Assessment of energy intake Estimates of food supply v measurement of food consumption
Elizabeth A. Dowler and Young Ok Seo
Nabonal consumption Indicators are frequently compiled using food supply estimates in the absence of reliable household or individual Intake data The authors examine the relationship between these three levels of mformation and in particular, the potential ‘losses’ of energy in the food system, comparing data from different countries’ and over time They demonstrate the unreliability of supply estimates as proxy indicators of consumption and quesbon their current usage in statements about global hunger and the links between health and food intake
The assessment of tood energv supply. whether at the mdlvldual. the h ousehold or natlonal level. 1s a major preoccupdtlon of mstltutes and agencies concerned with food and agriculture The purpose m mterpretIn_ethe data ~arles from predIctIon ot future movements m foodstuffs to
Keywords UK National Food Food supply, Food wastage
Larlous countries
Survey
E A Dowler IS with the Nutntlon Policy Unit, Department of Human Nutrition, London School of Hygiene and Tropical Medlclne, Keppel Street, London WClE 7HT, UK Dr Young Ok Seo was formally with the Nutrition Policy Unit and IS now working with the Korean Rural Nutrition Institute, 88-2 Suh Doon Dong Suwon. Kyung KI Province. Seoul, Korea Both authors are Indebted to PhIlIp Payne (Head of the Nutrition Department) for his continual support and advice throughout this work and to Professor T T Poleman of Cornell University for his comments on this paper Views expressed here are entirely thelr own ‘World Bank, World Tables. second edltlon. Johns Hopkins University Press, BaltConfrnued on page 279
278
attempts to assess nutrient Intake adequacv The reasons for the latter exercise depend to some extent on who IS makme the measurements national
lnstltutlons
agricultural
mdv
he concerned
and health sectors
consumption
with
pohcles
m the
food.
and wish to develop Indicators based on
data both to monitor Ttatus In whole populations
and serve
as warnings tor Interkentlon InternatIonal agencies. on the other hand. as well ds provldlng technical back up for such local actl\ltle3 often become
Involved
In publlshmg lntormatlon
series of social and economic
data on
on food supplq’ IS almost always included
otten under the heading of heAh or social Indicators Calorie suppI> per cdplta 15 that most otten quoted These figures usually derive from the FAO Prorllrcttor~ Yeurbooh When the data are presented as a percentage of requirements ’ the requirement figures are usually those published bv the FAO or WHO Such series ot data ‘Ire used both to mahe cross-country comparisons ‘md to follow trend\ over time HoHe\er this exercise In mahlng comparisons IS sometlmes further
than complllng
lists There
are instances where
tahen
the per capita
the lnequahtles ot calorie supplv figures dre used to emphasize consumption hetween countries - ,Ind even to support accusations ot Indulgent behablour (such as o\erconsumptlon and wastage) on the part ot the populations of richer countries ’ National average suppI> figures are also used In attempt? to estimate numbers malnourished m a particular region or the world as a whole, that IS. those whose consumption falls short ot requirements ’ Furthermore. the size of the gap - b! how much In aggregate. lntahes are below requirement - IS used as a measure ot the Increases In agricultural production needed to provide the necessary tood Several ,luthors habe questioned
0306-9192/85/030278-l
the theoretlcal
1$3 00 Q 1985 Butterworth
basis of the various
& Co (Publishers)
Ltd
-tssr,wwt1r OJ mrrg\ calculations
described
determInIng
energy
concept
filed
of
constitutes rlateness
of
seem Conbnued from page 278 Imore, UD, USA, 1980, FAO, Stale of Food and Agrrculture 1980, FAO, Rome, 1981 ‘D S Miller, Man s demand for energy, In K L Blaxter. ed, food Charns and Human Nutnbon, Applied Science Publishers Barklng UK, 1979 3FA0, The Fourth World Food Survey, FAO, Rome, 1977, S Reutllnger and M Selowsky, Malnutnbon and Poverty, World Bank Staff Occasional Paper No 23, Johns Hopkins University Press, Baltimore, MD USA, 1976 4P R Payne, Review of Malnutntlon and Poverty, Food PO/ICY, Vol 2, No 2, May 1977, pp 164-165 and Letter to the editor’, Vol 2. No 4 November 1977, pp 352-354, T T Poleman, A reappraisal of the extent of world hunger, Food PO/ICY, Vol6, No 4. November 1981, pp 236-252, P V Sukhatme and S Margen, Autoregulatory homestatic nature of energy balance , Amencan Journal of C//meal Nubt/on Vol 35. 1982, pp 355-365, TN malnutrition Measuring Snnivasan. Ceres, Vol 16 No 92, 1983. pp 23-27 M Lipton, Poverty, Undemutntron and Hunger, World Bank Staff Working Paper No 597, 1983
fact
requirements)
le use supply
and
the
data as intake
to contend
from national
suspect.
and
dlfftculty
diet Many comment whtch treat supply
calculations
to be grounds
derived
the problem ot ’ They discuss at length (and In particular the vahdlty of the
aboLe
requirements
an msufflclent
consumption
lrlruhr
food
thus
that
data
(Food
calculations
ot
defmmg
Hhat
on the mappropas rf they were
measurements
measurements
supply
that
of
briefly data
There would which are Sheets) are m
of Intake
Balance numbers
malnounshed.
adequacy of foodstocks and gaps’ and all the trend dnd ranking analyses In terms ot consumption which are based on these supply data, must be suspect
too
The food system There IS a deceptl\ely can be said to operate. Food
supply
simple series of levels namely Food
d\ailable
at national
level
Obviously level which
thrs system can be dmplifled contribute to the downward
+
household
dt bhlch
Food
dt
the tood
system
consumed
---) b\ lndlvldual
level
to Include subsectors at each flow (tor example food can be
grown or redred wlthm the country imported reledsed from storage) and cross linkages of markets which complicate the simple notion of movement
from producer
to consumer
For example
a household
can
grow Its own food, buy food on the market. recel\e tood as a image or exchange food as a gift with other households Figure 1 sho\is a stralghtformard point
out.
example the
of such dl\erslhcatlon
measurement
of
tood
Thus. as Burk and Pao
consumption
has
d different
ImtINtlons md
eatmgplacss.
Household usa m meals aacks.etc
Figure 1 The food system Source
flow of food from producer
to consumer
Modlfled from Burke and Pao, op cut, Ref 5
FOOD POLICY August
1985
279
4ssr,sr~lellr
01
errcrq\
rtmhc
‘M C Burk and E M Pao, Methodology for large-scale Surveys of Household and /nd/ndua/ f3efs. USDA Agncultural Research Semce, Home Eionomlcs Research Report, No 40 6There IS a wide literature on the technlcal and theorehcal issues involved In measuring an indivtdual s normal food Intake P V Sckhatme, The world s hunger and future needs In food supplles , Journal of Royal Sfafrsf/ca/ Socrefy America, Vol 124, 1961, pp 463, (one among many wntlngs by Sukhatme on the subject), J W Marr, lndlvldual dietary surveys Purposes and methods’, World Renew of Nufrrfron and Dlefefrcs, Vol 13, 1971, pp 105-l 64, G H
Beaton, J Mllner, P P Corey, V McGuwe. M Cousins, E Stewart M de Ramos, D Hernlt, P V Grambsch. N Kassins and J A Little, Sources of variance in 24-hr dietaryrecall data lmplicatlonsfor nutrition study design and interpretation 1Amencan Journal of C//nrca/ Nufrrbon, Vol 32, 1979, pp 2546-2559, K J Acheson. I T Campbell, 0 G Edholm, D S Miller and M J Stock. The measurement of food and energy Intake in man - an evaluation of some technaues’. Amencan Journal of C//nrca/ Nufnfion. Vol 33, pp 1147-l 154, E F Wheeler. Food con&mption surveys’, In J S Weiner and J A Laurie, eds, Pracflcal Human Bology. Academic Press, London, 1982, A Fei;o-Luzzl, Meaning and constraints of energy-Intake studies in free-lmng populabons , In ed. G A Harnson, Energy and Effort. Symposia of the Socrefy for the Sfudy of Human Bology Vol XXII, 1982 Taylor and Francis. London, 1982, J E Stuff C Garza E O’Bnan Smith B L Nichols and C M Montandan. A comparison of dietary methods in nutritlonal &dies’, Amerrcan Journal of Clmfca/ Nufrrf/on. Vol 37, 1983 pp 300-306, K S Todd, M Hudes and D H Calloway Food Intake measurement problems and approaches , Amerrcan Journal of C/fn/cal Nufnfron. Vol 37, 1983, pp 139-146, J
unpllcatlon
m different
consumption
parts of the system. and the concept
itself vanes wth
I&r bwilral iilefar\-
of the researcher
ot tood ’
sw ise\ s
Interest IS usuallg tocused at the right-hdnd end ot the tood system. dt the lndlwdual level and on Indl\ Idual dietdrv sune!,s hledsurlng energ! dnd nutrient
Intakes of Indl\lduals
with an> preclslon
~hlch
record all food eaten o\er several davs ‘Such
costly
In
time.
u hat a pxtlcular
measure
period Intake
le the drnount that might be called the true . normal Intahe
there
are
weighing
dnd
behdl lour Intahe no\elt\
mrthodolo~lcdl
Ingredients
Direct obsendtion of meals dnd Ilksl! to cause abnormal
problems
IS verb Intruwe
The more rxeclselv dn dttsmnt IS made to measure the actual the
behd\lourdl
less lIkeI>
It
IS this
dcconimoddtion
mdg
represents
incredse
the
in some
ot being \\atchrd
ot the experience
nlternatl\e Imolvlng
normal
Indeed
circumstances
ds the
and measured Hears ott The
to weighing and measurmg IS to use questionnaire techniques memory or estimates ot portion we. these are generalI\
considered
The
dre
HoHe\er. if the nlm IS to obtam an estimate of the hdbitual of food dnd nutrients tar &I particular Indl\ldual or group ot
indiwdudls then
requires sur\eJs
measurements
and money It IS nonetheless possrble to mdl\ldual has eaten o\er a nartlcular time
motl\atlon
to glie
unreliable
IlteratureL’on
food Intake
IS Important
research
mlphed
or
ot
true
dlfflcultles
Intakes’
In Indl\lduals
of measuring
because so many questlons.
whether
In pol~cv Issues, seem to hinge
much toad particular hnow about
estimates
these technlc,ll
Intakes
people are customarily
ot Indl\ldual
people
posed In
on hnowng c horn
edtlng
wIthIn
-
Indl\ldual
It IS necessary to
particular
populations
to
understdnd ISSLIZS ot glob‘~l hunger and mdlnutrmon - how this problem IS detlned and the Impllcatlons of the numbers ot people - and tor epldemlologlcal
Involved
studies ot the Imh beween
Intake ot
energ\ and nutrients trom Particular sources dnd Ldrious malor , LdlseJses In other \\ords there IS a need tor lndlc‘jtors of consumption that dllon d problem to be defined in terms ot mdgnitude dnd numbers of
people
’
lndlcators
and
’
to
predict
future
I_!nfortun;ltel\
Mullen. N J Krantzler. L E Grwettl M G
The suneys
Schulz’ H L Melselmai, Validity of a food frequency queshonnalre for the determination of Individual food intake , Amerrcan Journal Cfrnrcal Nufnhon, Vol 39 1984, pp 136-l 43 ‘Marr, Beaton ef al. Ferro-Luzzl, Stufi ef al Todd et al, rbrd, M Baloqh. H A Kahn and J M Medalie, Random repot-l 24-hour dietary recall , Amencan Journal of Clrnrcaf Nuhffon. Vol 24, 19711 pp 304-310 A Lechtlg C Yarbrough, fi- Marlorell. H Delaado and R E Klein. The one-dav recill dietary survey a review of its usefuiness to estimate protein and calone inArchwos Lafmo-amencanos de take’, Nufrrcron, Vol 26, 1976, pp 243-271 Both Ferro-Luzzi and Todd ef al discuss ways of ImprovIng the estimate of an lndlvidual s usual Intake by increasing the number of days measured Both suggest that by measuring one-day Intakes for five to SIXdays, the estimate can be brought to wIthin 10% of the true Intake with 95% Conbnued on page 28 1
neYt lekel up on Figure
280
the objectlkes
thdt hais
the heen
done
COIISW~~~IOIIslit I ~\~s
Household
intdke
Indl\lduals
measurements ’
dre
ot
changes
1s generally
\er!
otten
In
those
not a\allable
measurement\
at the
1 the household
HIM-ehold about
consequences
lntormatlon
Techniques
bl detlnltlon \drj
from
do not reveal anything
\\el_ghlng
or
documentmg
Intakes ot meals to diary recording and r&zall Intervle\\s Food eaten outside the home. and the presence of non-household members during measurement are sometimes recorded and ‘adjustments made Many of the technical problems (accurxq Lersus lntruslon sample size. repetition rates and so on) are slmllar to those of Indl\ldual dietar! surveys ’ Some are lntrlnslc to household studies -such a\ what constitutes a household’ and ho\\ to avoid double counting where food IS prepared m more than one kitchen or eaten In chdnges
in larder
stocks.
or NeIghed
more than one unit ‘) Further sources of error In Household
Consumption
Surveys (HCS)
Include potential sampling bias In selectlon of households and m time I” In mdnv places toad suppI!, or usage may be quantltatlcely dnd qualltatlvelv hlghlv dependent on season for some or all of the
FOOD
POLICY
August
1985
populations
concerned.
fact of course. regardless
and for some or all of the food consumed
most natlonal
of seasonal
surveys are carried on throughout
factors.
though
some puhhshed
In
the year
analyses may
Indicate their effects (In some circumstances these seasonal factors may become so marked as to alter sample accesslblhty Where this IS the becomes more speculative. especially ds this case, mterpretatlon problem IS often the most elusive to remedy ) A mean Intake expressed as for a whole year may Indicate actual Intakes for a particular group at any one time and therefore would not be representative of true Intakes of the
whole
potential Continued from page 280 confidence Thereafter, little Improvement of the estimate occurs with addItional days measured ‘For dIscussIon of methodology of household dietary surveys see Burk and Pao, op at, Ref 5 and Wheeler, op cit. Ref 6 ‘Wheeler, op c/t, Ref 6, M T Martin. MSc thesis, food fntake m hrvo Bambara V/f/ages m Ihe Segoo region of Ma//, Department of Human Nutntlon, London School of Hygiene and Troprcal Medicine, University of London, London, 1982, D J Casley and D A Lury, Data Coflecfron m Developrng Countrfes, Oxford University Press, Oxford, UK 1981 ‘°Casley and Lury. /b/d “The UK National Food Survey (NFS) publishes quarterly and annual results by factors such as geographical area, household size, occupational class, and by other factors of Interest such as age of housewife, ownership of freezers ‘*For a dlscussion of how the NFS allows for wastage and losses MAFF Household Food Consumption and Expenditure 1980 With a Review of 1975-1980, Annual Report of the National Food Survey Committee (NFS), HMSO, London, 1982, also 1981 Report ‘3E A Dowler, A pilot survey of domestic food wastage , JoJrnal of Human Nutrition, Vol 31. 1977. DCJ171, R W Wenlock and D H Buss W&age of edible food In the home a prelimin&y study’. Journal of Human Nutrition. Vol 31, 1977, pp 405, R W Wenlock, D H Buss and B J Deny, Household food wastage in Britain’, &itrsh Journal of Nutnbon. Vol 43, 1980, pp 53 141b/d 15E P Cathcari and A M T Murray, A note on the percentage loss of calories as waste In ordinary mixed diets’, Journal of Hygiene, Vol 39, 1939, pp 45, Studies on food wastage In the USA Include S F Adelson. E Asp and I Noble, Household records of foods used and discarded’. Journal of the American Defebc Assoaaf/on, Vol 39, 1961, pp 578. C G Harrison. W L Rathfe and W W Hughes, Food waste behaviour In an urban populabon’, Journal of Nutntron Educatron, Vol7,1975, pp 13 M V Zaehnnger and J 0 Early, Proceedings of National Food Loss Conference, University of Idaho, Boise, ID. USA, 1976
FOOD
POLICY
August
1985
time
period
Such means are also unlikely
problem periods for any particular
group
to Indicate
nor would
variations
from one year to the next necessarily show up even If the surveys themselves were annual In all circumstances. the potential use of natlonal
HCS
population a glken
to detect and predict
problems
of food Intake
over time will depend on the particular
country
theoretical
-
and
account
how
closely
In published
the
results
methodology
survey
practice
Few authors
within
a
used In
follows
the
discuss this Issue.
nonetheless, the potential for HCS to represent actual mean Intakes IS there. not least In that results can be dlsaggregated by geography and time
Intakes
obtained I’ A potential
of
particular
source
groups
of error
of
Interest.
In national
bv
HCS
season
which
can
be
has recently
attracted particular attention In parts of Europe and the USA IS the Issue of domestic food waste and losses In fact few accounts have been published
of small-
or large-scale
surveys
of mastage
at the domestic
level In most national surveys a somewhat arbitrary allowance IS made for wastage. In the UK for Instance when the estimates of consumption comerted
to energy
recommended
and
Intakes.
nutrients
are
a 10% reduction
expressed IS made
as percentages
to allow
for plate
of and
kitchen wastage. spoilage and feeding to pets’ ” A Food Waste Survey LJmt was set up In the UK by the MInIstry of Agriculture, Fisheries and Food (MAFF) In 1976 wlthm the National Food Survey It concentrated Its mqulr! with a particular concern for measuring
on the domestic tood tor human
level at tlrst. consumptron
fed to animals I3 Two pilot surveys and a national surbey were conducted to collect all food Intended for human consumption but not actually eaten o\er one week The energy protein and fat content of this wastage
was measured
and presented
as a proportion
of the energy
recorded as entering the household over the same perrod were done at different times ot the year and on dlfterent
” The sur\eys sample sizes.
nonetheless
only
the results
were
consistent
and showed
that
-l-6”.
of
potential food energy was discarded This result compares well with that of a UK surbe! In the 1930s of 2-3’4~ and with those available from the USA of about 7% ” Further
technical
errors
m HCS can occur when
recorded
Intakes
are
translated into nutrients and calories by means of local extraction rates and food tables, these sources of error are possibly those most amenable to regular
momtormg
Several throughout
countries carry out regular sample surveys of food Intakes the nation, using the household as the consuming unit desptte the mtrrnslc appeal of the prmclple of measuring of food actually consumed by the household - errors and
However, quantltles
uncertamties consumption
and potential
ellmlnatlon
notwlthstandmgmany countries do not mount household surveys at all, or only at irregular Intervals Thus, the
source ot InformatIon about Intakes m these countries even further back on the aggregation scale m Figure food
supply
Nlltlonal
or food
toad
balance
halunce
has been the lebel 1 th,~t ot natlondl
sheets (FBS)
sheets
Food balance sheets (FBS) are compiled regularly In man! countrtes to give a ‘comprehenwe picture of a country \ food wpply durln_r d specified
reference
Imports
and exports
for seed
animal
” Account IS taken of productlon ‘md stock changes for dll tood productlon.
period feed.
manufacture
storage aldlldble
and transport subtracted tor human consumption
decades
of eftort
It
sizes m estimates
IS
for non-toad
use dnd losses during
The residual IS tdken to dt the retail level Despite
still admitted
occur
ewmates H Ith uses
that errors
throughout
of kno\\n
the system
be th,u several
and unhnown
for reasons
discussed
else\\ here I7 and that these errors are therefore compounded In the tl_rures d\allable for human consumption Several authors suggest that the technques systematically underestimate the amount a\allable tar consumption
m less developed
countries
concerned “( The food a\dtlable tor consumption cdplta hms. tor \\hlch use IS made mid-Jeer
estimates
of
populdtlon
partIcularI\
\\here
IS
energy
IF
usuallv presented on a dally per ot the LJN Populatton DI\lslon
we
(These
estimates
allo\\
for
lmmlgrdnts tourists and special groups such ds retugee camp occupants, but nonetheless contain their own Sources ot error ) Food IS expressed In terms
of
Indl\ldual
commodltles.
dnd
ot
calories
and
nutrients
obtained from those commodltles bj reference to detdlled specifications ot each foodstuff (such as water content, mllllng extraction rates) and use of regional FAO Food Composition Tables Each of these stages Introduces further some extent the
errors. degree
another
time.
and over
some of which can be quantified However of ImprecIsIon varies trom one country so that generalizations
ments are dtfflcult to mahe The FAO publlcatlons all 16For descn@on of the construction of food balance sheets, FAO, Food balance sheets 1975-77 average and per caput food supplies 1961-65 average 1967-77 , FAO. Rome, 1980, for dIscussion of the and problems, PV methodology Sukhatme, op nl, Ref 6, FAO, Provisional food balance sheets’. FAO Rome. 1977. @leman. op crt, Ref 4 ’ ‘Ibrd ‘BPoleman, op CII, Ref 4, Quantifying the nutntlon sltuatlon In developing countnes , Food Research lnsfrtute Studres, Vol XVIII, No 1, 1981, pp l-58 ’ ‘/b/d 2o At present the FAO food balance sheet represents the only source of worldwlde InformatIon on Food supply/consumption levels National surveys provldmg food consumption data are still limited In number and usually correspond to different time periods’ These are the opening sentences In FAO A Comparatwe Study of Food Consumpbon Data from Food Balance Sheets and Household Surveys. Statistics Dmsion, FAO, Economic and Social Development Paper 34, FAO, Rome, 1983
estimates
ot per
capita
aterage.
vex-round
potential
LarIatIon
re.glon to another nor
IS It
kno\\n
basis
lmposslble
In addition
In supplq In a country
those
Furthermore
IS
of Hhat
on dn
It IS eLIdent
accounted IS
bought
such
e\en
that
and trom
one
for In dnv uav.
‘~ctuallv
consumed
as Polemdn.
let alone
to measure
The
of
data
uSes
precIseI\
FBS
such
seasonalitv
tor
m eftect
these
dlstmctlons
esttmatlng
Intormatlon. IS
measurement
and tor
\\lth
dnv xcurx\
consumption
changes
365 days
ocer
and
would
time
cannot
I’)
require
Equdllv
take
the
account
ot
in int&e
Per caplra
ciarly
mttakrr
the
method
that
dlfterent.
IS not
adlust-
taking
the year (In time)
(In place)
what proportlon
against
consumption
to the errors
throughout
countrtes.
Given
cnutlons
to represent
potential
among others point5 out It IS to dlsttngutsh quantities entering dnd Ie,wng the retail sector produced dnd used entirely for home consumption In some
or who eats It and
contain
supplies ..
about
to to
It
IS In
calorie
Intake
many
countries
A cornparlsorl of
collectmg
some ways surprtsmg
aj two data
data
In
the
sources tHo
that the results
methods
IS so
In terms of dallv
per capita should be thought comparable Ho&ever not have Invested m nattonal HCS. certdlnlg not on an! regular basis. and reliance has therefore been placed on supply estimates - FBS - ns a basis for consumption Indicators “’ Is this reliance
FOOD
POLICY
August
1985
Justlhed” countries
those \%ho ulsh to calculate Intakes for Furthermore that do not ha\e regular HCS sur\e>s use FBS data as a
reterence
for data from other
sources
such as household
budget and
e\pendlture surveys. market sur\evs ,~nd Indlvldual dletarv surkeys ‘I Is this use of FBS valid” If per capita dally calories from HCS and FBS dre compared for any g~\en countr) the estimates from FBS seldom match HCS,
for developed
Table I sho\\s data and FBS for We
countries
the) are almost alHd>s more than HCS
kIlocalorIes (kc&) per capita. per day from HCS South-east Asian countries - Indonesld. Jap‘m
011
hlala>w Phlhppmes and Th‘lrland It cdn be seen that there IS a dlscrep;lnq between the t\\o sets ot data rnnglng from ,ibout 750 kcals (Thailand) to about 610 kcals (Japan) This obser\atlon IS not neh It has been known tor man! years and 21Reutllnger and Selowsky, op c/t, Ref 3 present an estimate of numbers malnounshed In the world where the Intake side of the equation IS measured by exploltmg the known relationship between Indome -and calorie consumption from household budget surveys A fuller cntique of their methodology can be found In Ref 4. but It should be noted here that they force’ their household budget data to ht that from FBS They say of HCS data that they cannot be used directly for assessing the global dimension of food consumption by Income groups without first reconciling the extrapolated data from such surveys and available data about national total food consumption’,by which they mean FBS “M C Burk and E M Pao, Analysrs of Food Consumpbon Survey Data for Developrng Countnes Food and Nutrition Paper 16, op c/t, Ref 20 Z3MAFF, op c/f. Ref 12 24Further data from these countries, which were chosen to demonstrate the dlfferences In national development in welfare status for the poor, form the basis of an m-depth study of nutntlon Indicators at different stages of national development See Y 0 Seo. Patterns of nutrition indcators at different stages of national development , unpublished PhD thesis, Faculty of Medicine. University of London, London UK, 1981
literature
Discrepancies
gathered
nuts
and
oler or
truer.
dlscussed
IS usualll!
defmlrlon alcohol
ot are
food
III p,lsslng
(u hether
Included)
and
In the
casually the
non-
of time periods are the usual rr‘rsons glien
” Ho\ie\er. figures ‘Ire usuall! published Inclusive and e\clusr\s ot alcohol. and aild foods bvould make little substantial difference to mean energy figures compdrablllty
Another
posslblht!,
eaten
rxpldnatlon
IS
that HCS tall to tahe sufficient
IS
the
outside
home
Inadequate
Such evidence
as there
account of me,lls IS.
suggests this
to account tor the slzeable dlscrepancles
”
The five countries whose HCS dnd FBS data dre quoted were selected tor stud!
because
their
polItIcal
and
frze-marhet
economic
systems
\\ere
ethnic groups (Chinese. hIala> and Thai) and 111 the troplcal zone between 15” latitude north and south ot the
slmrldr.
as mere
location eyu‘lltor
Data
their
on J,lpan
similar chdracterlstlcs
were
collected
so that
a debeloped
country
Hlth
was Included In the stud) ” It the data trom these
Include an Indicator ot economic development. such as GNP per head. and the countries are ranked accordlng to GNP It cnn be seen that both consumption and suppl) tlgures rank them In the same order
countries
as GNP rhan
(Figure 2)
the
IncreasIng tionship
GNP
e\tensl\e
not only
national
the FBS estimate
dlwaq’s
higher
tood supply estimates
from FBS and mtdhe
not constant betiieen dlfterent countries Moreover. changes Mlth GNP this suggests that the dlscrepdnq
as the economic - and the food - system becomes more and complex It IS also hheli that \%ith Increasing complevtv.
Table 1 Energy supply and consumpbon countries (kcals per capita, per day)
Food balance sheets’ Indonesia Japan Malaysia Phlltpplnes Thalland
IS
but the dlscrepanq Increases ulth obser\atlon suggests that. Hhate\er the relameasure
IS. It IS
rel,itlonshrp
Increases
This
between
from HCS the
However
consumption
2116’ 2833 2521 2290 2282
measurements
for selected
Household consumption survey
South-East
Aslan
Discrepancy
1832= 2188’ 2085e 1804’ 2037g
284 645 436 486 245
Sources a FAO op ot Ref 16 except Phlllpplnes Food and Nutntlon Research lnstltute
Manila the Phlllppmes Pub No 75 1976 ’ FBS for lndonesla 1970 1972 kcals per capita This reduces the discrepancy to 140 kcals ‘Bureau of Statlstlcs. lndonesla Survey Social Ekonoml Naslonal (SVSENAS) 1970 d Annual Report Naf/onal Nulr~bon Survey, Mmstry of Health Tokyo Japan 1975 ’ lnstltute Medical Research Mmlstry of Health Kuala Lumpur Malaysia 1975 ’ Food and Nutntlon Research lnstltute Pub No GP-11 1979 and Income and Food Consumpf~oon Report 1975 Mmlstly of Agnculture Quezon City. Mamla. the Phlllppmes, 1979 g CornplIed from Mlmstry of Public Health Household Expenditure Sutvey!Soaoeconomlc Survey 1975-76 Nattonal Statlsbcal OffIce National Social and Economic Development Board 1977-76
FOOD
POLICY
August
1985
283
(Food balance she&
Uiousehold consumptmn survewl
220 350410
Figure 2. Cross-country
I
Indonesm Phlhppmes Thatland
comparison
the
system
%tttsttcal
becomes
more
which
sophisticated
760
2550
Malaym
Slwre
4450 Japan
documents the economic system Itself It may well reflect more accurately
Thuc
the food supply per capita within a country, that IS. as national Income Increases. FBS data become more rellahle Several writers. Poleman In particular. habe commented on the probablllty that FBS consistently and perhaps grossly underestlmates the reahty of tood supply In developing countries ” Indeed It IS hkelv that HCS also under report consumption for very poor households The latest study from FAO documents a number of studies comparing household 25Poleman, op crt, Ref 18 26FA0, op cut, Fief 20 “/b/d, p 14 “/b/d. p 8 =/b/d, p 15 30A J Culyer, Measunng Health Lessons for Ontarro, Ontano Economic Council Research Studies 14, University of Toronto Press, Toronto, 1978, N Baster, Socral hdrcators and Socral Sfatrsks MI Conlexf of FAO’s Concerns, ESS/MISC/78-5, FAO. Rome, 1978, R A Bauer, ed, Social Indrcators, MIT Press, Cambridge, MA, USA, 1986, E A Dowler. P R Payne, Y 0 Seo, AM Thomson and E F Wheeler, Nutntlonal status Indicators - mterpretatlon and policy makmg role’, Food fohcy. Vol 7, No 2, 1982, pp 99-l 12 In particular, this paper draws a dlstlnctlon between shorthand Indicators (ie those that act as an indirect measure of the variables of Interest because they are quaker, easier. cheaper to measure for example, the price of rice can be used as a shorthand Indicator for all food prices) and proxy Indicators (ie those Indicators which stand as proxy for a set of variables or aspects of a particular system which cannot as yet be precisely defined For example, the Infant mortality rate might be used as a proxy indicator of health - health being a concept open to multiple definetlons and Interpretation )
284
data with FBS estimates (Table
3) ” The discrepancy
varies
this time between -579 kcals to +llOO The authors suggest particular food group\ which thev speculate might contribute to the differences observed (mostly nuts and oilseeds. fish, sugar products. and roots and tubers) though with no attempt to quantify their contnbutlon However their main conclusion IS that since FBS are estimates at a different level In the system from household suneys, the latter by defmmon not accounting for non-household food use (which they think might be
substantial In ‘developed’ countries). FBS data should be larger than the data derived from household surveb\ The mdgnltude of the discrepancy varies trom one country to another’ 2T We would not disagree with them thus fx where we hould part cornpan! IS In their contlnual assertion that Hastage ‘It the household level especial& In developed
countries.
final conclusion tood balance
that.
for part of the dlscrepancv ” and In their In the absence of datd from household sunejs.
account\
sheet data provide
a good proxy
for food consumption
levels ot the population’ “) Much has been wntten else&here on the theoretlcal nature dnd “‘The essential characterlstlc of an Indicator practlcdl uses of Indicators IS that It should bear a predictable consistent relatlonshlp Hlth the quantity It IS representlng or descrlblng In other words. one could tolerate FBS as shorthand Indicator5 of food consumption It one could characterize the relatIonshIp between FBS measures dnd actual consumptlon surveys The realit! IS that this cannot be done the relatlonshlp between the two sets of data IS not constant between
countries
nor necessarily over time
FOOD
POLICY
August
1985
Table 2 Household data and food balance sheet esilmries Per capita dally calones
Country survey DEVELOPED
Difference As percentage of per capita daily calorie derived from FBS
HS
FBS
Number of calones
2640 2696 2694 2122 2454 277 I 2313 2254
3293 3049 3150 3208 3328 3165 3169 3111
- 653 -353 -456 - 1086 -874 -394 -856 -857
-198 -11 6 -145 -33 9 -26 3 -124 -270 -27 5
1986 1917 2170 2817 2514 1951 2054 2192 2633
1880 2249 2026 2249 2497 2223 2039 2363 2501
+106 -332 +144 +568 +17 -272 +15 -171 +132
+56 -148 +7 1 +25 3 +o 7 -122 to 7 -72 +53
2110 2292
2439 2860
-329 -56%
-13 5 -199
COUNTRIES
The mcome,expendrfure‘budget survey Auslna (1974-75) FInland (1976) France (1977) FR Germany ( 1978) Greece (1974) Hungary (1971) Norway (1977-79) UK (1979) DEVELOPING
COUNTRIES
The ~ncome,expend~lure~~udgel survey Bangladesh (1973-74) lndonesla (1978) lndla (1971-72) Iraq (1972) Morocco (1970-7 1) Pakrslan (1971-72) Phllrpptnes (1975-76) Sri Lanka (1969-70) Tnnldad and Tobago (1970) The food consumptron survey Nofe HS = Household Survey Balance Sheet See text op at, Ret 20
FBS = Food
Brazil (1974-75) Turkey (1974)
Where does the energy go? The relatIonshIp between natlonal food supply estimates from FBS ,md Intakes from HCS IS not constant among countries. and It changes with GNP We suggest that the Increase m discrepancy between the two with rlslng GNP IS a consequence of the mcreasmg comple\lty of the food system - In storage. transport. processmg and marketmg - and Its as yet undocumented losses. particularly of energy Support for thrs Idea rn addltlon to the cross-sectional data above comes from a fairly unique set of longltudmal
data
from
Japan ‘I These
enable
comparisons
to be
made over time A natlonwlde survey of household food consumption (seven-day weighed Intakes) has been carried out annually for the last 30+
years
period
During
the 1950s and
of rapid economic
growth
196Os, the country and was transformed
experienced
a
mto a highly
centralized, mdustrlahzed society Figure 3 shows the average calorie Intake from HCS tor vanous years and the corresponding value for average calorie suppI) from FBS for the same years, plotted against GNP (US$) on a log scale The average calorie Intake from the HCS remdined reasonably constant over the period considered However per capita calorie supply from FBS rose rapidly as GNP Increased. especially after 1958 when GNP Increased rapidly Before then. It was about lo_?-104% of the per capita calorie intake from HCS By 196-I. It had risen to 115% and by 1973 to 129% Thus the reldtlonshlp between
31Seo, op at, Ref 24
FOOD POLICY August 1985
measurements of food consumption and food supply m a country which experiences rapldly lncreasmg Income IS quite complex, and changes with the level of income
285
(Food balancesheets1
A FAO countt-v reouWement (Hourehold ConslJmptlon SUI-WVSI
Figure 3
Longltudmal
comparison
(Japan)
I
I
I
I
I
220 1963
350 1958
760 1964
2450 1972
4466 1975
The FAO Statlstlcdl Offlce the discrepancy at mdlvldual that
further
study
of the
GNP US t log scale
Study urges detailed stud) ot the nature of country levels ” It certainly seems to us
Japanese
system
would
yield
some
required Insights Data from the UK has been mvestlgated wds expected that a country llhe the UK, where the tendency larger
productlon
cfflclent records.
units.
centralized
an earlier productlon
and
reasonably
annual figures of National supplies of produced In a similar wa) to FBS for from statistics ot total tood supplies at
stage of the dlstnbutl\e chain rnther than These consumption lekel estimates (CLE)
more accurate and less of an overestlmate the) are denved by d more sophlstlcated energy
keeping
sytems of taxation crop yeld estimates and stock movement would be able to rndlntam d high degree of accuracv In FBS
estlmdtes The MAFF publishes food moving into consumption’. FA0.‘3 but In this case derived
32FA0, op c/f, Ref 20 336rdrsh Busmess. (prewously In Board of Trade Journal), for example, Vol 37 No 11, pp 573-575, 14 December 1979 gwes consumption supply estimates for 197578, MAFF, op c/t, Fief 12 34NFS. personal communlcatlon 350p crt, Ref 33
record
of the
turther It IS towards
as kcals per capita
for
1975-80
trom on-farm dre said to be
than the FBS figure because method ” The figures tor
are shown
In Table
3 and var!
between 2920 and 3850 ” However. the national HCS for the same period gives 2390 to X30 kcals per capita. per day the discrepancy IS around
6X1 kcals
That
IS,
the best estimate
of mean energy
Intake
m the
Notes NatIonal averages In kcals per capita. per day a These eshmales rnclude lood used In manulacture of soil drinks and sweets, food consumed In catenng estabkshments or rnstrtuhons(hosprtals boardmg schools, pnsons HM Forces) and purchased by mdrvrduals but not taken home lo the household supply Excludes kgures on alcohokc dnnk purchase shown under (v) Dna means not avariable, ’ NFS figures do not Include purchase of alcoholrc drinks sugar or chocolate conlectronery. sofi dnnks or meals purchased outslde the home Assumptrons and adjustments rn analyses concernmg the latter are set out rn the NFS publrcatron (op err. Ref 12) “Smce 1975 purchases of soft dnnks for the household supply have been recorded analysed and the figures presented separately, e Obtamed from the CLE ‘Thus figure skll excludes con sumptlon of alcohokc dnnk Source MAFF op crt Ret 12
288
Table 3 Nutritronal value of household food m the UK 1976-81
I) Consumptron level eslrmatesa II) NFSC kcals (MJ) (I) - (11) III) Soft dnnkf kcals IV) Sugar and chocolate confechonelya v) alcoholrc dnnks VI) (II) + (Ill) + (IV) (to nearest 10 kcals) (I) - (VI)
1975
1976
1977
1978
1979
1980
1981
2920
2920
2930
2920
2950
2850
na”
2290 (9 6) 630 I7 133
2280 (9 6) 640 21 139
2260 (9 5) 670 18 138
2260 (9 5) 660 19 144
2250 (9 5) 700 19 141
2230 (9 4) 620 21 133
2210 (9 31
160 2440
166 2440
164 2420
176 2420
181 2410
174 2380
166 2360
480
480
510
500
540
470
FOOD
POLICY
21 130
August 1985
.-lssrattwnr01
data IS 3SV0 higher
the natlonal
the discrepancy
HCS
than measured
The MAFF
for a number
mean energy
has been Investlgatlng
of years and one potentId
Intakes
the basis of
explanation
was
thought to be the underesttmatlon of true Intakes dt the household lebel because of waste and losses The findings of the Food Waste Surve! Unit published so fdr suggest that domestic unlikely to exceed the 10% allowance already Some further work examlnmg the accuracy
Hnstage and losses are made jh of the supply estimates
and documenting losses In the system (that IS. between farm gate and food entering the household) has been published. much IS as jet unpublished For Instance. a recent UK dnal>sls b> Rov suggests sources of error m FBS from losses on farm (nearlv XI”,) and during processing
(about
9%)
losses to the food
” The rrlatlonshlp
system,
hedged
between
about
with
Roy’s
careful
estimates
deflnltlons
of and
ranges of figures as the! are, dnd the process tor cdlculatmg CLE 15 not cle‘lr, but may well contribute to an understanding of the dtscrepanck One problem Roy faced was the absence of reliable relevant mformatlon the UK food system 15 tdlrly centralized and documentable surveys m the 19hOs or from 36The publlshedlosses of 4-6% potential food energy being dlscarded IS presumed to be a mlnlmumestimate and the deductlon has therefore remained at the conventlonal 10% level, op cut, Ref 6 37R Roy, Wastage m the UK Food System, Earth Resources Research Publlcatlons. London, 1976 The percentages given refer to wastage as a proportion of edible material entering supplies or of potential agricultural yield (~7) The Report examines the problems in defining waste and losses’ between farmgate/dockside and the household and, stressing how widely ‘percentage wastage’ vanes from one commodity to another, emphasizes how mlsleadlng average figures can be when they are used out of context of the data and calculations on which they are based Nonetheless, Roy gtves a good summary of then existant evidence for the proportion of food destined for consumptlon which IS dlverted to other uses at each stage of the food system and that which flnally reaches the plate and IS consumed =D D Singer. Food losses In the UK, froceedlngs of the Nufntron Socrefy, Vol 38, 1979, pp 181-186 39R Osner, ‘Food wastage , Nutntlon and Food Scrence, No 77, 7982, pp 13-17. Rov. OD c/t Ref 37 All these authors (Refs 37:39) summarize work from several surveys Including their own Figures given are for energy In plate waste as a proportion of energy consumed 4olb/d 4’See also Anne Chinnock, Food consumption in Great Britain 1969-1978’. MSC Thesis, Depattment of Human Nutntlon. London School of Hygiene and Tropical Medicine. University of London, London, 1981 42K Mellanby, ‘Wasteline , Nature, Vol 257, 23 October 1975, pp 639
FOOD
POLICY
August
1985
In another. briefer largely undocumented
contemporary Fophlstlcated.
let many ot the data he quotes the LISA
are from
summary rebleu Singer Instance\ examples ot food waste m food proces>lng and dlstrlbutlon
both of that Intended for human consumption but not reaching the household (as In the retail sector - unsold fresh tood such as bread or milk
\thlch
foodstuffs
might which.
be returned through
to the supplier
processing.
or destroyed)
become
unavailable
and
for consumption (such as sugar effluent In s\ieets preserkattie ture. losses In fish processing to produce tush fingers) ” Work
has also been conducted
on catering
kitchen
of
or unusable manufac-
and plate
u‘lstage
In mstltutlons m the LJK 3y Such data suggest that the losses as a percentage of Indl\ldual mean intakes ot energb are higher than In domestic households and more Larlable Not surprlslngl\. \iastage In hospitals (hotels.
tends to be highest (up restaurants and conterence
to XI”,). hdlls)
IS
that In prltate catering ne\t highest (10-17%).
wastage levels m schools and work/student retectorles to that m domestic households (3-6”0) “’ The
UK
N,ltlonal
consumptron outside
the
Food
surveys. home
does
The
Sur\ey. not
In common
measure
published
food
reports
seem to be slmllar with
man\’
purchased
national and
eaten
discuss this Issue and the
allowance made In terms of meals dnd also the purchase and consumption of sugar confectlonerq, sott and alcoholic drinks Table 3 contains their esttmates derived from supply tlgures. dddlng those tor sugar and soft drinks to the consumption tlgures and comparmg these with CLE still leabes a dlscrepancv the HCS averages ” This used Intakes
issue
IS
to assess across
obvIousI) the
adequaq
developed
ot around
Important of
given
Intahes
500 heals
le Jbout
the srtuatlons and
\h here FBS are
In particular
and de\elopmg countries
It
22% ot
IS
comparatl\e
also Important
given the prominence of the word ‘nastage’ m man! current dlscusslons of the problem It IS all too easy to use the emotlbe word waste’ wthout consldermg Its tmphcatlons The Chief Scientist at hl;l\FF. to take but one example, has been quoted as clalmmg that of the toad bought In Bntam. as much as a quarter may be wasted’ A?The moral lmpltcatlons of the Idea that profligate consumers of the developed world are
carelessly emptymg valuable calories down their throats, their sinks or Into their dustbins (or worse still teedlng them to their pets) are attractive to Journalists. pohtlclans and even academics who wish to castigate the wasteful, but the facts do not bear out the allegations ‘Waste as a teature of atfluent societies does not seem to apply to the member cltlzens at least where household food IS concerned Indeed. Roy comments as a result of his stud! that the evlstence ot a certdln IsLeI of wastage
m the tood
since It represents of necessity’
system
a slacb which
ma!
e\en
can to some
be considered
extent
be taken
desirable up In times
-I’
Conclusions Despite
several
years’
effort
to Improve
the
technical
aspects
ot the
methodologies. the two sources m common use as Indices of nutrient consumption gibe dltferent aberage per capita figures tor dally energ) Intakes constant 43Roy, op crt, Ref 37, p 29 It IS possible he was bemg misquoted, as the NatIonal Food Survey m the UK records food bought at the household level, and has used a 10% wastage estimate for many years, perhaps he said ‘brought which would be nearer the mark Alternatively he may have been mcludmg food purchased by restaurants, canteens and institutions, which would make the estimate more reasonable 440p at, Ref 1 ‘?Iavld Southgate, In an editorial in the Bntlsh Nutntlon Foundation, Nulr~tron Bulletm, No 41, Vol 9, No 2, 1984, pp 57-58 Consumption level estimates and the NACNE discussIon paper’, comments adversely on the relationship between current UK recommendations for healthpromotmg’ diets and the measurements of intake on which this advice IS based
288
cy
IS
Furthermore. the between countries.
related
and that
size of this or over tune
to at least one other
the nature
have demonstrated of the notion that
of tha
difference IS not necessanlq This change In the dlscrepan-
socloeconomlc IS
relationship
Indicator
as yet unknown
that of GNP. Flnallv.
we
our rejection (at least where the UK IS concerned) wastage at the household level can account for the
discrepancy Food balance sheet data are theretore an unreliable calorie Intakes and. In addition. are subject to systematic
estimate ot error which
becomes
of the food
progressively
larger
~lth
increasing
sophlstlcatlon
system Household consumption surveys. despite their Inherent sources of error, are more hkely to give an accurate picture of calorie intakes Statements concerning comparisons among countries’ consumption levels.
about
SOFAq4
or
about about
the relatlonshlp between food/nutrient intakes and health,J’ the size and nature of International ‘toad gaps or estimates
or of
the numbers
the slgmflcance
malnourished
data and not on consumption.
of trends
m time for example
In the world should
that are based on food
be regarded
FOOD
supply
with great susplclon
POLICY
August 1985