Tegemeo Institute of Agricultural

Policy and Development

Working paper 19

STAPLE FOOD CONSUMPTION PATTERNS IN URBAN KENYA:

TRENDS AND POLICY IMPLICATIONS

By

Milu Muyanga, T.S. Jayne, G. Argwings-Kodhek, Joshua Ariga

Tegemeo Institute of Agricultural Policy and Development, Egerton University.

P.O Box 20498 Nairobi.
Tel: (02) 2717818
Email: egerton@tegemeo.org

Support for this research has been provided by the Tegemeo Agricultural Monitoring and Policy Analysis
Project (TAMPA) between Tegemeo Insitute/Egerton University and the Department of Agricultural Economics
at Michigan State University. Financial support for this project is provided by the Kenya Mission of the United
States Agency for International Development. Additional support is also provided by MSU through resources
from the Food Security III Cooperative Agreement.
 Table of contents

PAGE

ABSTRACT ................................................................................................................................. IV
INTRODUCTION ..................................................................................................................... 1
DATA AND METHODS .......................................................................................................... 1
Data ........................................................................................................................................1

Weighting Procedure .............................................................................................................2

Methods..................................................................................................................................4

RESULTS .................................................................................................................................. 5
Staple Carbohydrate Consumption ........................................................................................5

Staple Carbohydrate Expenditures.........................................................................................8

Disaggregating the Forms of Maize Consumption ..............................................................11

Reasons for the Recent Shift toward Sifted Maize Meal.....................................................15

Consumers’ Use of Alternative Marketing Channels for Primary Staple Products.............18

CONCLUSIONS AND POLICY IMPLICATIONS ............................................................... 20

REFERENCES ............................................................................................................................. 23

ii
LIST OF TABLES

Table 1. Consumption of Primary Staples (kgs per adult equivalent) ..............................................6

Table 2. Staple Carbohydrate Expenditures Shares among Nairobi Consumers, 1995 vs. 2003. ........8

Table 3. Expenditures on primary Staple Commodities, Nairobi......................................................9

Table 4. Staple Carbohydrates Real Prices (Ksh per kg), 1995 and 2003. .......................................10

Table 5. Consumption and Expenditures of Main Carbohydrate Products ......................................13

Table 6. Primary Reasons for Consuming Alternative Forms of Maize Meal, 2003 .......................16

Table 7. Consumers’ Use and Market Share of Alternative Retail Channels for Primary Staples .....19

LIST OF FIGURES

Figure 1: Per Adult Equivalent Consumption of Staples (Kgs).................................................7

Figure 2: Expenditures on Primary Staple Commodities . .......................................................9

Figure 3. Expenditure Shares in Per Adult Equivalent Income ..............................................11

Figure 4: Household Heads Economic Activities and Maize Meal Consumption ..................12

Figure 5. Extra Costs If Households Consuming Posho Meal were to Shift to Sifted Meal ..14

Figure 6. Inflation-adjusted Prices of Retail Maize Grain, Posho Meal, and Sifted Meal. ....17

Figure 7. Alternative Retail Channels for Sifted Maize Meal, 2003 ......................................20

iii
 Abstract
This study examines current consumption patterns of the main staple carbohydrate products
in Nairobi -- maize, wheat, rice, and cooking bananas -- in an effort to update policy makers’
knowledge of current urban food consumption patterns. The study also identifies the factors
driving changes in the amount and form of urban maize meal consumption, in order to better
understand how food security policy should be designed to respond to the needs of low-
income consumers. The study reveals how consumption and expenditure patterns differ
according to household income. The study also examines whether the marketing channels
used by the poor to secure their staple carbohydrate products differ from those used by
relatively high-income consumers. The data used in this study comes from a cross-sectional
random survey of 542 households in Nairobi’s urban areas and environs. The Tegemeo
Institute in collaboration with the Central Bureau of Statistics (CBS) using the CBS’s
NASSEP IV frame implemented the survey in November/December 2003. An earlier analysis
of household survey data conducted in 1995 involving a random survey of 549 in Nairobi was
used was a baseline.

The results indicate that while generally the volumes of staple carbohydrate consumption
have declined, it is the poorest section of the urban population that has been affected the
most. Also, on average, there has been a decline in consumption of maize products and rice.
The poorest have experienced the greatest decline. Consumption of wheat products has
grown significantly among all groups, but particularly among higher income groups. These
results also indicate a significant shift in maize meal consumption patterns. Even though
posho meal consumption has declined, it continues to be a preserve of the poor because it is
relatively inexpensive compared to sifted maize meal. Retail channels used to procure staple
carbohydrate products vary substantially by income. The poorest income groups rely mainly
on small shops (dukas) while well-off households rely both on dukas and supermarkets.

These findings hold some implications for food security policy. Because wheat is emerging as
an important expenditure item among the urban households, even the poor, the duty on
imported wheat and wheat flour, which currently stands at 35 percent and 60 percent,
respectively, may have adverse effects on urban poverty. Because Kenya is a member of
COMESA and WTO, there will be external pressures to reduce these duties over time.
Measures aimed at enhancing productivity and marketing of domestically produced wheat so
as to lower wheat prices for the benefit of the poor should be considered.

iv
STAPLE FOOD CONSUMPTION PATTERNS IN URBAN KENYA:
TRENDS AND POLICY IMPLICATIONS

INTRODUCTION
Since the early 20th century, maize has been the main staple crop of Kenya.
Historically, urban food security has depended on ensuring adequate supplies of maize at
tolerable prices. Maize consumption in Kenya has been estimated to be 98 kilograms per
person per year, and this figure has for many years served as the basis for the computation of
food balance sheets and other estimates of national cereal import requirements (Nyoro et al
2004).
This study examines current consumption patterns of the main staple carbohydrate
products in Nairobi -- maize, wheat, rice, and cooking banana -- in an effort to illuminate
policy issues affecting urban food security. We also identify the factors driving changes in
the amount and form of urban maize meal consumption. To better understand how food
security policy should be designed in order to respond to the needs of low-income consumers,
we disaggregate consumption and expenditure patterns for low-, medium- and high-income
groups. The study also examines whether the marketing channels used by the poor to secure
their staple carbohydrate products differ from those used by relatively high-income
consumers. The results hold important implications for policy makers by revealing how
staple food consumption patterns are changing, and by identifying the types of investments
that are needed to strengthen improve low-income consumers’ access to food.

DATA AND METHODS

Data
The data used in this study comes from a cross-sectional random survey of 600
households in Nairobi’s urban areas and environs. The survey was conducted in November-
December 2003 and implemented by the Tegemeo Institute in cooperation with the Central
Bureau of Statistics (CBS). The survey uses the CBS’s NASSEP IV frame established using
the 1999 nationwide population and housing census database. Census Enumeration Areas
(EAs) were used as the primary sampling units (PSUs). The first step in developing the

1
frame involved allocating the PSUs to the districts considered as the strata. This was followed
by selection of the PSUs using probability proportional to size.
Due to socio-economic diversity in the urban centers, the six major towns (Nairobi,
Mombasa, Kisumu, Nakuru, Eldoret and Thika) were stratified into five income classes
(strata): upper, lower-upper, middle, lower-middle and lower. Nairobi was allocated a total
of 108 primary sampling units out of the 1800 units in the national frame. These were then
allocated to the five strata using optimal allocation and the PSUs selected with probability
proportional to size. The allocation of PSUs among the five strata is as follows:

Income Primary Sampling units (PSUs)

Strata
1. Upper 28
2. Lower Upper 12
3. Middle 16
4. Lower Middle 36
5. Lower 16
Total 108

For the purpose of the household consumption survey, 30 PSUs were selected in Nairobi
using systematic random sampling with the following distribution:

Income Primary Sampling Units (PSUs)

Strata
1. Upper 8
2. Lower Upper 3
3. Middle 5
4. Lower Middle 10
5. Lower 4
Total 30

For each of the 30 primary sampling units, 20 households were then systematically selected,
giving a total of 600 households covered in the city. Because of missing information on
some surveys and other sources of attrition, the final sample size for analysis was reduced to
541 households.

Weighting Procedure
The weighting of the data from the household consumption survey takes into account
the sampling procedures at each stage of selection and non-responses. Weights for each
cluster were calculated based on their selection probabilities. Household weights were also

2
calculated based on their probabilities of selection. In cases where some selected households
did not respond, the weights were adjusted by the following factor:

wi = ki/ni

where ki = Number of selected households in the ith cluster

ni = Number of households that responded in the ith cluster

Thus the overall household weights were calculated as follows

Wci = Dij * Hsi / Hri

where Dij is the sample weight of the jth household in the ith cluster.
Hsi is the number of selected households in the ith cluster
Hri is the number of households that responded in the ith cluster

Surveyed households were asked about their purchases and consumption of an array of maize
products as well as wheat, rice, and other carbohydrate products that have traditionally
constituted the important sources of calories in urban diets. The specific maize products that
respondents were asked about include a) highly-refined sifted maize meal (e.g., “Hostess”
brand); the less-refined packaged maize meal brands (e.g., “Jogoo,” “Pembe,” “Jimbi,” etc);
the less-refined posho meal (both dehulled and straight run); grain for posho milling
(dehulled and straight run); grain for other dishes; and green maize. For wheat, respondents
were asked about their consumption of bread, flour, spaghetti, macaroni, and pasta products.
Consumption figures exclude food commodities consumed at the urban household premises
but produced at households’ rural farms and transported to town, as well as the relatively few
cases of food commodities grown and consumed from households’ urban plots.
As a baseline, we draw upon an earlier analysis of household survey data conducted
in 1995, also in Nairobi (Jayne and Argwings-Kodhek, 1997). This study involved a random
survey of 549 urban households implemented in October 1995. This was 20 months after the
adoption of important maize marketing policy changes, the most important of which was a
significantly smaller role of the National Cereals and Produce Board in the maize market and
the elimination of controls on maize trading and maize meal pricing.
The time series information on monthly retail maize grain prices is drawn from the
Market Information Bureau of the Ministry of Agriculture. Retail maize meal prices and the

3
urban CPI data are collected by the Central Bureau of Statistics. Estimates of monthly prices
for unrefined maize flour (“posho” meal) were obtained based on estimates of posho milling
charges at various intervals over the 1993-2003 period. Periodic visits to Nairobi markets by
Tegemeo Institute researchers indicates that milling charges have been roughly constant over
this period at between 4-6 Ksh per 2kg tin of maize grain. Thus, assuming a mean posho
milling charge of 2.5 Ksh per kg, this fee is added to the retail price of maize grain to derive
monthly posho meal prices. These prices do not account for the costs of consumers’ time and
transaction costs of acquiring the maize grain and standing in the posho mill queue.

Methods
Consumption patterns were converted to “adult equivalents” to standardize
consumption units within households. Adult equivalents are commonly used technique that
determines consumption on the basis of the sex and age of each specific household member.
By aggregating the determined adult equivalents of the respective household members, the
household’s number of adult equivalents was derived.

Conversion Factors to Compute Adult Equivalents

Adult Equivalence
Age Males Females
Under 1 year 0.33 0.33
1 - 1.99 0.46 0.46
2 - 2.99 0.54 0.54
3 - 4.99 0.62 0.62
5 - 6.99 0.74 0.70
7 - 9.99 0.84 0.72
10 - 11.99 0.88 0.78
12 - 13.99 0.96 0.84
14 - 15.99 1.06 0.86
16 - 17.99 1.14 0.86
18 - 29.99 1.04 0.80
30 - 59.99 1.00 0.82
60 and over 0.84 0.74
As per the World Health Organization (Jayne and Argwings-Kodhek 1997).

Household income was derived as the sum of proceeds from employment and
business earned by household members in the previous month. Remittances from household
members not residing in the household and pension accruing to retired household members

4
were also included. Households in the sample were ranked by income per adult equivalent
and then stratified into five income quintiles to assess potential differences in consumption
patterns by income.
Estimates of household monthly expenditure on particular food products were derived
as quantities purchased multiplied by the median price paid by all households purchasing the
food item. The descriptive tables and graphs presented later in the paper are based on these
expenditure computations. The surveys contained recall questions to explicitly record
consumption levels in different periods of the year to account for seasonality. However, it
was found that little seasonal variation existed in Nairobi consumers’ consumption of the
primary staple foods examined in this study. Hence, we report monthly consumption patterns
of the most recent period (mid November to mid December, 2003) covered in the survey.
Descriptive time series analysis of monthly price trends and marketing margins for
sifted vs. posho maize meal is carried out to explore whether potential shifts in consumption
patterns are correlated with changes in relative prices.

RESULTS
Staple Carbohydrate Consumption
Maize continues to be the primary staple food in Nairobi in terms of kilograms
consumed per adult equivalent. About 97 percent of the respondents reported to have
purchased maize products (maize meal, grain or green maize). About 12 percent reported to
have obtained maize grain or green maize either from relatives or their own farms in the rural
areas during the month preceding the survey. Two percent of respondents had acquired
maize products from their urban field.
Generally, total staple carbohydrate consumption has declined by 3 percent between
1995 and 2003 (Table 1). The extent of the decline decreases with income with the poorest 20
percent experiencing the largest decline (a 22% decline in gross kgs consumed between 1995
and 2003). The wealthiest 20% of urban households, by contrast, increased their
carbohydrate consumption by 19 percent.
Nairobi households consume a monthly average of 5.67 kgs per adult of maize
products, 3.83 kgs per adult of wheat products, 1.58 kgs per adult of rice, and 1.43 kgs per
adult of cooking bananas. However, there significant differences in consumption patterns
when the sample of Nairobi households are stratified by income group. After ranking all
households by monthly income per adult equivalent and then stratifying them into five equal

5
 groups, we find that households in the poorest income quintile consumes the most maize
(5.90 kgs per adult), declining by 7 percent to 5.47 kgs per adult for households in the
wealthiest income quintile. The inverse relationship between maize consumption and
household income suggests that maize is an “inferior good.”1
By contrast, the consumption of wheat products rise steadily as income rises. Wheat
products in this study include bread, flour, spaghetti, macaroni, and pasta products. Data in
Table 1 show that consumption of wheat products rises from 2.25 kgs per adult among
households in the lowest income group to 5.58 kgs among households in the highest income
group.

Table 1. Consumption of Primary Staples (kgs per adult equivalent) per month, Nairobi

2003 [a1] 1995 [b1] 2003[a2] 1995[b2] 2003[a3] 1995[b3] 2003[a4] 1995[b4] 2003 [c] 1995 [d]
1 5.90 8.95 2.25 2.11 1.16 1.45 0.75 0.34 10.06 12.85
2 5.61 8.39 3.08 2.66 1.55 1.56 1.19 0.12 11.43 12.73
3 5.74 7.90 3.47 3.73 1.38 1.82 1.88 0.23 12.47 13.68
4 4.32 6.97 4.76 3.56 1.79 1.86 1.81 0.63 12.68 13.02
5 5.47 5.78 5.58 4.06 2.01 2.13 1.52 0.28 14.58 12.25
Total 5.67 7.48 3.83 3.27 1.58 1.78 1.43 0.33 12.51 12.86
As a Percentage of Total Consumption of Primary Staples (2003 and 1995 respectively)
(a1/c)% (b1/d)% (a2/c)% (b2/d)% (a3/c)% (b3/d)% (a4/c)% (b4/d)% (c/c)% (d/d)%
1 58.65 69.65 22.37 16.42 11.53 11.28 7.46 2.65 100.00 100.00
2 49.08 65.91 26.95 20.90 13.56 12.25 10.41 0.94 100.00 100.00
3 46.03 57.75 27.83 27.27 11.07 13.30 15.08 1.68 100.00 100.00
4 34.07 53.53 37.54 27.34 14.12 14.29 14.27 4.84 100.00 100.00
5 37.52 47.18 38.27 33.14 13.79 17.39 10.43 2.29 100.00 100.00
Total 45.32 58.16 30.62 25.43 12.63 13.84 11.43 2.57 100.00 100.00
Percentage Change in Staple Consumption from Base year (1995)
(a1- b1)/ b1% (a2- b2)/ b2% (a3- b3)/ b3% (a4- b4)/ b4% (c- d)/ d%
1 -34.08 6.64 -20.00 120.59 -21.71
2 -33.13 15.79 -0.64 891.67 -10.21
3 -27.34 -6.97 -24.18 717.39 -8.85
4 -38.02 33.71 -3.76 187.30 -2.61
5 -5.36 37.44 -5.63 442.86 19.02
Total -24.20 17.13 -11.24 333.33 -2.72
Proportional Change in Consumption of Primary Staples (Base year 1995)
(a1- b1)/ d% (a2- b2)/ d% (a3- b3)/ d% (a4- b4)/ d% (c- d)/ d%
1 -23.74 1.09 -2.26 3.19 -21.71
2 -21.84 3.30 -0.08 8.41 -10.21
3 -15.79 -1.90 -3.22 12.06 -8.85
4 -20.35 9.22 -0.54 9.06 -2.61
5 -2.53 12.41 -0.98 10.12 19.02
Total -14.07 4.35 -1.56 8.55 -2.72

1
In economic jargon, inferior goods refer to commodities for which consumption declines as income levels rise.

6
 Consumption of rice also appears to be positively related to income. Consumption of
cooking bananas is not strongly correlated with income, except that it appears to be relatively
popular among households in the middle-income group. Consumption of cooking bananas
has also risen moderately in absolute kg terms and substantially in terms of percentage
change between 1995 and 2003. Rice and cooking bananas overall account for roughly 13
and 11 percent of the total kgs consumed of the staple food products included in this study.
The prominence of maize in urban staple grain diet has declined by 22 percent from
58 percent share in staple food 1995 to 45 percent in 2003. Maize accounts for 59 and 38
percent of the staple carbohydrate consumption among the 20 percent poorest and richest
households (respectively). It is still the dominant staple food among the urban poor.
However, among urban households in the top 40% of the income distribution, wheat has
overtaken maize in terms of amounts consumed (Figure 1). Consumption of wheat products
has grown significantly in the past decade, particularly among higher income groups. Its
share has risen by 17 percent from 25 percent in 1995 to 31 percent in 2003. Wheat
consumption accounts for 22 and 38 percent of the 20 percent poorest and richest
(respectively) staple food consumption. The share of rice declined while that of cooked
bananas tripled over the same period.

Figure 1: Per Adult Equivalent Consumption of Staples (Kgs)

!
%)
,% / 0 ( , %
.
-

"# $ %& "' ( ) %&

* + ( %(,

(1 2 % 3 ) %2 % 4( 5( 6

7
Staple Carbohydrate Expenditures
The survey results provide a perhaps surprising picture of expenditure patterns.
Although maize products provide the greatest contribution to Nairobi staple food diets in
terms of kilograms consumed (except for households in the highest income quintiles), wheat
is now emerging as the most important carbohydrate expenditure item. This is because wheat
is more expensive per kilogram than maize, and because the consumption of wheat products
has risen over time while maize consumption has declined since 1995 (Table 2). Wheat
products in 2003 accounted for 43.5% of total expenditures on the main staples (maize,
wheat, rice and cooking bananas). Maize products have declined from 41.8% to 32.4% of
staple carbohydrate expenditures between 1995 and 2003. Nairobi consumers’ expenditures
shares for rice were roughly constant between 1995 and 2003, while the expenditure share of
cooking bananas has tripled, albeit from a small initial share in 1995.

Table 2. Staple Carbohydrate Expenditures Shares (Expenditure as a Percentage of Total

Expenditures on the Top Four Staples) among Nairobi Consumers, 1995 vs. 2003.
1995 2003
Maize Products 41.8 32.4
Wheat products 34.8 43.5
Rice 22.1 20.2
Cooking bananas 1.3 3.9
Source: Tegemeo Urban Household Consumer Surveys, 1995 and 2003.

In 2003, expenditures on wheat products were Ksh 170 per adult per month (Table 3).
By contrast, maize and rice products accounted for Ksh 126 and Ksh 79 per adult per month
respectively. Cooking bananas were relatively low at Ksh 15 per adult per month (4 percent
of expenditures on the staple food products covered in this study).
When consumer expenditures are disaggregated according to household income, we
observe some notable differences. As with wheat consumption, expenditures on wheat
products increase greatly with income (Table 3 and Figure 2). Wheat expenditures were Ksh
98 per adult per month for households in the bottom income quintile, rising to Ksh 255 per
adult per month for households in the highest income quintile. Rice expenditures are also
positively related to income. By contrast, expenditures on maize products first rise then
decline moderately as income rises. Maize products account for less than half of total
expenditures on staple carbohydrate products for all income groups, ranging from 43.8% for
the poorest 20% of Nairobi consumers, to just under 22% for the wealthiest 20%. Wheat

8
products rise from 34% of total expenditures on staple products for the poorest group to over
50% of expenditures for the wealthiest group. Rice products consistently account for roughly
20% of staple food expenditures across the income spectrum.

Table 3. Expenditures on primary Staple Commodities (Ksh per adult equivalent per month),
Nairobi.

- ) 7 8% % , - ) 7 8% % , - ) 7 8% % , - ) 7 8% % ,
", $ %& 9 ! 9 ! 9 9 ! ! !!
! 9 !!
99 9 9 !
!9 9 9 9 9
") ( ) %& ! 9 ! ! ! 9
% , !" ! !# # $% &! &' %' '& " $ ' !# !' # # #
2 % % 7 $ :$ (

Figure 2: Expenditures on Primary Staple Commodities (Ksh per adult equivalent per month),
Nairobi.

300

250
-

200
,

150
+

100

50
*
()

0
1 (Lowest) 2 3 4 5 (Highest)

(1 2 % 3 ) %2 % 4( 5( 6

Expenditures on maize products exceed those on wheat for households in the bottom
two-income quintiles (40% of the poorest households in Nairobi), but for the top three
income quintiles (60% of Nairobi’s population), wheat dominates staple carbohydrate
expenditures. If these results are not simply a one-off phenomenon but reflect robust trends
in consumption patterns, then there is need for a general updating of the widespread notion
that urban Kenyan consumers are exclusively dependent on one crop – maize – for their
staple food requirements.

9
 There has been a general decline in staple carbohydrates real prices between 1995 and
2003 (Table 4). Even though maize products prices appear to have declined by a bigger
percentage (16%) compared to wheat and rice, the difference in price in the same period is
largest for rice and wheat products. It appears that the increased consumption of wheat
among the higher income groups could reflect, to some extent, substitution in consumption
on the basis of a decline in the wheat/maize price ratio.

Table 4. Staple Carbohydrates Real Prices (Ksh per kg), 1995 and 2003.
1995 2003 % Change
Maize Products 26 22 -16
Sifted maize meal 28 25 -11
Posho meal 24 22 -8
Maize grain 22 18 -18
Wheat products 50 44 -12
Wheat flour 45 34 -24
Wheat bread 52 46 -12
Rice 57 50 -12
Cooking bananas 16 11 -31
Source: Republic of Kenya, CBS, Statistical Abstract (various issues) and Author Calculations

Expenditure Shares in Total Income

In some countries, the dominance of a particular staple commodity in consumption
means that rapid increases in its price (due to drought or other supply disturbance) can have
far-reaching economic reverberations on real wages and even macroeconomic effects. Such
commodities are called “wage goods” because their prices tend to indirectly affect wage rates
throughout the economy. An example of this would be rice in Indonesia, which accounts for
over 75 percent of total cereal consumption and over half of total calorie intake (FAO, 2004).
Maize has been considered to be a wage good in Kenya, and if rural consumption patterns are
accounted for, the primacy of maize in the overall Kenyan diet would presumably still be
clear.
However, the importance of maize in urban consumption appears to be declining over
time. Overall, the expenditure share of all maize products in total income per adult
equivalent in Nairobi is 4 percent (Figure 3). Among the poorest 20 percent of urban
consumers, the expenditure share of all maize products in total income is 11 percent. In their
1995 survey, Jayne and Argwings-Kodhek (1997) found that the poorest 20 percent of urban
households in Nairobi spent 17 percent of their total income on maize meal. This means that
even among the relatively poor consumers of Nairobi, the importance of maize expenditures
as a fraction of overall income has declined somewhat. Among the highest income quintile,

10
maize expenditures constitute a negligible 0.7 percent of total income, which is exactly
consistent with the findings of the earlier 1995 study.

Figure 3. Expenditure Shares in Per Adult Equivalent Income

30

25
% ,*

20
(

15
)

10
(%
/ ;.

0
1 (Lowest) 2 3 4 5 (Highest)

Total Maize Products Wheat Products Rice

Source: Tegemeo Urban Household Consumer Survey, 2003.

We are not in a position to determine whether the general decline in maize

consumption reflects a long-term secular decline or is a short-term result of temporary shifts
in maize-wheat price ratios. However, based on secondary sources indicating a general rise
in wheat and rice consumption in the east Africa region, it is quite possible that urban staple
food consumption patterns are indeed becoming more diversified.

Disaggregating the Forms of Maize Consumption

There are three main types of products in which maize is consumed in urban Kenya.
There are two main types of maize meal products: relatively expensive packaged sifted meal
(with varying levels of refinement), and a less expensive and less refined “posho” maize meal
(also with varying levels of refinement). Packaged sifted meal is produced by larger
commercial mills and sold in retail outlets, while posho meal is typically obtained when
consumers buy maize grain and take it to a small local hammer mill for grinding into meal.
Green maize (roasted or boiled on the cob) is also popular. Whole or cracked maize grain is
also prepared in some dishes.
In 2003, 92 percent of the households had bought maize meal. Out of these, 80 and
12 percent had bought sifted and posho meals, respectively. The number of households

11
purchasing both sifted and whole flour stood at 8 percent, which is double the percentage in
the 1995 survey.
Sifted maize meal is the most common form of maize consumption in urban Kenya.
As shown in Table 5, over the entire sample, sifted maize flour accounts for Ksh 91 of the
Ksh 126 expenditure per adult per month on maize products (72 percent of expenditures), and
83 percent of the total quantity of maize meal consumed. Posho meal accounts for only Ksh
19 per adult per month, which amounts to 17 percent of consumption both in kgs and
expenditure of total maize meal expenditure. Green/whole maize accounts for Ksh 17 per
adult per month.
These results indicate a significant shift in maize meal consumption patterns since the
1995 survey by Jayne and Argwings-Kodhek (1997). In their analysis, posho meal
constituted 46 percent of total maize meal consumption in Nairobi. The finding that posho
meal consumption has declined from 46 to 17 percent of maize meal consumption is
consistent with rapid appraisals of posho millers and other participants in Kenya’s maize
marketing system conducted in late September 2004.
The form of maize consumption varies substantially by income. Among households
in the poorest income group, posho meal accounts for 2.03 kgs of the total 4.80 kgs
consumed per adult per month (42 percent). This indicates that posho meal, because it is
relatively inexpensive compared to sifted flour, plays an important role in poor people’s diets
in urban areas. Households headed by less professional individuals are more likely to
consume posho meal (Figure 4).

Figure 4: Household Heads Economic Activities and Maize Meal Consumption

100%

80%

60%

40%

20%

0%
/ ( < % % ( % ( (, ) . %% % ' $5 3 %)
(

(8% 2 )

Source: Tegemeo Urban Household Consumer Survey, 2003.

12
 To understand the importance of the informal markets and small-scale posho millers
in reducing the costs of acquiring maize meal products for the poor, we ran some simple
simulations to estimate the additional costs that consumers would pay if they had to
exclusively purchase sifted meal. This would simulate the conditions that would exist if
maize grain and posho millers were not available for consumers to purchase maize and mill it
into posho meal. To estimate this, we multiply the kgs consumed of posho meal by the
survey median price difference between sifted and posho maize meal for the various income
groups shown in Table 5. We find that if household consuming posho or both posho and
sifted meal were to shift their maize consumption to sifted maize meal (holding other factors
constant), on average they would be forced to pay an additional Ksh 5 per adult equivalent
per month (Figure 5).

Table 5. Consumption and Expenditures of Main Carbohydrate Products (kgs per adult
equivalent per month), by Household Income Quintile

Median monthly income Mean consumption Mean expenditure

Commodity Ksh per adult equiv. Kgs per adult equiv. Ksh per adult equiv.
per month per month
1 1,127 2.77 70.51
Maize Sifted 2 2,112 4.00 100.01
3 3,548 4.03 101.80
4 5,563 3.64 92.17
5 15,337 3.47 89.21
Total 3,548 3.58 90.74
1 1,127 2.03 39.90
Posho 2 2,112 1.06 21.34
3 3,548 0.87 16.78
4 5,563 0.72 14.10
5 15,337 0.14 3.02
Total 3,548 0.96 19.04
1 1,127 4.80 110.42
Maize Meal 2 2,112 5.06 121.35
(Sifted+ 3 3,548 4.89 118.58
Posho) 4 5,563 4.36 106.27
5 15,337 3.61 92.23
Total 3,548 4.54 109.78
Total Maize 1 1,127 5.90 128.21
Products 2 2,112 5.61 136.30
(Meal+ 3 3,548 5.74 131.29
Grain+ 4 5,563 4.32 130.78
Green) 5 15,337 5.47 104.79
Total 3,548 5.67 126.30
Source: 2004 Tegemeo/MSU urban survey.

13
 Households in the lower income quintiles would be the most affected, with the
poorest 20 percent of Nairobi households paying Ksh 11 per adult equivalent per month,
while the wealthiest 20 percent would pay 43 cents per adult equivalent. However, even
among poor consumers, this is a very small amount, accounting for roughly 1% of household
income among the poor. The importance of posho meal in maize consumption and its role in
providing a cheaper alternative to sifted meal appears to have declined since the 1995 survey.
This is because of two changes between 1995 and 2003: (1) the price difference between
sifted meal and posho meal has declined, perhaps resulting from greater competition in the
milling sector (there has been rapid expansion in maize milling investment in the past 10-12
years); and (2) a reduction in the amount of posho meal consumed by the poor in 2003.

Figure 5. Extra Costs If Households Consuming Posho Meal were to Shift to Sifted Meal

12

10
Extra cost (KSh)

0
1 (Lowest) 2 3 4 5 (Highest)
* + ( %(,

/ ;% %2 < ,% / 0 ( , % %)

Source: 2003 Tegemeo Urban survey results.

Over the same period (1995 to 2003), maize meal consumption per adult equivalent
dropped from 6.9 to 5.8kgs per month. The expenditure and consumption per adult equivalent
varies inversely with household income per adult equivalent. The highly refined meal
(Hostess brand) seems to be a preserve of wealthier households. Roughly 57 percent of
households consuming this brand were within the wealthiest 20 percent income group.

14
 Box 1: A Typical Posho Consuming Household in Nairobi

Esther Waithera is divorced. She is 31 years old and lives with her five children in Maili Saba slums. Three of
her children are in primary school while the other two are yet to attain school-going age. Even though Esther is
a form four leaver, she has not managed to secure a well paying job, thus turning to hawking as a source of
livelihood for her family. She is making Ksh. 2,400 on average monthly. Her family live in a single room iron-
sheet roofed and walled rented house with earth floor. Her family neither has a television set, a transistor radio
nor a mobile phone.

During the month prior to the interview, Waithera had spent Ksh 1,000 on family food; Ksh 200 on food outside
the household; Ksh 300 on house rent; and Ksh 100 on transport (fare). On water and fuel, she had spent Ksh
240 and Ksh 256 respectively. Over the year preceding the interview, she had spent Ksh 200 on medical bills.
Since her children are not yet in the secondary school, which is relatively expensive, and given that primary
education is now for free, she had only spent Ksh 500 on school fees.

Maize is the main Waithera’s family source of carbohydrates. Over the last one month she had purchased ten
gorogoros (2kg cooking fat tin containing approximately 2.25kg of maize grain) of maize grain from the posho
mill (hammer miller) at KSh 35 per gorogoro, which was subsequently milled into posho. Waithera prefers
posho meal because it is because it is cheap compared to sifted maize meal. She had also bought two gorogoros
of maize grain for githeri (a mixture of maize and beans meal) and 20 pieces of green maize at Ksh 4 per piece
from the local market. Waithera’s household had also bought 4 kilos of rice at Ksh 30 per kg and 4 loaves of
bread at Ksh 20 from the local duka (shop).

Waithera concurs that posho meal is cheaper compared to sifted meal but quickly adds that it is not convenient
to obtain considering that one has to procure the grain and take to the posho mill. It takes her approximately 15
minutes to buy maize grain, about 5 minutes waiting time while in the queue at the posho miller and about 15
minutes to walk home with posho meal. Besides, during the months of April, May and June maize grain is
scarcely available in the market forcing prices to shoot.

She occasionally receives farm produce from her relatives living in the countryside. For example, over the year
preceding this survey, Waithera household had received 2 gorogoros of irish potatoes; 10 cabbages; half
gorogoro of green peas and 3 litres of milk.

Reasons for the Recent Shift toward Sifted Maize Meal

These results from the foregoing analysis indicate a significant shift in maize meal
consumption patterns since the 1995 survey. Respondents were asked the reasons behind
their current maize meal consumption patterns. Thirty-seven (37) percent of the households
compared to 58 percent in 1995 consumed sifted maize meal because it is convenient to
procure and cook (Table 6). While sifted maize meal can be obtained from retail outlets
easily, with posho meal (69% of the posho consuming households) one has to procure the
grain first, then take it to hammer miller for milling. Most of the posho-consuming
households (86%) could at least access a hammer miller within 15 minutes of walk. From the
households consuming sifted meal solely, about 35 percent could not access posho mills
within 15 minutes walk.

15
Table 6. Primary Reasons for Consuming Alternative Forms of Maize Meal, 2003
) .
*
* = . = .
> > % ," & % ," & > > % ," & % ," &
4 , %( , ( ; . ( ! !
( %% . 5 ! 9
% (%(
) ( ( ! 9 9
% %% 9
? (,, % ) %% 9 !
' (% !
@ %)
Source: 1995 and 2003 Tegemeo Institute/Michigan State University Urban Household Surveys.
Note: percentages in the columns add up vertically to 100

Taste (18%) and sheer habit (16%) are also other important considerations behind the
consumption of sifted meal. Compared to 1995, 12 percent of the households are finding
sifted meal less expensive. Primary reasons behind the consumption of sifted meal do not
vary with income.
Less households, 56 percent, compared to 67 percent in 1995 are considering posho
meal reasonably priced compared to sifted meal. Preference of posho meal on basis of price
varies inversely with income with 61 and 44 percent of the poorest and wealthiest 40 percent
consumption of posho meal based on affordability.
The profound ability of posho meal to “fill the stomach” for a longer time was the
primary reason behind 27 percent of the households consuming posho meal. This attribute
(filling stomach better) appears to vary directly with income, with more households (46%) in
the 40 percent wealthiest income group preferring posho meal on that basis.
Considerable cost-competition by large millers has lowered sifted meal prices
motivating preferences for sifted meal over posho meal. Prior to market deregulation, the
large millers were oligopolistic in nature due to due to regulatory barriers. Since then, many
new entrants have come aboard and competition stiffened. Interviews with millers revealed
that to compete effectively, large millers had to reduce the grain-to-meal extraction rate. Also
bulk grain storage enables millers to smooth their annual grain supply to guard against meal
price fluctuations. They have also adopted other strategies like bulk packaging (90-kg) of

16
sifted meal which can be further broken on demand to small quantities as small as a quarter
of a kg by the retailers.
At the time of the 2003 survey (November/December), the price of sifted in real terms
was relatively low, and the price difference between sifted and posho was lower than in most
years between 1993 and 2004 (Figure 6). When the price difference is narrow between sifted
and posho meal, a shift in consumption toward sifted meal is to be expected, and we might
also expect a shift back from sifted to posho in years when the price discounts for posho
become relatively large again, if consumers (especially the poor ones) are conscious to price
variations.

Figure 6. Inflation-adjusted Prices of Retail Maize Grain, Posho Meal, and Sifted Meal, Nairobi
Markets, 1993-2004.

60

)
s
e 50
ci
pr
4
0
0 40
2
t
n
at
s
n 30
o
c(
g
k
r 20
e
p
h
s
K
10
1994 1996 1998 2000 2002 2004

maize grain, retail markets

unrefined "posho" meal, retail markets
packaged sifted meal, retail

Source: CBS for CPI and sifted maize meal prices; Ministry of Agriculture for retail maize grain prices;
authors’ derivation for posho meal as described in data section.

Posho millers are also operating on an uneven playing ground with the large millers.
While the large millers can procure grain cheaply and in bulk (thus enjoying economies of
scale) either directly from farmers or the NCPB, the posho millers are relegated to grain

17
procured through the informal market channels that is relatively expensive. Furthermore, the
duty free maize being imported currently is likely to disproportionately benefit the large
millers.
Interviews with the posho millers indicated that the initial cost of starting the
business, the increasing cost of diesel and electricity, and over-investment as some of the
major reasons bedeviling the posho milling industry. Fuel and electricity costs account for
about 50 percent of the posho mill operating costs. Due to a lack of alternative standby
power arrangements, the posho mill industry has also been bedeviled by frequent power
failures. Cost of licenses and rent are the next most important costs to posho millers.
Most of the posho millers are operating at 10 percent of their installed capacity, doing
between one and two 90-kg bag in a day on average. This means the milled volume in a day
is lower than what can be done in one hour. Inter-posho miller competition is based on
milling charges and ability to source relatively cheap maize grain. Posho mill owners are
also employing persons to aggressively for their business. If one is able to stock maize grain,
then he/she stand a chance of milling more compared to the one who relies on customers
coming with their own maize grain. Thus, for competitiveness, posho millers struggle to get
and stock cheap maize grain. Locating a mill near a grain market is another business
strategy.

Consumers’ Use of Alternative Marketing Channels for Primary Staple Products

Nairobi consumers use a range of alternative marketing channels for procuring their
main staple food products. These include duka shops (small retail stores), large national
supermarket chain stores, smaller supermarkets, open markets, kiosks (stalls), and posho
millers. Table 7 presents the importance of various retail channels by which urban
consumers procure their primary staple products. The table also shows the respective values
of primary staples (maize products, rice, wheat products and cooking bananas) purchased
from each retail channel. Staples worth Ksh 473 million are sold through duka/shop in a
month representing 50 percent of the market share. Large supermarkets follow handling 17
percent of the market share valued at Ksh 160millions in a month. Other important channels
include small supermarkets (12%), markets (10%) and kiosks (9%).
The importance of these retail channels varies substantially by income. The majority
of the households (64%) in the poorest income quintile prefer duka/shops, spending Ksh105
million in a month on this channel (Table 7). Open markets are the next most important
means by which the poorest consumers obtain staple products, accounting for 14% of their

18
expenditures. Among the 2nd poorest income group, the most important retail outlets are
dukas/shops, roadside kiosks, small supermarkets, and markets. Only among the top income
quintile did national supermarkets account for more than 20% of total expenditures on the
main staple food products. Among this relatively wealthy group, duka/shops accounted for
38% of expenditures on the top four staples, while the national supermarkets accounted for
37%.

Table 7. Consumers’ Use (Ksh Millions) and Market Share (%) of Alternative Retail Channels
for Primary Staple Products2, by Income Group, 2003

/0 1 2 ! & $ /3 2
4 % (, ) , - ) 7 - ) 7 - ) 7 - ) 7 - ) 7 - ) 7
A 5 ) . 9 ! ! 99 9 !
# . 5 % ! ! ! 9 9 ! !
,, . 5 % 9 9 !
5 % !
- ( 5 5( 9 ! 9 9 99
2 ) (,,
@ %) 9
$ "" $ "" "" "" ! "" #$" ""
Source: 1995 and 2003 Tegemeo Institute/Michigan State University Urban Household Surveys.

Considering the different retail channels specifically for sifted maize meal, we find
that about 60 percent of the 20 percent poorest and the wealthiest obtain their supplies from
duka/kiosk and large supermarkets, respectively, even though there doesn’t seem to be a
marked difference in the retail price (Figure 7). Small supermarkets are offering a slightly
lower price but that not withstanding; only about 18 percent are using them on average.
Majority of the households (70 percent) consuming the extremely refined brand (Hostess)
procure it from large supermarkets while 15 percent buy from small supermarkets.

2
Maize and wheat products, rice and cooking bananas
3
Hawker, manufacturer, factory, green grocer, neighbor, bakery, place of work and colleagues

19
Figure 7. Alternative Retail Channels for Sifted Maize Meal, 2003

80
. "7& 70
60
=

50
( %) *

40
30
20
10
% ,B

0
"# $ %& "' ( ) %& 2 (

Shop/Kiosk Small Supermarket Large Supermarket

Source: 2003 Tegemeo Institute/Michigan State University Urban Household Surveys.

CONCLUSIONS AND POLICY IMPLICATIONS

This study has identified the factors driving changes in the amount and form of urban
staple carbohydrates consumed. Wheat products, maize products, rice, and cooking bananas
were the main staple carbohydrate products consumed by Nairobi consumers. The results
indicate that the volumes of staple carbohydrate consumption have remained roughly
constant between 1995 and 2003. However, among the relatively wealthy group,
consumption has increased while among the poorest groups, consumption has declined. On
average, there has been a shift in consumer expenditures from maize to wheat products.
Wheat now appears to be the dominant staple in terms of expenditure, accounting for 43.5
percent of total expenditures of the main staple carbohydrate products, while maize products
account for 32.4 percent in 2003. Consumption of wheat products has particularly increased
among higher income groups. However, among the poorest 20 percent of Nairobi consumers,
maize meal remains the dominant staple, although even among this group, wheat
consumption has risen. The expenditure shares of rice and cooking bananas are roughly 14
percent and 4 percent, respectively.
In some countries, the dominance of a particular staple commodity in consumption
means that rapid increases in its price (due to drought or other supply disturbance) can have
far-reaching economic reverberations on real wages and even macroeconomic effects. Such
commodities are called “wage goods” because their prices tend to indirectly affect wage rates
throughout the economy. An example of this would be rice in Indonesia, which accounts for

20
over 75 percent of total cereal consumption and over half of total calorie intake (FAO, 2004).
Maize has been considered to be a wage good in Kenya, and if rural consumption patterns are
accounted for, the primacy of maize in the overall Kenyan diet would presumably still be
clear. However, in urban areas, staple food expenditures appear to have become considerably
diversified. In some respects, this is a positive trend as it reduces the strategic political
importance of a single commodity and the likelihood of civil disruption when the price of that
commodity rises.
We are not in a position to determine whether the general decline in maize
consumption reflects a long-term secular decline or is a short-term result of temporary shifts
in maize-wheat price ratios. However, based on secondary sources indicating a general rise
in wheat and rice consumption in the east Africa region, it is quite possible that urban staple
food consumption patterns are indeed becoming more diversified.
These results also indicate a significant shift in maize meal consumption patterns.
Even though posho meal consumption has declined from 46 to 17 percent of maize meal
consumption on average, it continues to be an important commodity among the poor. The
driving force behind the consumption of posho appears to have little to do with its nutritional
supremacy. Primarily, low-income earners consume posho meal because it is relatively
inexpensive compared to sifted maize meal. However, because the price of posho meal (price
of grain plus milling fee) relative to the price of sifted meal has narrowed between 1995 and
2003, this may account for why consumption of posho has declined somewhat over this
period. Part of the reason for the narrowing of the relative costs of posho meal vs. sifted meal
has been because the large millers have reduced the degree of refinement of their products in
response to great competition from the posho millers since the market liberalization process
of the mid-1990s. The development of informal maize markets in urban areas induced rapid
investment of small-scale mills in urban areas, taking away market share from the large
millers. The large millers’ attempts to cut costs by reducing the level of refinement of sifted
maize meal products has put downward pressure on maize meal prices, to the benefit of poor
consumers. This response has narrowed the price difference between sifted and posho meal,
and it appears that this is a major factor accounting for the loss of posho millers’ market share
in recent years after having enjoyed strong demand for their products in the mid- to late-
1990s, especially among poor consumers.
These findings hold some key policy implication imperative to assisting low-income
consumers access to staple carbohydrates. Wheat is emerging as an important expenditure
item among the urban households. This is a clear signal to the policy makers that wheat

21
should not be left out when formulating food security policies. The country domestic
production is 67,000 metric tones and imports 570,000 metric tones annually (republic of
Kenya 2004). The domestically produced wheat costs US $224 per ton. Duty on imported
wheat and wheat flour stands at 35 percent and 60 percent respectively. The imposition of
duties on wheat imports is forcing consumers to take expensive wheat products. However,
import duty is bound to diminish in the long run because Kenya is a member of the Common
Market for Eastern and Southern Africa (COMEAS) and the World Trade Organization
(WTO) and has trade agreements with the European Union under the African Caribbean
Pacific (ACP)-European Union (EU) Cotonou Agreement. These agreements require
lowering of tariffs.
This should also send a signal to wheat farmers that potential market exist for
increased production and enhanced productivity. Thus, measures to aid farmers improve on
productivity to lower wheat prices include encouraging wheat farmers to grow higher
yielding varieties and application of improved agronomic practices; improved infrastructure
development (roads, railway and storage); farm inputs supply (fertilizer and agro-chemicals);
fuel; and reducing local authorities fees and direct and indirect taxes (estimated at between
US$23-35 per ton) farmers pay in the course of meeting their production costs.
The findings indicate that retail channels use for staple carbohydrates vary
substantially by income. Shops (dukas) are popular with the poorest income group. Mostly,
large supermarkets are the preserve of well-off urban households, and even for them, dukas
were the primary source for their staple carbohydrate products. It is possible that this is
changing rapidly. Regardless, policy measures aimed at reducing the costs incurred by dukas
in their operations and assisting these shops to continue serving the poor sections of the urban
population efficiently should be explored, given their primary importance for urban
consumers, and especially the poor.

22
References

Argwings-Kodhek, G. and Thom S. Jayne, 1996, Maize Market Liberalization and Food
Consumption Patterns in Urban Kenya. Tegemeo Institute, Working Paper No. 2

Donovan, G. 1996. Agriculture and Economic Reform in Sub-Saharan Africa, AFTES

Working Paper No. 18, Washington D.C.: The World Bank, Africa Technical
Department.

Jayne, T.S. and G. Argwings-Kodhek, 1997, Consumer Response to Maize Market

Liberalization in Urban Kenya." Food Policy, v. 22, no. 5 (Oct.): 447-458.

Jayne, T.S. and Stephen Jones, 1997, Food Marketing and Pricing Policy in Eastern and
Southern Africa: A Survey, World Development, 25(9).

Mukumbu, Mulinge and T.S. Jayne, 1994, Urban Maize Meal Consumption Patterns:
Strategies for Improving Food Access for Vulnerable Households in Kenya," in
Proceedings of the Symposium on Agricultural Policies and Food Security in East
Africa, East African Association of Agricultural Economists, May 1994

Nyoro, J.K., Lilian Kirimi and T.S. Jayne, 2004, Competitiveness of Kenyan and Ugandan
Maize Production: Challenges for the Future.

Republic of Kenya, 2004, Economic Survey, Central Bureau of Statistics, Government

Printer, Nairobi

World Bank, 1994, Adjustment in Africa: Reforms, Results and the Road Ahead (N.Y.;
Oxford Univ. Press)

23