a-.: -j* -i 1. F} -ff -~* ..............., IK The Dmad for Calories in Developing C( ain to Countries i n| K. Kndsenand L Sdte Odin K. Knudsen and Pasquale L. Scandizzo th tv te This paper employs characieristic dcemand theory to estimate demand functioi's for oe calories for a set of developing countries and to investigate the potential impact of income growth, redistribution, and price changes on alleviating underconsumption of calories. The analysis finds that, although calorie elasticities with respect to income are R suibstantial for the poorer consumers, income growth above historical rates is required SC for the food needs of the entire population to be satisfied within the next fifteen years, tri even if calorie prices remain constant and income distribution does not become more le, unequal. m da Key wtorilds: calorie consumption, demand theory, economic development, income distribution. CC D. ne This paper uses household survey data on sizable calorie price increases would not rule ch consumption to analyze the determinants of out elimination of malnutrition. In this case, di: calorie intakes in developing countries. The price increases could generate production in- pr paper relies on characteristic demand analysis creases without ill cffects on the nutritional cs for a demand function specification for status of the poor. calories. It explores the effect of calorie price p1. differences, income, and other socioeconomic m factors on the intracountry and intercountry The Theoretical Model distribution of calorie intakes. Using the esti- mated functions, wex make broad macroesti- ..s mated func tio, make brad macoe The recent focus on malnutrition by econo- cY. mates of the potential impact that income it a entruhicm-ru-pcfcW growth and redistribution could have on al- mists has been through income-group-specific e m o demand analysis. Three approaches have ta eviahng malnutl,fon. The paper reaches three broad conclusions currency. One method used by Pinstrup- First, both income and price elasticities o Andersen, Londofio, and Hoover is to esti- as maedmn untosfrsecfccmo-gc demand for calories are below unity and tend mate demand functions for specific commod- to cluster around 0.60 for the poorer consum- ities by income class using the Frisch method. st. ers They are much lower for higher income By assuming want independence among com- eroups. Sey aremeven a mfoderate increase in modities, a matrix of direct price and cross groups. Second, eeamoeaeicasinelasticities is derived using estimated income( calorie prices implies a large nutritional sac- elasticities of dend udgetprotins,m( rifice for the poor if present income growth elasticities of dematd, budget proportionse (2 and distribution trends continue. Only a con- and a coefficient for the flexibility of money siderable acceleration of economic growth This demand matrix is then converted to a matrix of direct and cross-price elasticities for (3 trends would permit calorie needs for the en- .. ,. . .. .nutrients and used to assess the nutritional tire population to be satisfied within the next fifteen years. Third, if moderate redistribution status of low income groups under various income and pr-icing policies. The drawbacks of In policies permit a substantial portion of the in- iscmethod are the extesive data reuie come increase to be allocated to the poor, this method are the extensive data require- . ments, the crucial role of the separability as- co sumption, and the need to use an estimated co Odin K. Knudlen and Pasquale L. Sc3ndizzo are ;iatf members of coefficient of money flexibility. co the World Bank. WaQington. D.C. ea The authors are grateful to two anonymous refere. . for helpful A second method by Timmer and Alderman comments and ug'ge-hionN on an earlier draft. The World Bank uses Indonesian household data to estimate, should not be held responsible for views expressed in this paper. for four income groups, calorie elasticities tio YIL Copyright 1982 American Agrictltural Economics Association ~~q~**'E7i*7- < .~-~ Kn uclh COan td Scandizzo DeLn(ind ftr Calories 81 I with respect to income and rice, cassava, and (4) L = U[f,(q * * . qe, )q . . . qft(q) o cor-n priices. Because calories and quantities , - are directly linked, this approach corresponds -jA( -pq Y to analysis of commodity demand functions by /=J income class, with the composite commodity Differentiating with respect to qj(f = 1, 2, ... , estimates being redundant. Although this n) and equating to 0 yields technique is preferable to the Frisch method, there are problems with cross elasticities be- (5) 3 (aU/ax ) (aJj/aq ) - = 0, tween many commodities and other such in- - teractions. At also requires extensive data, often unavailable in developing countries. where j = 1, 2 . . n. Indicating with pi = The third method, pioneered primarily by (aU/ax1)/X, the shadow price for the ith Reutlinger and Selowsky, uses a less elaborate characteristic, and with aij = fi ,the margi- scheme in wlhich relationships between a nu- 'aq3 tritional characteristic, calories, and income nal "yield" of goodj in terms of characteristic levels are made. Using this characteristic de- i, we can rewrite expression (5) as mand function along with income distribution m data, the nutritional status of the poorest in- (6) p = aijp, come group can be estimated and projected. i= Demand functions for individua,l commodities where j = 1, 2 ... n. Expression (6) states that need not be estimated. This paper uses the the market price of each commodity is equal to 10ot rule characteristic demand function method but a weighted linear combination of the prices is case, differs from the Reutlinger and Selowsky ap- of its characteristics, the weights being the tion in- proach by using household data instead of amounts of each characteristic provided by :ritional averagecountrywide data and by estimatlng a one additional unit of the commodity. characteristic price elasticity. This allows At a level of income Y and prices p, let a price effects to be added to the demand esti- commodity bundle of food (ql, q2, * *, qj) be mates and predictions. purchased containing only characteristics i= Characteristic demand theory as presented 1 to I. Then from equation (6) total food ex- by Lancaster assumes that products are con- penditure E(Y,p) will be sumed because of the utility derived from their J J I econo- characteristics or properties. For food these (7) E(Y,p) = 3 pjq = Piq3 - -specific would include nutrient content, texture, color, j== 1=1 s have taste, etc. Following this approach, the utility instrup- of a representative consumer can be expressed Because of the technical relationship between J to esti- as a function of the characteristics of the characteristics and commodities,. )mmod- goods consumed while the budget constraint is (c)haracterstcs and commodatses xi lajjqj, nethod. stated in terms of commodities: dg com- (8) E(Y,p)= xiPi income (1) max U = U (x, * x, income ortions, (2) subject to xi =fi(q, . * * q.); Expressing expenditures in terms of the money. i = 1, 2 . . . m, and characteristic "energy" measured in kilo- ed to a n calories per day (indicated for simplicity as ities for (3) pjqj - Y = 0. "calories"), (8) can be rewritten as tritional l I various 3p pixi acks of In ()-(3), U is the utility indicator: xi(i = 1, 2, (9) P(Y, P) = Y,) + c require- . .. ., ), the ith characteristic; qj the quantity Xc(Y,P) x= ility as- consumed; and pj the market price of the jth where P(Y,p) denotes the average expenditure .timated commodity; Y represents the consumers' in- per calorie; pc, the implicit price of calories; come; andJ', the functional relation that links and x,(Y,p), the amount of calories consumed. derman each characteristic to the i goods. Expanding P(Y,p) with a truncated Taylor itimate, Forming the Lagrangean for the maxiimiza- series yields the linear approximation, sticities tion in (l)-(3), after substituting (2) into (1) yields (10) P(Y,() = P(Y0,p) + a(Y - Y0), 1', 82 Februarv 1982 Atf'r. J. Agr. Econ. where P(Y0,p) = lim P(Y,p), YO min Y, and ture groups using dlata from household con- P(y Y) y- sumption surveys (expenditure is used as a a y'p evaluated at YO- proxy for disposable income). For each coun- ay try in the study, the average quantity of the As income goes toward its minimum level, various food items consumed by households in we can expect food expenditure to be devoted each expenditure group is converted to more and more lo procuring the most funda- calorie levels using FAO calorie consumption mental nutrient, calories. Therefore, we can tables. Second, we estimate a weighted re- write gression relating the average expenditure per .J calorie to per capita total expenditure. We > E Pjqj then use the sum of the constant term and the (11) P(Y0,p) = lim Pco, residuals as an estimate of the basic price of y- , XC calories. Third, we estimate a series of func- - e . d tions relating calorie intake to total expendi- where Pco denotes the pi-ice of calories in what ture levels and to the basic price estimate. We can be termed a basic bundle. Substituting e . * * use weighted least squares because sample -.~ (11) into (10) and the resulting expression Into sizes are different in each expenditure class. (9), we finally' obtain The ratios between sample and population (12) P(Y,P) = Pco + a(Y - Yo). sizes (sample "shares") are used as weights. Because our information consists of income In practice, consumers will face the same class data for six different countries, all re- basic calorie price except for a random factor gressions are also subject to the test that (a) reflecting differences in information, location, the coefficient of any variable and (b) the and other factors. Therefore, assuming that Yo value of the constant are significantlY different is sufficiently small, we can rewrite (12) for from one country to the other. Furthermore, the /.th consumer as follows: we test for the effect of five country-specific (13) P(Yk,Pck) Pck + aYk characteristics: food production per capita, =;- Pco ± ak + '1k, urbanization rate, literacy rate, population size, and adult population to total population where Yk indicates the income of the kth ratio sample unit, Pck iS the basic calorie price, and Table 1 presents the estimates for the aver- Ilk is a random disturbance having mean zero age basic prices of calories for each country, and constant variance. computed according to expression (14) and to Equation (13) says that the average expen- the estimated cross-country regression be- diture per calorie can be decomposed into tween average per capita expenditure per three parts: (a) a basic calorie price equal for 1,000 calories (converted in U.S. currency at all consumers, (b) a term depending on the parity exchange rates) and per capita total ex- consumer's income, and (c) a random ele- penditure. The basic prices vary from 70 to ment. Jndicating with hats appropriate esti- 24¢ per 1,000 calories at the parity exchange mates of equation (13), we can conclude that rate and from roughly 3¢ to 12¢ at official (14) p,(k) =P( - Pca + Uk rates. In both cases there is no apparent corre- lation between the estimated basic prices and is an estimate of the basic price of calories the selected socioeconomic characteristics, facing the kth consumer. This estimate can be (i.e., food production per capita, urbanization obtained as the residual of the regression rate, literacy rate, population size and adult I through the origin of If, against P(Y,,), In turn, ratio), this estimate can be used to evaluate the re- Table 2 presents selected estimates of . .- sponsiveness of calorie consulmption to varia- calorie diemand functionls based on the cross- tions in its own price. section sample. The functional form is -1'Ml.;semilogarithmic, a useful form because of the >- ; 4implied3 inverse relationship between absolute Esuiviiates of Calorie Demand Ftnctions elasticities, income, and prices. For the esti- 4alo..imated functions, the basic p-ice indices esti- The method for estimating the calorie demanld mated according to equation (14) are used as function is oummnrized in three steps. First, price ariables. The per capita e fpenditures wTe cte corie consumtption by expendi- (in national currency) are used as income - r A - 's :'e . p r Tce L'~ ..- wecluat ao-i osmtinb xpni inntoa Crec) r sd sicm Econ. I j~nand scandizzo Dcmnuud fior Calorie's 83 d conl Table 1. Estimated Average Basic Prices of 1,000 Calories d as a coun- In Domestic Parity Official of the Country Currency Currencya Exc[ ange Rate' Exchange Rate, olds to Bangladesh (Takas) 0,48 11.9 6.0 Indio (RupLc2S) 0.54 18.9 6.8 nptionl I ndonesia (Rupiah) rates2 I2 (R up3.+.8ee +0.2s).D, 0.229 3.£4 l t. , 6.9 ~-. rependi-Lnk (20.6)e (68)0(-.1) (-4.5 5 5.6 e. We ampl thereUnits: 95 P Dms i s curerncapit00a ofiodexpniueis...l00clre;1 sepnitr/aiamnh(S tprt xhne.l ~iceofhnts: vriblS. Se00aoietparat ifntion ecarge ralesoa esti-aedb duction ietn , epainingummresida.rs nr Ucome:mated for,00 peloplesa beflo ecande raboes The dsiffter eqaincues.o hbveetmtsi ta ()exediue6evljutsufcintt prhaetal 2 and .5 the7 elsiiissow ntbe3 wfrethegression.9; s pl)er inpitablodexen2dhowthat the.. , calrieasesY in epnincomecatiaothe po Ser pty li excvaries eciplc satan: and the picedmm coefiens for Bangladesh, ,i ooc,D,min aimumii of 0.iso Sri Lanka.Urtor Tis iarndicates. ilationnicatydfeetfoeahohrFute-teFGcorerqieetwlinrae. dainmore, facors pother thanw incoe andv pries caoifeeintaes bten. nd6 ovr allppartoafec caloripvrtlie whconisumptione acos 10% Conider inncoe,sifn the clori pric is-' con- aver-n counries.sTisns 3 indicatled by thew siniiant tecastant n Moreover, t thecunres wiveth thne mostes intr,cdmmvaibe coefficie ntsfo oangltehe intercPtk fo malntiion-Bangladesh,r ndoia, ccIndonesa- andfi totand byd the siniicancefoficet per capitadfood pro-all sow .I conidral hrigher.hi income teas- enpercabtley2 Caloerien Deomeand Funtions acrossr Cutry FAOd EilopendtreqClasses-etwl.ices Jcyiat iltaon ex- atosohr hnicoen pindos caloie-ntk Srivee 2Indad6c o ee apea o fetolre osu p ioniacrs 1esi rstan inrincomeLinkh: clre rc In is cnO: havger cepndntVriabesThsi idcondb h sgiiatstant L(ID, D i MD,rov , th0cutre wih0hms dum)fialfoffceticihaneretmlntiinBagaeh,IdalIdnsa Ountryaorc ntk corde-bI thlo.ce sinii cance. 5,10. 664.9pit 362.7 217.5 -90. sh.o- 82v co sie72.5 ig e incom7 e e .as5. andand riablse2 2.al oriea~a,a -5,21n . 698,7on 139.9s 183.8 ry a .d Exedtr 3 C.lasses41 05. 'adl t 152e12x(.) (.1 t3-t -12 indo- Paki2Sri I(2.1 Ovlow calorie intake 4ore . CAlries'c apita.d-y -6,7110.3 7266 28644.9 327 17 5 679.0 429.4 3. - 1.49257 1.077 7. 34) 4. (6o Blw R ~.6 10.9) 12.0) (10.8) t41 I.9t 3-3.4 12.3) 1051) 2rss . Calories/capitaiday -6,262.0 6968.9 3. 101.3 4)5.7-90 5 -919.026 1.276,5 6.risticst:ay-,642 89. 8.4 t26 3. -7. -3. (f9he elo CR12 .0) (2.7) (3) 0'4 21 2 I - .0 lOS I- 61 zatio 7. Caduriesc.apita, d, -183,412.7 6534.4 293.5 345.1 33.3 453.) -17it -77 850 1.7351] 927~.9 e tAb eCR(52.1 (1.01 (1.3) (0.1) II 01 I., ) -2 25i 12.34 t.4) Abeove CR I 126 tturs of4 taortes c.spitztday 46041.7 470.3 0. 5. 6367 397 79-22.49.1 .179 3 comeAowe CR (3.7) (109) 2,O )'5 1- 1.1 .4t 23 O5 ,~ro s- . Caoris'caita ay - ,26.0 98.9101. 487 905 -919. 1.71. K7 Belv! R (8.2 (44) 7.) 1 3Z8 -1.r i 2. m is.~ &- .f~- . .-~ '.*~ 84 Februiarv 1982 Ane'r. J. Agr. Econ. Kli Table 2. (Continued) T Paki- Sri Food stan Morocco Lanka Indo- Paki- Sri Produc- U,bani- No. of LN(IP)P LN(II-') LN(1P)- India nesia stan Morocco Lanka tion zation Obser- Dependent Variable D, Da D, D, D, D, D, D, /Capita Rate RI vations - Overall caloric intake 1. Calories/capita/day 898.7 -509.0 1,868.4 1,901.5 2,512.8 3,823.9 -3,768.6 7,528.7 0.99 60 - (2.4) (-0.9) (1.7) (0.5) (2.3) (3.2) (-2.3) (2,3) Ba 2 Calories/capita/day 931.2 2,544.0 3,564.0 -2,456.0 2,285.4 0.99 60 In( - (332) (2.7) (3.5) (3.4) (2 6) In 3. Calories/capita/day 514.9 822.4 -1,370.2 84.1 -7,470.2 0.98 60 M (1.15) (1.3) (-2.2) (2.1) (-0,9) M Low calorie intake Pa -below CR Sri 4. Calories;capita. day -58.3 530.0 4,619.7 -631.0 -3,152.2 -59.2 0.99 23 Below CR (-0.1) 10. 81 (3.2) (-0.7) (-1.0) (-0. 1) 5, Caloriesicapita/day 4,273.3 0.99 23 Below CR (2.5) 6. Calories/capita/day -7.0 -2,725.5 0.99 23 bu Below CR (-.6) (-1.9)al High calorie intake --above CR wi 7. Calories/capita/day 592.4 -961.1 1,568.7 -232 995.1 2,064.4 -4,372.1 5,760.7 0.99 37 Above CR (0.9) (1. 1) (1.02) (0.0) (0.3) (0.6) (-lI (1) 8. Calories'capitaiday -2,080.4 -4,120.4 0.98 37 Above CR (-2.6) (-2.2) hrn 9. Calories/capita/day -1,868.1 -4,041.4 0,63 1,077.2 0.98 37 Above CR (-2.2) (-2.21 (.01) (1.1) *Note: Whete LN(Y is logarithmn or expenditures per capita conversed to U.S. 5 at parity exchange rates: LN(/PP,. (ogarithrnn of~ the implicit price of' calories T consverted to U.S. S at parity exchange rates: D,, dummy variable for India-, D., dummy variable for Indonesia-, D,,, dummy variabie for Pakistan-, D,, dummy Pii variable for Mvorocco: D,, dummy variable for Sri Lanka: CR, FAO caloric reottirement: and t, statistics in parentheses.es re] ticities and, hence, higher calorie response as increases. A 10% increase in price would in(~ income increases. The poorest 25% in these cause consumption to fall about 1%', while a uti countries have calorie income elasticities be- 10% increase in income would cause con- gri tween 0.61 and 0.74, indicating that a 10% rise sumption to raise by amounts varying between all in income will increase calorie intakes by 6%o 1.6% and 3.3%. Therefore, consumers with a ob to 7% ortepoetg u. calorie intake above FAQ requirements ex- ha The calorie responses to price increase also hibit price elasticities of demand that are low ag show a similar pattern. The poorest 25% of the with respect to a calorie price relevant to the on, population from all countries show an average lowest basis of consumption. This is to be siz 4.5%9o 88 ali osm to nrsos expected as the consumption bundle for these Ta to an increase of 10% in the calorie price. As consumers is comiposed of different and higher tw for the top 25i"l of the income distribution, it quality types of f,ood than for the lower income su. appears that price responsiveness decreases consumers. The fact that the elasticity is not more than income responsiveness as income zero, however, suggests that the low income Su! tio .2' Table 3. Calorie Income and Price Elasticities Ta Calorie Income Elasticity Calorie Price Elasticity At the At the Povertya Lowest Highest Povertya Lowest Highest Country Line Quartilet' Quartilec Line Quartileb Quar-tileeBa Bangladesh 0.35 0.67 0.17 -0,5 1 -0.63 -0.11I India 0.44 0,61 0.16 -0.54 -0.88"1 -0.09 Indonesia 0.39 0.74 0.28 -O,5i -0.6d-0.9at Morocco 0.606 .3-.5-0.60 --0.07 Ind Pakistan 0.34 0.53 0.38 -0.45' -0.48 --0.107 In SiLna0.1 8 0. 17a 0.17 -0.51 -0.45 -0.11 M aComputed from regression 2, (able 2. Sri bComputed from regres,~io n 5, table 2. Consputed from regression 8, table 2. Not Computed from regression 6, table 2. P -u- Econ. Knud-lst'l bt/llI ,SCa,n,dzzo Deia,tel for Calotries 85 Table 4. Distributioni of Calorie Intake .. ...... - -tV ,. No. of FAO/WHO- Percentage ofPopulation Consuming below: Obser- Recommended . Per Capita 2,400 2,200 2n000 1,800 1.600 1.400 Calorie Intakea Calories Calories Ca-loties Calories Calories Calories 60 Bangladesh 2,020 91 ,.75 55 35 i 8 60 India 1,910 60 48 35 22 12 5 i 60 Indonesia 1,920 74 64 53 41 2 6 12 Morocco 2,276 48 39 30 21 13 7 Pakistan 2,050 97 90 44 2 0 0 Sri Lanka 2,000 84 41 7 0 0 0 23 Adjusted to account for individual variability. 23 9 23 bundle (the "basic" bundle) is also consurmed, tween food (calorie) availability to the house- although in different proportions and jointly tholds and their calorie requirements (table 5). with other goods, by the higher income group. This measures focuses on the overall nutri- tional deficit of a country and can be related to X 37 (a) tihe size of its undernourished population, X 37 Implications of the Estimates (b) market demand, and (c) the amount of total food needs (i.e., market demnand pius the of caloies These estimates can serve wo interrel.ated nutrition gap). ,dummy purposes. First, they can be Msed to derive Given these estimates, wve can investigate estimates of the calorie intake distnibution and the prospects of alleviating malnutrition by C related measures of malnutrition e; rectly from closing much of the nutritional gap by 1995. would income distribution data. Seconid, iey can be We conduct this analysis under four scenarios: vhile a utilized to explore the ability fot iincome (a) constant income distri'oution and constant con- growth. redistribution and price adjustment to calorie prices- (b) "optimal" income distribu- Atween alleviate underconsumption. The estimates tion (defined belowv) and constant calorie with a obtainable from income distribution statistics prices. (c) constant income distribution and its ex- have two basic components: (a) the percent- calorie prices increasing at 1% per annum, re low age of households consuming below the rec- and (d) "optimum" income distribution and to the ommended nutritional standard and (b) the calorie prices increasing at 1% per annum. to be size of the nutrient deficit by income group. Under the "optimal" income redistribLution r these Table 4 contains estimates of the first of these scenario, the income of each group with aver- higher two components for the case of calorie con- age calorie consumption below the recomi- ncome sumption. mended intake is alloxved to growv at the rate is not A second, possibly more significant, mea- necessarv to close the nutritional gap. The per- income sure of the extent of malnutrition is the nutri- capita inconmes of the other groups are as- tion gap, which is the aggregate difference be- sumed to grow%, at 1% per annLum. The resulting Table 5. Nutrition Gaps (Grain Equivalents) _ Proportion of Nutritional Gap to: ighest Average Per artilee Capita Gap Miarklet T,;al Total Gap Per N'earr' Demand Food Needs 0.11 ~ -0.09 (millions metric tons) (kg)--------------(S) -0.09 Bangladesh (1974) 2.65 46 16.9 14.4 - 4 0.07 India (1974) 9.43 39 6.8 6.3 -0.10 Indonesia (1970) 3.75 47 13.2 11.7 -0.11 Pakistan (1971) 1.56 66 11.5 10.1 Morocco (1971) 0.50 26 12.4 11.0 Sri Lanka (1970) 0.06 16 2.1 2.1 Note: Grain equivalents are converted at 3.5 million calories per metric ton. a Population below the recommended calorie intake. 8tt-'' - - *-----:-h 86 Februarv 1982 Amer. J. Agr. Econl. Table 6. Per Capita Income Growth Rates Necessary to Close the Nutritional Gap by 1995 with Rising Calorie Prices Average Growth Rates Required to Close the Nutritional Gapa Constant Income Optimal Distribution Redistribution" Historical Growth Rate Constant Calorie Price Constant Calorie Price Country 1960-76 Prices Kicrease 1%9 Prices Increase 1% Bangladesh -0.4 3.9 4.4 1.24 1.55 India 1.3 2.7 4.3 1.09 1.27 Indonesia 3.4 3.2 4.1 1.09 1.23 Morocco 2.1 4.0 5.0 1.20 1.36 Pakistan 3.1 1.6 1.6 1.00 1.15 Sri Lanka 2.0 1.00 1.00 Note; Based upon regression 4 in table 5 except for Sri Lanka, which used regression 8 and India, which used regression 5. a These are the growth rates necessary to bring the mean consumption of the bottom 10% of the populations up to the FAO/WHO recommended calories intake, adjusted for individual variability. Those groups with consumption below the adjusted FAO/WHO recommended caloric intake receive an income growth rate necessary !L , .to close their'nutritronal gap; those consuming above the recommended intake are giVen a 1% growth rate. average growth rates in per capita income can of the population if the bulk of the additional then be interpreted as minimum grow'th rates income growth is channeled to the poor. necessary to close the gap assuming a reason- This conclusion, however, also holds in re- able redistribution of incremental income. verse. If the poor's participation in economic 'As shown in table 6, the per capita growth growth is less than the rest of the population, * - rates necessary to close the nutritional gap their nutritional status is likely to suffer pro- - without a change in income distribution vary portionally more. Thus, policies are needed to from 1.6%k to 5% with constant calorie prices. ensure (a) that the undernourished are not These rates appear rather high and unlikely to excluded from the general improvement in liv- be achieved, especially for Bangladesh and ing standards, and (b) that they are specifically India and when calorie prices are permitted to helped to achieve minimum consumption increase. standards if the process of growth slackens or * If we assume an "optimal" redistribution of if income distribution deteriorates. growth, however, the necessaiy growth rates [Received February 1980; revjisionj accepted decline to about 1.0-1.6% per annum, even Augutst 1981.] with sizable price increase. Thus, if income growth were focused on the poor, malnutrition could be eliminated with modest aggregate . growth even with a concomitant rise in calorie w n t a Kravis, Irving, Alan Heston, and Robert Summers. "Real prices. This growth would be well within his- GDP Estimates for More than One Hundred Coun- .torical rates. tries." Econ. J. 88(1978):215-42. Lancaster, Kelvin. Consitnier Demtand: A Newt, Ap- proach. New York: Columbia University Press, 1971. Pinstrup-Andersen, Per, Norha Ruiz de Londono, and Conclusion0s Edward Hoover. "The Impact of Increasing Food Supply on Human Nutrition: Implic.tions for Com- This paper has focused on the determinants of mdy Pirtes in Agricur Research and-P5 l vcalorie intakes for an important sample of de- iy"Ae.J g.Lo.6(98:0-5 Reutlinger, Shlomo, and Marcelo Selowsky. Malnutrition veloping countries. Using aggregate data from andcl Poverty. Washington, D.C.: World Bank Occ. household budget surveys, the paper has Pap. No. 23, 1976. shown that neither foreseeable price increases Timmer, Peter C., and Harold Alderman. "Estimaling nor a slackening in economic growth would Consumption Parame[ers for Food Policy Analysis." hamper improvement in the nutritional status Amer. J. Agr. Econ. 61(1979):982-7. -RU . V. ,