet arrivals
of the several staples in Accra.
Chart 7 is particularly worthy of examination: it contains the inly direct
evidence we have of the extent and nature of the seasonal changes Nx which occur
in the diets of urban Ghanaians. These, it suggests, are quite sizable Arrivals
of both maize and processed manioc, the two main suppliers of calorie s in Accra,
varied greatly over the year, as did those of plantains, the third-rank ag source.
Plantains may be harvested virtually all year, the explanation for the varia-
tion in shipments probably lies in alternative uses for trucks. Most cf tie plan-
tains consumed in Accra come from Ashanti and the forested area lyil.g between
that region and the city. These two areas also account for the bulk o the coun-
try's cocoa production. The main season during which cocoa is harvested and
shipped to the coast extends from September through early January. It is during
these months that plantain shipments to Accra were at a minimum.
The apparent variations in maize and processed manioc consump ion almost
surely reflect a combination of seasonal supply conditions and consumer prefer-
ences. Maize is harvested in southern Ghana and Togo during two periods of
the year: midsummer, when the early crop of the forested regions m: lu:es; and
October-January, when both the second crop from these areas an. the only
crop of the Coastal Savanna are collected. The result is that maize is iil decidedly
seasonal supply in the Accra market. Prices respond accordingly (Chart 7, bot-
tom section). During the census year the price of maize was compare itively low
between August and January, while it rose sharply during the sprint and early
The effect of these price fluctuations on processed-manioc consume tion seems
clear. Though manioc is available for harvesting throughout the yelr, arrivals
of the processed products in Accra were markedly seasonal. As the price of
maize rose in the spring (to the point where it became rather more expensive
CHART 7.-GHANA, PRODUCE MOVEMENT CENSUS, 1957/58: NET MONTHLY
INFLOW OF STAPLE FOODS INTO ACCRA; AND AVERAGE MONTHLY
WHOLESALE PRICE OF MAIZE AND GARI IN THE ACCRA MARKET*
O N D J F M A M J J A S
Net inflows based on unpublished returns from the 1957/58 Produce Movement Census; price
data from Ghana, Dept. Agr., Monthly Report-Foodstuffs Supply Position (various issues); caloric
values calculated using conversion factors given in FAO, Food Composition Tables-Minerals and
Vitamins-For International Use (Nutr. Studies 11, 1954), pp. 10-12, 18.
a Includes bananas.
b Includes sorghum.
c Rice, millets, yams, cocoyams, "other cereals," and "other roots and tubers."
--i Manioc products
THOMAS T. POLEMAN
in terms of calories than gari), imports of the dry manioc products increased
greatly-to the extent, even, that they appear to have temporarily supplanted
maize as the chief source of calories-only to drop off again a few n months later
when the relative price of maize declined. This complementary price-inflow
relationship between maize and processed manioc is highly revealing. Not only
does it offer further evidence of the less preferred position the dry m nioc prod-
ucts occupy in the diet, it also strongly attests to the efficiency with which the
marketing system responds to local price and supply conditions. Th f exibility
of the system seems such that urban consumers may confidently expe t to receive
adequate supplies of the cheapest energy sources available, no matt r what the
season of the year and no matter what their place of origin.
The foregoing picture of the food economy of Ghana's cities is, if course, a
partial and introductory one. Given the paucity of evidence on eco lomic phe-
nomena in Middle Africa, it could hardly be otherwise. Nonetheles;, examina-
tion of the limited data now available has provided us with certain useful in-
sights. Several of these, because of the extent to which they run contraryy to
previously held notions, seem important enough to justify reemphas s.
Among these, to be sure, are the broad conclusions which ma) be drawn
regarding the functioning of the supply system. Despite its technological primi-
tiveness, the efficiency of this system seems beyond doubt. Far from limiting
Ghana's urban centers to hinterlands within a radius of twenty or t ir-y miles,
it has, in conjunction with the recent improvement in the road network, enabled
them regularly to draw staple foodstuffs from supply areas 100 milus or more
away. Moreover, it has permitted producers in these areas to special ze in those
crops for which their lands are best suited, and at the same time demonstrated a
remarkable agility in responding to changing seasonal conditions. Manifestly
it is a thing of value, as are the incentives which make it possible; caut ous indeed
should be any attempts (such as appear to be contemplated) to sup elementt or
replace it with state controlled "cooperative" marketing ventures.
But if the operation of the supply system provides ample indi action that
Ghanaians are capable of responding efficiently and rationally to ec ,nomic in-
centives, it does not follow that one can safely anticipate their futui e behavior
merely through analogies with Western experience. This, perhaps, s the most
significant revelation of the budget surveys. Particularly with respect to the type
of dietary adjustments which accompany rises in income, a point of cru cial future
importance, we have observed a sharp divergence between reality andcl expecta-
tion: instead of incorporating greater amounts of animal proteins into their diets,
the wealthier urban consumers apparently eat essentially the same ty pe of meal
as their less fortunate neighbors. The implications of this, of course, are all to
the good from the point of view of production. Because the starchy staples are
much easier and cheaper to produce than animal products, a continuation of
present dietary patterns will lessen the demands on agriculture posec by urban-
ization. But why this rigidity should obtain remains obscure. Conceivably a
whole new dietary must evolve (comparable to that which occurred a generation
GHANA'S URBAN FOOD ECONOMY
or so ago in the United States when the salad course achieved widespread ac-
ceptance) before any appreciable changes in its composition can take place. But
whatever the explanation, the pitfalls of blind extrapolation are obvious. Middle
Africa is a world of its own. Essential to the accommodation of the problems of
its future growth is an understanding of its present.
1 Ione Acquah, Accra Survey (London, 1958).
2 P. T. Bauer, West African Trade (Cambridge, 1954).
3 M. K. Bennett, The World's Food (New York, 1954).
4 W. B. Birmingham and G. Jahoda, "An African Public Opinion Survey," in
Proceedings of the Fourth Annual Conference of the West African Institute of Social
and Economic Research (Ibadan, 1956).
5 W. L. Booker, Government Statistician, Accra, Ghana, letter to M. P. Miracle,
dated Aug. 5, 1960.
6 Friedrich Burgdorfer, ed., World Atlas of Population-Distribution of the Popu-
lation on the Earth about the Year 1950 (Hamburg, 1954).
7 K. A. Busia, Report on a Social Survey of Sekondi-Takoradi, Gold Coast 1947-48
8 A. W. Cardinall, The Gold Coast, 1931 (Accra, n.d.).
9 M. Crossley, "Some Puzzling Spending Habits in Ghana: A Comment," Econ.
Bull. (Accra), Mar. 15, 1958.
10 M. Fortes, R. W. Steel, and P. Ady, "Ashanti Survey: 1945-46: An Experi-
ment in Social Research," Geog. I. (Roy. Geog. Soc., London), Oct.-Dec. 1947.
11 Ghana, Dept. Agr., Annual Report . for the Year 1955-56 (1958).
12 "Miscellaneous Information 1958/59 Estimates" (mimeo, n.d.).
13 Ghana, Off. Govt. Stat., Quarterly Digest of Statistics, June 1960.
14 Benjamin Gil and K. T. De Graft Johnson, "The Post-Enumeration Survey in
Ghana," Econ. Bull. (Accra), April 1960.
15 Gold Coast, Dept. Agr., Annual Report . 1949-50 (1951).
16 Ibid., 1953-54 (1956).
17 Gold Coast, Depts. Med. and Ed., Gold Coast Nutrition and Cookery (1953).
18 Gold Coast, Off. Govt. Stat., Census of Population, 1948, Report and Tables
19 Kumasi Survey of Population and Household Budgets, 1955 (Stat. and
Econ. Papers No. 5), March 1956.
20 -- 1953 Accra Survey of Household Budgets (Stat. and Econ. Papers No.
2), December 1953.
21 Sekondi-Takoradi Survey of Population and Household Budgets, 1955
(Stat. and Econ. Papers No. 4), March 1956.
22 Gold Coast, Railway and Harbour Authority, Administration Report... 1953-
23 L. M. Goreux, Income Elasticity of the Demand for Food-Household Survey
Analysis (UN, Econ. Commis. Europe, Com. on Agr. Prob., AGRI/WP.7/2, 22 June
24 Faye W. Grant, "Nutrition and Health of Gold Coast Children," 1. Am. Diet.
Assn., July 1955.
25 Polly Hill, "Some Puzzling Spending Habits in Ghana," Econ. Bull. (Accra),
June 1957, Pt. 1; November 1957, Pt. 2; and December 1957, Pt. 3.
26 B. F. Johnston, The Outlook for Wheat and Flour Imports in Tropical Africa
(U.S. Dept. Agr., For. Agr. Serv., February 1959).
27 -- The Staple Food Economies of Western Tropical Africa (Stanford,
THOMAS T. POLEMAN
28 W. O. Jones, Manioc in Africa (Stanford, Calif., 1959).
29 S. LaAnyane, "Aweso-A Manya Krobo Huza" (Ghana, Dept. Ag;., mimeo.,
30 C. W. Lynn, Agriculture in North Mamprusi (Gold Coast, Dept. Ag r. Bull. 34,
31 Lucius Nicholls, Tropical Nutrition and Dietetics (3d ed., London, 1951).
32 B. M. Niculescu, "Preliminary Report to the National Food Board on Food
Supplies to Sekondi-Takoradi" (Econ. Res. Div., Univ. Coll. of the G ld Coast,
mimeo., July 13, 1955); and "Second Report . (Dec. 15, 1955).
33 K. A. J. Nyarko, "The Development of Kumasi," Bull. of the Ghana Geog.
Assn. (Legon), January 1959.
34 Astrid Nypan, "Market Trade in Accra," Econ. Bull. (Accra), Mar:h 1960.
35 E. J. Pedler, Economic Geography of West Africa (London, 1955).
36 E. R. Rado, "A Social and Economic Survey of Bentsir Quarters, Cape Coast,"
in Proceedings of the Third Annual Conference  of the West African Institute
of Social and Economic Research (Ibadan, mimeo., 1956).
37 Royal Institute of International Affairs, Ghana: A Brief Political ana Economic
Survey (London, May 1957).
38 Dudley Seers and C. R. Ross, Report on Financial and Physical P oblems of
Development in the Gold Coast (Gold Coast, Off. Govt. Stat., 1952).
39 H. H. Smythe and Mabel M. Smythe, The New Nigerian Elite (S:anford,
40 A. C. Sutherland, "Private Enterprise Housing in the Urban Area o0 Kumasi,"
in Proceedings of the Fourth Annual Conference  of the West Africaen Institute
of Social and Economic Research (Ibadan, mimeo., 1956).
41 Uganda, E. Afr. Stat. Dept., The Patterns of Income, Expenditure and Con-
sumption of African Unskilled Workers in Mbale, February, 1958 (1958).
42 United Nations (UN), Dept. Econ. and Soc. Aff., Economic Surve o Africa
Since 1950 (1959).
43 W. J. Varley and H. P. White, The Geography of Ghana (London. 1958).
44 H. P. White, "Environment and Land Use in the South Eastern S vannas of
the Gold Coast," in W. Afr. Inst. of Soc. and Econ. Res., [Fifth] Annual Conference
Proceedings, March 1956 (Ibadan, mimeo., 1956).
45 "Internal Exchange of Staple Foods in the Gold Coast," E on. Geog.,
GHANA'S URBAN FOOD ECONOMY
MANIOC IN AFRICA: CORRECTION
Thomas T. Poleman's intensive study of the food economies of Ghanaian
cities, published in this issue of Food Research Institute Studies, has uncovered
two errors in the construction of Table 8-4 of my Manioc in Africa (Stanford,
Calif., 1959). This table, entitled "Ghana: Average Daily Purchases of Manioc
and Other Starchy Staples in Five Communities, 1954-56," is based in part on
the same sources as Mr. Poleman's Table 5 and Table 7. My figures for Kumasi
and Sekondi-Takoradi were based on summary tables in the original sources
which failed to include some foodstuffs purchased in small amounts-krako,
manioc dough, shelled corn (maize), prepared plantain, and prepared rice in
both cities and gari in Sekondi-Takoradi. Inclusion of these products increases
the total calories derived from starchy staples from 1,318 to 1,364 in Kumasi and
from 1,104 to 1,260 in Sekondi-Takoradi. Calories from manioc as a per cent of
all calories from starchy staples, however, are unchanged.
The data shown for South-East Akim Abuakwa in my table are also in error,
apparently because they were calculated with an American short ton rather than
the English long ton. The correct figures are given in Mr. Poleman's Table 7
and in my Table 8-5. They show manioc to account for 50 per cent of all calories
from starchy staples instead of 46 per cent as shown in my Table 8-4.
n1s"N.Yn "0 - : 1,