The Household Budget Surveys collected extensive information on household
consumption and expenditure. This includes information on which items were
consumed, on their source and their cost.This chapter examines the overall level
of household consumption and how it has changed over the 1990s. It also
examines the structure of household consumption, presenting the share of
consumption that is taken by the various categories of goods and services.
6.2 Measuring Consumption and Expenditure
The 2000/01 HBS collected information on household consumption and
expenditure in three ways. A diary was kept in each household to record items
purchased and consumed each day for one calendar month.This was filled in by
a household member or, where household members were illiterate, by the
interviewer. The interviewer was expected to visit households every two to
three days to ensure that the diary was being correctly completed; visits were
expected every day in the case of illiterate households. In addition to this diary,
individual adults were provided with a personal diary that they used to record
their personal expenditure outside the household. This information was later
added to the household diary. In a separate part of the questionnaire,
respondents were also asked to recall expenditure on non-food items in the last
year.The same procedures were used in the 1991/92 HBS except that personal
diaries were not provided.
The consumption recorded in the diary included both items that were purchased
and items that were consumed but which had not been purchased: home
produced items, transfers received by the household (such as gifts or support
from other households) and payments in kind for work done. Both the quantity
consumed and its value were recorded. Items that had not been purchased were
valued at local market prices.
Since all items that were consumed were priced, a single monetary measure of
household consumption can be calculated, denoted household consumption
expenditure. This includes food consumption, health and educational expenses,
expenditure on consumer durables and expenditure on other non-durables. It is
standardised to 28 days.
68 HOUSEHOLDBUDGETSURVEY Final Report
The surveys also collected information on household incomes. However,
household income in developing countries fluctuates more than expenditure
over the short term and is often reported less accurately. Consumption
expenditure provides a more reliable measure of household income, reducing
reporting errors and smoothing out short-term fluctuations. For this reason,
consumption expenditure is used as the basis for a monetary measure of welfare.
The collection of consumption data in a monthly diary generates a large volume
of data, unavoidably producing some errors, which necessitated a careful data
cleaning procedure. This cleaning was carried out for both the 1991/92 and
2000/01 data to ensure comparability. For some items, it was possible to derive
information on consumption either from the diary or from the twelve-month
recall section, the choice being based on an analysis of relative frequencies and
values in the two sources. These and other technical issues are discussed in
Chapter 7 compares household consumption levels to a poverty line. There, a
more restricted consumption measure is used to provide a more precise
measure of welfare. However this chapter focuses on describing consumption
patterns more broadly.
6.3 Average Consumption Expenditure Levels
Table 6.1 shows the average levels of consumption expenditure for 28 days by
area. Mean expenditure per capita is 10,120 Shillings.The median – which shows
the value below which half of the population falls – was substantially lower,
because a small number of high values have more effect on the mean.
Mean per capita expenditure in Dar es Salaam is 2.6 times higher than in rural
areas. Differences in average consumption expenditure per household were
slightly smaller because rural households tend to be larger.
Comparing the regions, average expenditure is highest in Dar es Salaam, Mbeya
and Mtwara and lowest in Rukwa, Singida and Kigoma (Map 6.1).
TABLE 6.1 AVERAGE CONSUMPTION EXPENDITURE LEVELS IN 2000/01 (28 DAYS,TSHS)
Measure Dar es Salaam Other urban Rural areas Mainland
Mean expenditure per capita 21,949 14,377 8,538 10,120
Median expenditure per capita 16,349 11,561 6,860 7,523
Mean expenditure per household 117,893 78,079 52,649 59,935
Note: Consumption expenditure in nominal prices.
69 6. Household Consumption and Expenditure
An adjustment for price inflation is required to compare expenditure in 2000/01
with 1991/92.This is calculated using price information contained within the HBS
data itself, because both the quantity consumed and its value were recorded.
From this data, a Fisher Ideal price index is calculated. This suggests that an
average consumption basket has increased in price by a factor of 2.49 since
1991/92, well below the increase in the consumer price index.The Fisher Index
has the benefit of taking into account changes in consumption patterns that have
occurred over the period15.
Table 6.2 presents mean and median per capita consumption expenditure for
2000/01, together with their equivalents for 1991/92 inflated by the Fisher Index.
The inflated figures give the values that would be expected if average
consumption levels had remained constant over the period and only prices had
This comparison shows that average real consumption levels have increased over
the 1990s. Mean consumption levels have increased by around 17 per cent in real
terms. Median levels have increased by around 10 per cent. The increases have
been largest in Dar and smallest in rural areas. The increase in real rural
consumption levels is lower still if prices are adjusted separately. Nevertheless, in
all areas there has been an increase in household average real consumption over
TABLE 6.2 TRENDS IN REAL PER CAPITA CONSUMPTION EXPENDITURE
(28 DAYS, 2000/01 TSHS)
Dar es Salaam Other urban Rural areas Mainland
2000/01 21,949 14,377 8,538 10,120
1991/92 14,896 12,733 7,661 8,686
Ratio (00/01) / (91/92) 1.47 1.13 1.11 1.17
Ratio with area-specific price adjustment 1.48 1.18 1.09 1.16
2000/01 16,349 11,561 6,860 7,523
1991/92 12,106 9,622 6,300 6,816
Ratio (00/01) / (91/92) 1.35 1.20 1.09 1.10
Ratio with area-specific price adjustment 1.36 1.25 1.07 1.10
Note: 1991/92 values are inflated to 2000/01 prices using the national Fisher Index calculated from the HBS data (2.49).
The second ratio for each measure gives the increase in real consumption when prices are adjusted separately for each of
the three areas.
15 Further details on the calculation of the Fisher Index are given in Appendix A2.
70 HOUSEHOLDBUDGETSURVEY Final Report
6.4 The Structure of Consumption
In addition to providing information on the overall level of household
consumption, the HBS can be used to look at the structure of consumption.Table
6.3 shows the mean expenditure per capita on food, household durables, medical
and education costs and on other non-durables16. In this table, 1991/92 data are
left in 1991/92 prices.
TABLE 6.3 MEAN EXPENDITURE PER CAPITA BY CATEGORY OF ITEM (NOMINAL
FIGURES, 28 DAYS,TSHS)
Category 1991/92 2000/01
Dar es Other Rural Mainland Dar es Other Rural Mainland
Salaam Urban Tanzania Salaam Urban Tanzania
Food - purchased 3,873 2,862 1,011 1,397 10,301 7,114 3,118 4,085
Food – not purchased 36 361 1,175 1,011 368 876 2,375 2,051
Total food 3,910 3,223 2,186 2,409 10,668 7,989 5,492 6,137
Durables 577 541 262 314 2,565 1,501 706 923
Medical expenditure 52 65 26 32 569 338 190 232
Education expenditure 66 47 19 25 974 431 138 227
Other non-durables 1,377 1,237 583 708 7,172 4,118 2,012 2,602
Total 5,982 5,114 3,077 3,489 21,949 14,377 8,538 10,120
Note: ‘Food - not purchased’ includes food produced for home consumption, received as payment in kind or gifts etc. Figures
are not adjusted for inflation.
In both years, mean expenditure in all categories is highest in Dar es Salaam and
lowest in rural areas, with the exception of home-produced food.
There has been a fall in the share of household expenditure being spent on food
(Table 6.4) Within this, the proportion of food that is purchased has increased
while the proportion that is home produced has declined, suggesting a decline in
the importance of subsistence farming. Other items have seen substantial
increases in their share of household expenditure – particularly medical and
educational costs and other non-durables17.
The proportion of household expenditure that is devoted to food is usually
related to a household’s income. Households with higher incomes spend a lower
proportion on food. Households in Dar es Salaam spend the lowest proportion
on food (54 per cent), while rural households have the highest food share (67
per cent).The share of expenditure on food has declined in all areas.The decline
is largest in Dar es Salaam and smallest in rural areas, consistent with the growth
in real consumption levels.
16. Other non-durables include personal effects, personal care, recreation, fuel, transport, utilities and
services, clothes, alcohol and other items. More details of expenditure by item are given in Appendix
17. The decline in the share of expenditure on food will in part reflect increases in expenditure on some
items that households may have been unable to avoid – for example, in education and medical costs.
Food share declines even if these elements are removed, though it declines less than in Table 6.4.
71 6. Household Consumption and Expenditure
TABLE 6.4 MEAN PERCENTAGE SHARE OF CONSUMPTION EXPENDITURE BY
CATEGORY OF EXPENDITURE
Category Dar es Salaam Other urban Rural areas Mainland
91/92 00/01 91/92 00/01 91/92 00/01 91/92 00/01
Food - purchased 67.1 52.2 56.9 52.8 30.5 35.2 35.8 38.6
Food – not purchased 0.7 2.1 9.4 7.9 41.8 31.8 35.5 26.8
Durables 7.6 7.8 7.4 8.0 7.2 7.1 7.2 7.3
Medical expenditure 0.9 2.9 1.2 2.4 0.9 2.1 0.9 2.2
Education expenditure 1.1 4.0 1.1 3.0 0.8 1.6 0.8 2.0
Other non-durables 22.6 31.1 24.0 25.9 18.9 22.1 19.7 23.1
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Of which, total food 67.8 54.2 66.3 60.7 72.3 67.0 71.3 65.4
Note:These budget shares are means of the shares calculated for each household.
Average consumption expenditure per capita is around 2.6 times higher in Dar
es Salaam than in rural areas. Mean real expenditure levels have increased by
around 17 per cent between 1991/92 and 2000/01, suggesting that household
real incomes have been rising. The proportion of expenditure devoted to food
has also declined, as would be expected if real incomes have risen.There has also
been a decline in the proportion of food that is home produced, as dependence
on subsistence has declined. The share of expenditure devoted to health and
education has increased.
All areas have seen a real increase in household consumption and a fall in the
food share. However, there has also been a substantial widening of the gap
between Dar es Salaam, other urban areas and rural areas. Consumption levels
have risen most in Dar es Salaam and least in rural areas. The share of
expenditure on food shows a correspondingly larger fall in Dar es Salaam and a
smaller fall in rural areas.