World Bank Reprint Series: Number 176 Marcelo Selowsky Income Distribution, Basic Needs, and Trade-offs with Growth: The Case of Semi-industrialized Latin American Countries Reprinted with permission from World Developmzent, vol. 9 (1981), pp. 73-92. IVorld Depelopmenit, Vol. 9, pp. 73-92 0305-750X/81/0101- 0073 $02.00/0 Pergamon Press Ltd. 1981. Printed in Great Britain Income Distribution, Basic Needs and Trade-Offs with Growth: The Case of Semi-Industrialized Latin Amnerican Countries MARCELO SELOWSKY* The World Batik, Washington, D.C. Summary. - This paper addresses income distribution issues and policy options to eliminate ex- treme poverty in a particular typology of middle-incomiie semi-indtustrialized developing coun- tries of Latin America, The countries included in the typology are characterized by a relatively higih per capita income (above US$700/yr in 1977), a relatively highi degree of industrialization (industry representing over 30% of GNP), a rather large size both in population and area, and relatively well endowed resources in agriculture. A distinctive feature is a strong inequality in the distribution of income and wealth relative to other countries of similar levels of per capita income. The countries included are Brazil, Colonmbia, Ecuador, Mexico and Peru, representing two-thirds of the population of that region. 1. POVERTY AND INEQUALITY the poverty group ranges from 58'% for Colom- bia to 82.7% for Peru, Brazil's figure falls This group of semi-industrialized developing between 64.3 and 72.9%, whereas the Mexico countries has experienced an impressive growth figure is approx. 75%. In several countries, during the last decades, GNP per capita reaching rural poverty is associated with particular 3.8% per annum. Mean per capita income regions; in the case of Brazil half the families reached $1 168 in 1977 (Table 1), mean life ex- in the poverty group are located in the rural pectancy reached 60yr and mean calorie con- Northeast, and in the case of Peru, in the sumption has become equal to the per capita Mancha India. calorie requirements. These figures exaggerate the participation What is striking is the coexistence of these of rural households because the reported high means with an extremely unequal distri- income figures must be corrected by different bution of these welfare indicators across the price levels, particularly those of food. In the population, i.e. the problem of inequality. In spite of high mean levels of per capita income, * almost 28%7 of the population suffers from a Thlis papecr is acondensed version of 'Balancing trickle a t 2. odown and basic needs strategies', World Bank Staff caloric deficit. Because poorer families have Working Plaper No. 335 (1979). It was prepared as a more children, 36% of all children suffer these backgrouind stuidy for the World Bank's Wiorld Devel- caloric deficits (Table 2). opmelnt Report (1979). That report addressed the The same can be found for welfare indica- structural transformations of middle-incomc devel- tors other than income. In Brazil, differences in oping countries and policies to pursuie the twin objec- life expectancy of 21 yr arc found between the tives of growtlh and poverty alleviation. poorest groups in the rural Northeast and the Several people provided valuable comments dtring richest groups in urban areas. In Mexico, infant the preparation of this paper. Sebastian linicra comn- *ortality per 1000 live births ranges between mented on an earlier draft and provided important 392 or lthe perichest1 s rtat ranges and109forthee suggestionis. Conversations with P. Isenlinan, B. B. King, 39p2 for the richest states and 105.9 for the P. Knighlt, C. Lluch, T. N. Srinivasan, L. Squire, poorest. In Fetador it ranges fromn 45.5 to M. UrrLltia, J. Villarzti, R. Webb and other colleagues 112.1. in the Bank are gtracfulll acknowledged. Of course, Most surveys indicate that when households none of tlhemn is responsible for remaining deficiencies. are classified by per capita household income, Tlh: views and interprelations in this paper are those most of the poorest 20-40% of families live in of the autlhor and should not be attributed to the rural areas. The share of rural households in World Bank. 73 74 WORLD DEVELOI'PIMENT Table 1. Per capita income and growt/) rates Growth rates (1960-1977) Population Per capita GNP 1977 income, 1977 GNP Population per capita Country (millions) (dollars) (8X¢) (M) (e) Brazil 116.1 1360 7,8 2.9 4.9 Colombia 24.6 720 5.5 2.8 2.7 Ecuador 7.3 790 6.1 3.0 3.1 Mexico 63.3 1120 6.1 3.3 2.8 Peru 16.4 840 5.2 2.9 2.3 Average (weiglhted) 1168 6.8 3.0 3.8 Source: World Development Report 11 (IBRf), 1979). Table 2. PercLntage of indil iduials with caloric dleficit, 1976 Aggregate income gap Country Population Children (%7) Brazil 24.0 32.4 2.2 Mexico 24.0 32.4 2.3 Ecuador 40.0 48.8 8.9 Peru 35.0 42.7 4.1 Colombia 26.5 35.8 3.6 Average 27.75 36.0 3.0 Source: M. Selowsky, 'Balancing trickle clown and basic needs strategies', World Bank Working Paper No. 335 (Junle 1979). case of Peru, such adjustments lowered the per- subsistence farmers of distinct ethnic back- centage of rural households in the poverty ground. 'Subsistence farnmers are mostly Indian, group to 59%,b.' The use of differential poverty Quechua or Aymara speaking, and in 1961 lines provided by ECLA for urban and rural about 70% were illiterate. Their principal source areas show that approx. 60% of households in of livelihood was on the average about 0.9 ha the poverty group are rural.2 of Sierra cropland, three heads of cattle, and How are these households linked to the some other livestock. Most earned some cash urban and rural economies? What are their income by seasonal labor on larger farms or sources of income? These are important ques- occasional sales of livestock products.'3 In tions for identifying points of policy inter- Colombia, approximately half the rural house- vention. Ideally, one would like to classify holds in poverty are small producers, the rest rural households according to the relative landless labour. Nevertheless, even small pro- importance of their share of income from ducers received at least two-thirds of their ownership of land, sharecropping and the pro- earnings as wage labour in larger farms.4 vision of wage labour to other farmers. Because In contrast to Peru, where most rural poor the typical population census or household sur- are subsistence farmers, the evidence for Brazil vey from which these poverty groups are con- shows that most of the poor are landless labour. structed usually does not provide this data, According to the SUDENFi--1BRD survey in sectoral (i.e. agricultural census or urban Northeast Brazil, only 17 mlillion of the 6 employment) surveys must be used to derive million rural labour force (heads of house- this information. holds plus working family members of land- Scattered evidence tends to show that the owners) own lanid. Of this 6 million, 'about mixture of sources of earnings for the rural three million have no access to land, and sub- poor varies considerably in the countries sist on temporary employnment or scrounge explored. In Peru, 80% of the rural poor are out a living on lands so poor or remote that INCOME DISTRIBUTION, BASIC NEEDS AND TRADE-OFFS WIThI GROWTH 75 they were not captured by any of the censuses, line. This income gap amounted to 8% in 1960, cadastral survey or the SUDENE-IBRD survey. 4.5 o in 1970 and 3.0%.. in 1976 (see Table 2)7 These three million and their families constitute The above figures clearly show that in these the hard core of poverty in the Northeast. Most countries poverty is, and will increasingly be, a landowners (even if they own only two or three purely distributive problem rather than the hectares of poor land), sharecroppers, renters result of a lack of aggregate resources. Elimin- and permanent workers, realize real incomes at ating extreme poverty while maintaining a high or above the absolute poverty line, but the growth ra^e should be feasible. It is the option remaining persons presumably do not, and this explored in this paper. situation is easily traceable to their lack of ac- cess to land.'s5.AESO OIC NrRETO This evidence shows that rural poverty 2. AREAS OF POLICY INTFRVFNTION cannot be characterized by a single, simple Areas of intervention for poverty alleviation occupational classification. Small farmers do are grouiped here into three categories. The first not seem to represent the core of rural poverty, includes the elimination of distortions that are as is usually believed. Landless labour is at adverse to a stronger trickle-down effect of least equally represented in the poverty grotup, growth; its objective is toicreasethe employ- and wage earnings represent an important com- ment conitent of rowto The second area ponent of income, even in the case of sub- refers to investment policies to increase the sistence farmers. T'his has important policy im- .endo ment oflhuman an p incacapitao plictios, a wil b see laer,endowmient of human an(i physical capital of plications, as will be seen later, the poor. Hlere we are onaly interested in high- Two patterns seem to emerge from the data thelding Herestents, namly those highr on the urban poor: most poor are located in yielding investments, namiely, those that are non-metropolitan areas, particularly in small good for the poor but also for the growth rate. towns;abouthalfof theas, particulassociatd s l Both sets of policies have two features. First, towns; about half of them are associated with they incrrease the welfare of thle poor(z by in- commerce and the service sector. Depcn(ding on they inc the earefile poo, by io the country, between half and two-thirds of creasing their earning capacity. Second, they do lieas o bouehods i th poo fimilis ae not imipose a t rade-oft between dist rihlt ion and( eads of households in the poor familoes are growth or efficiency. On the contrary, they can salaried employees, the rest self-employed, benefit both equity andl growth. There is no single relationship between The third area is 'basic needs', poverty allevi- unOemployment status and households in ation must also be measured by welfare indica- poverty. Open unemployment does not seem to tors other than income. Increases in health characterize the households in the poverty status, life expectancy and civic participation group. A recent study at ECLA shows that dcpendL on the consumption of goods and ser- most heads of households in the poverty vices like food, health, housing and minimum group are employed, although alarge percentage edlucation. Analytically, it is convenient to of them are underemployed, working fewer separate this area from the second area of inter- than 39 hr a week, and are willing to work ad- vention: first, they might not represent a 'good' ditional hours.6 iinvestment and, second, their consumption The evidence indicates that relative income might not rapidly respond to the increase in the inequality has, at best, remained constant over income of poor families; in other words, the these long periods of sustained economic growth. first and second options do not automatically Although the fraction of people in poverty has guarantee the 'basic-needs' objective of poverty declined, the force of population growth has alleviation. resuilted in a constant absolute number of peo- ple in poverty. In Latin America, 100 million people live and will remain in absolute poverty (a) Policies withlout trile-o]fs.' eliminating during the next two decades. What is of interest distortions that are adv7erse to both emplov'- is that despite a constant number of poor, the mnenit and growtl economic capability of solving the probjem of poverty has increased substantially. (i) This section reviews policies to eliminate Let us define as the poverty line the per those distortions that not only have, by defini- capita income at which families purchase an tion, a negative effect on output, but also ad- amount of food yielding the required amount versely affect employment. Of special concern of calories. Therefore, we can define as the is the employmenit generation effect pf GNP aggregate income gap the fraction of GNP that, growth, because it is the basic mechanism by if transferred to households below that poverty which economic growth trickles down to the line, would lift these families to that income poverty groups. 76 WORLD DEVELOPMENT Although some distortions are the result employment is negative. of historical forces unrelated to public policy, On several occasions employment creation most have been policy-induced and, ironically, was the objective of tariff protection policies were initially conceived to improve income dis- to the industrial sector. The question arises tribution and employment growth.8 The central why government chose such policies. One ex- point here is that the distortions introduced planation is that the positive and negative ef- were not the result of the objective per se but fects on employment resulting from industrial of the specific policy instruments selected to protection are not symmnetric in terms of timing achieve it. Thus, the trade-off between these or visibility. New industrial activities emerge objectives and economic efficiency was the out- quickly when they are the result of tariff bar- come of choosing the wrong instruments rather riers; the barrier is simply the result of nego- than an inherent incompatibility between them. tiations with domestic investors or multi- Why were these instruments chosen in the national enterprises, i.e. the new investment first place? Why were they not replaced by has been 'latent'. The positive effects on em- instruments capable of achieving the desired ployment are immediate, highly visible, and redistributive effects without efficiency losses? concentrated in urban areas. Urban labour The answer is fiscal expediency. The trade-off usually has a stronger political militancy than between employment and distributive objec- rural labour. The negative effects of such tives, on the one hand, and economic efficiency policies usually occur over the long run and are on the other, resulted from having used fiscally less visible: the mechanism is the exchange rate, cheap instruments. The cheaper the instrument, which tends to increase less than it would the higher the inefficiency generated in the pro- otherwise have, and the effect is spread across a cess of achieving a redistributive goal. The main large number of productive units and regions (it thrust of this section is to show that, today, is less visible from a political point of view). these countries can afford 'expensive' policy Trade restrictions adversely affecting the instrzments and, therefore, avoid such trade- agricultural sector have also been undertakren to offs. keep consumer food prices below world prices. Two sets of distortions that have adversely When this is achieved by food import subsidies affected the employment generation effect of or prohibitions on food exports, prices to growth are explored: first, those that have domestic producers go down below world distorted the composition of output either by prices. A policy that lowers consumer prices trade policies or by internal price policies; and without discriminating against domestic pro- second, those that have increased the relative ducers involves a subsidy on total consunmption, cost of labour over and above the one corre- independent of the source of supply. Obviously, sponding to the relative abundance of labour in this entails a larger fiscal burden than that of the economy. alternative instrUrnents. (ii) Trade restrictions provide a different Unless consumption subsidies are used, premium to the export sector (when they gen- policies to keep the price of food down will erate one dollar's worth of exports) than to the discriminate against agriculture. The cheaper import-competing sector (when they substitute (fiscally) the instrument used, the less success- one dollar's worth of imports). The distortion ful it is in reaching the objective without is defined by the end result of the policy: the adversely affecting agricultural output and import-substituting sector expands relative to emiployrnent. the export sector and ends up by using more (iii) Most countries place ceilings on nominal resources per dollar substituted than those used interest rates that, utnder inflation, yield low or by the export sector per dollar generated negative real rates and strong excess demands through exports. If iniport restrictions are con- for credit. Thuls, credit becomes alloLated to centrated in particular iniport-comrpeting indus- pressure groups and to large entcrprises where tries (those requiring higher protection) the risk is minimal (interest rate dlillerentials are import-conmpeting sectors without (or with not allowed legally so as to compensate for dif- smnall) protection also tend to he discriminatled terential risks). Large enterprises in the indus- against. trial sector rvceiving that credit are perlhaps the Agricultuire, either through its txports or its less labhoir-intensivC acltivities in the econonmy. imiport-conipetinig activities witbout protection, I Indervalhed excliange rates anid imports of received most of the negative effects of suchI capital goods without imnp)rt duties are further policies. To the extent that its labour-intensity so-urces of a cheap use of capital, as are tax is greater than that of the highly protected credits and tax exemptions for investinp in industrial activities, the net effect on global hbackward repions, INCOME DISTRIBUTION, BASIC NEEDS AND TRADE-OFFS WIThI GROWTIH 77 (iv) Minimum wage laws, social security legis- The employment effects of extending labour lation, and laws governing the stability of em- legislation, minimum wages coupled with legal ployment have been typical interventions to social securit.y contributions by employers, can increase the real income of labour.9 Such poli- be significantly more important than the effects cies have been enforced in particular sectors of of that legislation in the present. Minimum the economy, in registered enterprises and wages will increasingly become binding and particularly in the urban economy. For some therefore cover a much larger fraction of the labour skills, especially the lower ones, this labour force; the joint effect of both minimum legislation has become binding and thereby in- wage pltus social secuLrity will involve a more creased their real wages. To the extent that the important increase in the cost of laboLur than demand for labour has some price elasticity, presently exists. If denmand elasticities are this increase has been achieved at the expense higher in these new sectors, because of higher of a lower employment in these sectors. labour-intensity and higher factor substitution, From a political point of view, the room for the potential negative employment effects are manoeuver in this field is limited. Of inmport- reinforced. ance is to avoid adverse employment effects as Table 3 shows some orders of magnitude by governments extend wage legislation in the which social security legislation increases the future to latter sectors of the economy, sulch as cost of labour. In some countries it reaches 20% agriculture and the informal urban sector, parti- of the wage. If the price elasticity of demand cularly services. The issue is how to increase the for the output as welt as the elasticity of factor coverage of these benefits without substantially substitution is 1 and the share of labour is 0.5, increasing the cost of labour to these other sec- an increase of 20', in the cost of labour will tors. diminish employment in those activities also by Present minimum wage legislation and social 20%>. security contributions cover only a small f'rac- To avoid such increases in the cost of labour, tion of the labour force, i.e. that in the modern minimum wage laws should be complemented urban sector. Because this sector uses more by subsidies (or tax exemptions) on the use of skilled labour and is subject to more labour labour, and social security financed by the cen- union pressures, only a fraction of their employ- tral budget instead of by the employer. This is ment is bound by the minimum wage law (most well known; the issue is selectivity so as to workers earn a higher wage): the basic effect of make this policy fiscally feasible. such legislation is to increase the cost of labour One solution is to use these instruments at through social security legislation, which sets a the margin and in sectors where labour de- percentage contribution as a fraction of the mands are price-elastic. Subsidies could be given wage. Extending legislation to more traditional on the basis of additional employment, relative sectors, such as agriculture and the informal to a base year or relative to a fraction (i.e. 0.90) urban sector, implies reaching activities of of the employmnent over, say, the last 2 yr. The higher labour-intensity, higher factor substi- same would apply to social security contri- tution, and employing labour of lower skills. butions; employers would be exempt only for Labour not covered by legislation has the additional employment, the contribution to be lowest wages and incomes; only 8.5% of wor- financed by the central budget. Sectors with kers covered by the social security system high labour-intensity and high capital-labour belong to the poorest quintile of households in substitution such as certain activities in agri- the case of Colombia. culture, construction and commerce are likely candidates for these policies. Table 3. Social security contributions The notion of policies at the margin can also as a per cent of the wage, 1970 (%) be applied to other types of legislation. Laws imposing high penalties on the dismissal of Country Range workers, for example, have become a typical Brazil 20.7 -2 .1 feature of wage legislation, even more so Colombia 10.8-18.3 than in developed countries. The result is that Ecuador 13,5-14.5 the hiring of labour has become a capital Mlexico 15.2-20.8 investment whose cost is governed by law. Peru 9.5- 13.0 These laws could be replaced by unemployment insurance schemes financed by the public bud- SOLurce: OAS, Guidelines for Ac/hie'ingi get, providing unemployment payments that Mlaxinmum F.mployment and Growth in gradually decline over time and only last for a latin A merica ( 1973). limited period. 78 WORLD DEVELOPMENT (v) In most of these countries, wage levels (vi) The literature shows a growing pessi- are strongly associated with the degree of mism about the possibility of the manufactur- 'modernity' of the sector of enmployment and ing sector absorbing the high growth rate of with the size of enterprises. Labour market labour supply in these countries. It is argued segmentation, the extent to which labour of that iiianutifactiring employment is still a equal skills receive very different salaries, is sm,. ! share of total employment (15 -S20%;), also associated with these characteristics. The and that its growth, therefore, cannot absorb empirical evidence shows that suchsegmentation imiiportant fractions of the total labour force. not only has not diminished, but has perhaps It is also argued that the rate of job creation in increased over time. These wage differentials these sectors is low for a given output growth, are partly explained by the different.al en- that is, employment ouitput elasticities are forcement of labour legislation described earlier lower than one in other words, employment and partly the result of labour contracts in grows more slowly than output. enterprises with strong labour unions or multi- The second argument, however, overlooks national firms paying higher wages than (lomes- the fact that job creation has taken place with tic firms. quite substantial increases in real wages. Clearly, The end result is that 'protected' or formal one cannot have both: given the high growth labour markets face an excess supply of appli- rates in real wages, the employment growth has cants, those failing to obtain employnent ending been substantial, particularly relative to other up in the 'informal' or free-entry sectors where historical experiences. the wage tends to clear the market, reflecting Table 4 sh1ows that, on average, emlploymnent supply- demand conditions. and real wages have been growing at 3.6 and What are the implications of this segmen- 3.7',, respectively. The question is what the tation in labour markets for income distribution in',rease in enmployinent would be if additional and the returns to unskilled lahour? Iwo studies effort were nmade to translate these increases in undertaken for Colombia shed some light on labour demand into additional employment this question. By eliminating wage differentials rather than into additional wage increases, and(i for labourers of equal edlucation in urban what policy instriuients shIould be emlployetd in Colonmllia, the real wdge of the most unskilled such elfort. The answers to these questions are category of labour was estimated to rise by relevant to the issue of increasing the trickle- 8=' .0 This is a lower bound as only factor sub- down effect of an increase in labour demand. stitution takes place in this exercise, If the increases in labour demand that led to When additional substitution and rural the conibinaticn of increases in wages and em- urban migration is introduced (that is, long- ployment observed in Table 4 are maintained in run adjustments are allowed), the results can the future, the following trade-off between be substantially stronger. By introducing sub- wage increases and employment increases will stitution in the final goods market, as well as occur.'2 If wages were kept at a 2--3%' growth the possibility of international trade, and by assuming that urban-rural wage differentials . in w, are maintained (which limits the degree of Growth In wages (%) Growth n employment (%) migration), the elimination of urban wage dif- 2 7 ferentials was estimated to increase the real 3 5 wage of unskilled labour, urban and rural, by 4 3 35'T( 1 - Table 4. Anntual growth rates in rinploynent and wvages in the industrial sector, 1963 +-1 972 (percentages) Country Employment Real wages Brazil 4.1 2.8 Colombia 4.2 2.7 Ecuador 5.8 2.5 Mexico 1.9 6.9 Peru 1.8 3.7 Untveighted average 3.6 3.7 Source: U.N. Growth of World Industry, reported in P. NI eller, 'Eiifoques sobre la demanda de trabajo: relevancia para America Latina', (CIEPLAN, June 1978). INCONOE DISTRIBUTION, BASIC NEEDS AND TRADE-OFFS WITH GROWTH 79 rate, the einployment growth rate in these sec- Correcting such underinvestment does not tors could reach 5-7%. introduce a trade-off between distribution and The figures in Table 4 are averages for wor- growth; on the contrary it has positive effects kers of all skills. As mentioned previously, evi- on both equity and growth. dence indicates that real wages for educated Why have not governments taken care of this workers have grown faster because of the short- underinvestment if they indeed were interested run inelasticity of the supply of educated man- in growtlh? Two reasons are possible: either few power. Expansion of the educational sector of these good investments remain, or the mar- (discussed in the next section) will increase this ket inmperfections facing the poor were a way of elasticity and assure that additional labour channelling credit and investment toward demand results in a wider employment effect higher-income groups. In other words. credit rather than higher rents of already employed rationings allowed income groups with access workers. For lower skills, wage increases co- to credit to undertake investments with returns exist with elastic labour supplies stemming below the opportunity cost of capital to the from the rest of the economy. The issue of economy. labour market segmentation discussed pre- The first -,sk, therefore, is to exhaust the viously is central to the explanation of this search for gcod investments in the poor. Only phenomenon. by first identifying and evaluating the contri- bution of these investments to their incomes can we then proceed to other interventions involving a higher degree of trade-off. The availability of 'good investments' (b) Increasing the earning capacity ofJ tihe poor depends on the particular characteristics of by augmenting their endowment of assets: the poor: the existence of such investrnents is identifying the good investments higlly country-specific, highly interdcpepident and highly dynamic; it depends on the age characteristics of the poor (tthere are fewer (i) Introductionz good investments to be made in the very old), One way of increasing the earning capacity their hunman capital potential (for example, the of poor households is to augment their owner- fraction of physically and mentally handi- ship of productive (human .and physical) capped in the poverty group), and their initial assets. This option is appealing because it ownership of physical capital (for example, the automatically takes care of the long run: first, fraction of artisans and small farmers or share- by increasing present ownership it increases croppers among the poor). Sonic empirical evi- the future flow of productive capacity; and dence helpfujl in identifying these investments is second, it is perceived as less paternalistic than discuissed later. policies that directly subsidize the consumption of the poor. A more explicit statement of the limitations (ii) Investment i1i lhumani capital and trade-offs of such an option is needed. For These investments include all interventions example, what are the limits of usinig this in- that, by influencing the characteristics of indi- strument for distributive purposes without a viduals, affect their productivity as labour substantial trale-off with growth objectives? inputs. Investment in education and on-the-job What are the criteria by whichi this trade-off training are interventions whose productivity can be judlged? Can an optimal level of inter- effects have been well studied. Interventions at vention in this area be id(entiried, that is, a preschool age amiong poor children, infant level of additional asset-creation anmong the nutrition, and other cnvirolinmental variables poor that does not adveisely affect growth? such as health and psychlological stimulation, Investmeniits in the poor that do not adversely have been analysed but not yet quLiantified for affect growvth are, by diefinition, those whose their economic impact.13 rate of return (valued at shadow prices, witllomt These investmiients are particularly important a distributive weight correction) is lhiglher than becauise half of the poor individuals arc children the opportunity cost of capital to the ecoiiomy. below age 15, whose future productive poten- Therefore, the magnitude of the underinvest- tial is influenced by early interventions: raising ment in the poor can be defined as that amount the income of households in the present does of high-yielding asset formation which does not not guarantee a sufficient investmenit in children inateriali7e because of the market imperfections on the part of their parents: in most societies facinig the poor, particularly in capital markets, the iipossibilitv of borrowing against huLmian- 80 WORLD DEVELOPMENT Table 5. Stuzdies on the effect of malnutrition on learning Deficit in Probable effect on Authors performance consequent learning Cravioto and De Licardie Auditory-visual Reading ability (Mexico) integration Cravioto et al. (Mexico) Visual-kinesthetic Writing and drawing integration abilities Champakam et al. (India) Visual identification Reading abilities Cravioto et al. (Mexico, Kinesthetic-visual, General learning Guatemala); Guthrie, kinesthetic-haptic, abilities et al. (Philippines) haptic-visual, and auditory-visual integration Source: J. Cravioto and E. de Licardie, The effect of malnutrition on the individual', in A. Berg et al. (eds.), Nutrition, Nationial Developmtient and plannin,g, (Cambridge, Mass.: MIT Press, 1973). capital formation (there is no collateral) in- later age (l -3 yr later) than higher-inconme stu- creases the possibility of underinvestment in dents, Educational programmes for young education by poor families. women concerniing childrearing practices would Growing empirical evidence shows that pre- provide another type of solution. Two kinds of school children of poor families have a lower eduLIcational programnine are important. First, performance in most ability tests than matching nutrition education, particularly that relevant controls from higher-income groups. Much of to infant-feeding and breastfceding practices. this difference can be attributed to the nutri- Growing evidence indicates that the decline in tional-health-psychological environment of breastfeeding practices in low-income families preschool age children. If environmental defi- of urban areas is a main determninant of infant cits affect the future productivity of the indi- nmalnutrition. The resource cost of substituting vidual as well as his future capability to benefit breast milk appears quite large; preliminary from schooling, the question arises whether estimates suggest that if 20'.' of the mothers countries are not indeed underinvesting in pre- in the urban areas of developing countries do schlool age interventions for poor children, not breast feed, the loss in breast milk is Table 5 presents a survey of studies whose spe- around $365 million/yr; if half of the other cific aim was to evaluate the effect of infant 80% discontinue breastfeeding after the sixth malnutrition on specific types of abilities in month, the total loss -eaches $780 million.14 young children that are crucial to basic learning. Second, education on childrearing practices They show the mechanism by which early mal- with particular emphasis on early stimulation is nutrition can affect the effectiveness of later another course of action. Implementing edu- schooling. cational programmes of this type re(uires some Policies to change the out-of-honme environ- prelimlinary research, wlhich has not taken place ment throuiglh widescale preschool compen- on a wide scale in developing couintries. Ques- satory progranimes are difficuilt to undertake in tions that arise are: Hlow different are child- the short run, unless such programmes are sim- rearing practices across families in developing ply extensions of existing elenmentary schooling coLintiies? What are the factors dletermilining that can draw children into kindergarten pro- these differences? Are they relatedl to incoIlie gramrnes I or 2 yr earlier. Experience in the or to particular ethnic group% of thle popll- United States shows that because extensions of lationi? the 'kindergarten type' are either insufficient in In spite of spectacular expansions in enrol- themselves or improperly designed, they cannot irients and in the supply of educated labour, the compensate for the environmental deprivation evidence suggests that lbasic education is still an suffered by low-income children. A more com- effective and prodLuctive way of increasing the plex programme is required. earnings of poor households. First, still today, A partial solution, at least in the short run, poor households are characteri7ed by extreniely might be a Lorrective policy geared to lower- low levels of schooling among their workinig income children who enter prinmary school at a members, as well as low levels of sclhool enirol- INCOME DISTRIBUTION, BASIC NEEDS AND TIRA)AD-OVFS WITII GROWTHI 81 ment among their children; data by incomiie lower boulnd for the capacity of education in groups show that 30'; of heada of household in explaining productivity differentials."7 Similar the poorest quintile of families in Colombia results are reported for several cities of Colom- have zero yr of schooling. Almnost 60% have only bia. Controlling for occupation, ecoiionmic sec- 2 yr orless. In the sameincomegroup, only 54t tor and social stratum, primary education in- of children age 6-1 1 attend sclhool.15 Cross- creases earnings by 40, 16.4 and 38.3%' in the section regional data for the other countries cities of Bogota, Barranquilla and Cahi.'8 show that an important number of provinces Basic edlucationi is undloubtedly an inmportant still have enrolnment rates under 60% for clhil- way of increasing the incomes of poor house- dren below age 10. holds. Wheother it is also a good investment Second, basic education has an important from the point of view of resource allocation net impact on the earnings of labour, even if depends on the economic costs of achieving in- unemployment is taken into account and only creases in earnings capacities, the rate of return possibilities of work in the informal sector are to investment in primary e(dtucatioll. Several considered; arguments against, based on pre- studies show that not only the rate of return on sumably high levels of utnemploymiient among primary education is higher than any conceivable educated labour and on the notion that edu- opportuLnity cost of capital (the mean rate of cation is simply a screening device (to allocate -return is 25.1'.t) but it is substantially higher 'scarce' jobs in the modern sector), do not seem than the return on other levels in schooling. 19 to be confirmed by the existing evidence. Clearly, these countries are underinivestinig in In the case of Colombia, adjusting the rates basic education. of return to education by the probability of Why do not poor liouiseliolds increase their unemployment lowered the rate of retuLrn from enrolment in basic education as a reaction to 33 to 28%;7.16 Althouglh unemployment lowers its high rate of return? Why do rich house- the return it still remains su'bstantiatlly high. holds still invest in higher education in spite of Controlling for other factors positively associ- its low rate of return? Although public primary ated with schooling and having an independent education is free, other costs of schooling -- effect on labour earnings, the effect of schooling basically foregone earnings - are important is still important. Table 6 presents a summary costs for poor families. Ilhcir imiiplicit discouLnt of results of nine Latin American countries; rate is also high, even in com>parison with the even controlling for occupation and type of highi rates of returni to primary education. activity (employmnent variables), years of Higher education is highly sutbsidized in most schooling explains between 7 and 25% of earn- countries, consequently the private rate of ings inequality. This is a strong test for edu- retuLrn faced by rich househoids is substantially cation: education also changes productivity by higher than the social. In Colombia the social inducing cross-occupational and cross-sectorial rate of return is 8%' while the private is 25%; mobility, an effect that is not captured by these not only rich lhoustsliolds are being subsidized estimates. These percentages are, therefore, a but this subsidy encourages investment in a Table 6. Latin A merica - contributioni of different characteristics to tite explanation of earning ineqzality: ranges encompassing all countrv results* (%) Personal variables Employmcnt variables Education Areas (years of schoolii,z) Age Occupation Activity Urban Employees 7-17 10-21 6-12 5 Self-employed 8-22 15-26 3-9 3-10 Rural Employees 9-13 10-18 3-6 3--7 Self-crnploycd 13-25 12-20 0-8 1-4 Source: Oscar Altimir and Sebastian Piriera, 'Decomposition analysis of the inequLality of earnings in Latin America' (ECLA-World Bank, August 1977). * Countries included are: Argentina, Brazil, Colombia, Costa Rica, Chile, Mexico, Panama, Peru and Venezuela. 82 WORLD DEVELOPMENT sector with an economic return barely inatching Two features of these countries limit the the opportunity cost of capital relevant to most effectiveness of such policies: location and countries. country size, and the problem of targeting with- The policy implications of this are that the out substantial leakages to high-income groups. incentives for poor farnilies to increase their Subsistence farmers are rarely concentrated enrolment rates must be increased. The con- so that regional policy packages could be straint is not the capacity of schools but the designed. Because in most countries subsistence demand for schooling by poor househlolds. farmers coexist with large farms, only specific Additional incentives to compensate for fore. programmes can be targeted (credit, teclinical gone earnings (particularly in rural areas where assistance); other prograrnmes such as roads or employment opportunities for children are irrigation would sLubsidize other farmers as well. higher) and transportation costs are easy to Unless 'non-target farmers' are charged for the implement; school breakfasts and lunches, programme, the financial cost per 'subsistelnce free transportation by school buses are some farmer reached' becomes prohibitive. examples. Such measures can be easily financed Because of their crop pattern, irrigationl is by a redistribution of subsidies within the edu- perhaps the main constraint to increase the cation sector, specifically by eliminating productivity of the small farmers. Their location subsidies to higher education. These measures (basically the high plateaus), however, makes will be analysed in detail in Section 2(c). investment in irrigation a costly undertaking and economies of scale resulting from co- (iii) Increasing the physical assets of the poor operative arrangements are prevented by in the rural areas: identifying the good invest- distance and geographical dispersion. The con- ments clusion is that with present patterns of Policies to increase the ownership of physical ownersliip and spatial distribution of land, assets by poor liouseholAs are particularly rele- possibilities of increasing poor farmers' income vant in rural areas. Their potential depends on by mneans of high-yielding investments are the initial occupational distribution of rural limited. households. Without considering transfers of Countries that hlave undertaken land reform existing land, its effectiveness depends on: the include Mexico (1940) and Bolivia (1955), fraction of existing landowners in the poverty where large 'haciendas' farmed with traditional group, such as the ratio of subsistence farmers technology were subdivided into small farms. to landless labourers among the poor; the possi- Peru is still implementing the land refornm bility of finding high-yielding investments begun in 1973. The general evidence as to long- among subsistence farmers, and the yield on term effects of such reforms indicate that small the investment in extending cultivable land farmers have increased their previously low eventually to be transferred to poor house- income, and farm output has increased overall. holds. When a redistribution of existing land is Several specific lessons seem to emerge from feasible, such investments can then be calibrated those experiences: the iminimum farm size to take into account the complementarity required to sustain farmers' incomes and to between investment and transfers. take advantage of minimum economies of scale In the absence of redistributions of land the is still high, consistently ranging between 30 effectiveness of investment policies will depend and 40 ha, of which at least two-thirds is crop- on how important is the fraction of landowners land and pasture; given this required farm size, or small farms in the poverty group. Unlike in only 30--50%10 of the rural population benefits. Asia, nmost rural poor in Latin America are land- Over timne, high poplulatioll growth erodes the less, accounting for two-thirds ol the rural relative initial effect of the reform by increasing poor. The question arises, therefore, whether the relative number of landless fanilies, main- the present enthusiasm for a 'small-farmers tenance of productivity gains not only requires strategy' to poverty alleviation can be justified high initial levels of investment, but also high in the context of Latin America. Without levels of technical assistance and cretlit over redistributions of land, that strategy reaches time. The positive effects are less clear when only a small fraction and the better off of the large conmrmercial estates or imodern livestock rural poor. operations were subiject to land reform. There Credit, extension services and investnment in is a difference between redistributing land from roads and irrigation schemes are typical inter- 'haciendas' with traditiondl crops and redistri- ventions that could increase the productive buting land fronm high-technology, highly capacity of small farmers. The question is tiow nmechaniized estates. many of these investlnents have a high rate of What can we learn from these experiences return and could easily be targeted] to the poor. so as to evaluate options available today? In INCOME DISTRIBUTION, BASIC NEEDS AND TRADE-OFFS WITH GROWTH 83 countries where some reforms have taken place where this is more probable. and where unreformed land consists basically of What is the possibility, in some of these large, commercial estates, for which economies countries, of redistributions of land that of scale and technology are important (live- either maintain their production level without stock, sugar, cotton), or of large traditional additional investment resources or, if additional haciendas with small cropland potential (large investment resources are required to raise the estates in Mexico and Peru have only 10%o of incomes of the beneficiaries, they do it wvith a the cropland but 35% of pasture land) agrarian high rate of return. I reform options are less clear. Redistribution of Several simulation exercises have been made highly commercialized estates can entail a high assuming that the high productivity per hectare, efficiency cost unless co-operatives are im- usually observed in smaller farms, will be repli- plemented to maintain the gains of economies cated when land is redistributed. Obviously, of scale; but the latter involve changing difficult this always yields an increase in outptut, 'rhe social and institutional constraints. Redistri- problem is that the associated investment cost buting traditional haciendas of lower-quality to maintain this replication - i.e. roads, irri- land, basically suited for pasture, requires high gation, water, extension is not bi.-ught ex- additional investments to support smaller plicitly into the picture so as to allow the family units. computation of the net return to the reform. The agrarian reform option becomes more The IBRD SUDENJI study on Northeast important for countries that have not experi- Brazil provides such an information. Tbhs study enced redistributions of land and where the develops a model simulating the effects of land increase in output from reformed land could be redistribution where 790,000 new farm modules achieved at good rates of return on the addi- of 34.8 ha each are created. 20 The initial con- tional investment required. Brazil, Colombia, ditions and the results of the land reform are and to a lesser degree, Ecuador, are countries summarized in Table 7. Wnrkers without access Table 7. Rural Northeast Brazil* 1. The rural poor in the present: a functional breakdown Dependent population Poverty level Classification Number of agents (million) Less than US$50 per capita Temporary workers 2.7 million or less 3.5 Not substantially Sharecroppers 240,000 1.2 above US$50 Permanent workers 895,000 4.5 per capita Smallest farmers 67,000 0.335 Above US$50 per capita but less than $130 Medium farmers 384,000 1,92 Totals 4,3 million 10-12 million 2. Potential for land redistribution Number of existing large farms 37,038 Average size (ha) 742 Module size 34.8 Potential number of modules 790,000 3. Effects of the land redistribution Net money income per module US$781 ($130 per capita) Net increase in value added* US$219.2 million Required additional investment per module (US$3500 per module) I IS $ 2 7 65 million Rate of return of the investment cum redistribution 8% Source: Kutcher and Scandizzo, The Agricultural Economy Northieast Brazil (forthcoming). 4' Net of accruals to all factors other than land and over and above the opportunity of labour. 84 WORLD DEVELOPMENT to land currently represent two-thirds of the supplying them, particularly when the supply rural labour force and the main core of Brazi- of these services is characterized by strong lian poverty. The establishment of such 'module complementarities and economies of scale. farms' gives land access to 20% of the landless Two sets of issues are central in evaluating labour in the poverty group (temporary and these policies. First, at the macrolevel, they permanent workers and sharecroppers), increas- involve some trade-off with growtlh to the ing their per capita family income from 50 to extent that these services may have a more 130 dollars/yr. The net increase in value added limited effect on the productivity of the poor. of the region, net of accruals to all factors other What becomes important is to calibrate this than land and over and above the opportunity trade-off at the present (high) level of per cost of labour, is 219 million dollars. The re- capita income in these countries; this trade-off quired additional investment per module is appears d ffercnt today at per capita incomes of $3500, a total of $2.7 billion. The implied rate $1000 than it did 25 yr ago when these coun- of return to this redistribution cum investment tries had only half this income level. The is 8%. second set of issues is microeconiomiiic: (a) how The effects of this rather radical agrarian can these services be delivered withiout major reform on the regional distribution of income leakages to other income groups, (b) without are as follows. The percentage of households inducing an effect simply equivalent to an below a poverty line of $78 per capita declined income transfer, and (c) how can children be from 86.6 to 77.8%, a substantial change in reached when intra-family leakages also occur? absolute numbers, but clearly not the solution Nutrition and child care, consumption of for the problem of poverty in the rural North- health services, shelter with water and sewerage, east. Although the absolute number of bene- and basic education are typical services that ficiaries is high - around three-quarters of a determine 'quality of life'. The specific policy million households - the absolute base of choice and mix are clearly country and regional people in poverty is equally large: 4 million specific, and involve a high degree 1f value members of the rural labour force still remain judgment. Even objective judgments 'rom the in absolute poverty. medical field, such as 'optimal calor,e require- ment', are still highly controversial. (c) The direct provision of services: the 'basic- The choice of a specific basic-needs strategy needs approach' depends on the initial deficit, the age compo- sition of the poor, the effect of rural-urban (i) Introduction: toward a children-oriented migration in redefining basic needs over time, basic-needs approach and the income elasticity of the poor for dif- Measuring poverty alleviation in terms of ferent basic needs, namely, the rate at which welfare indicators other than income opens a the deficit will be closed by income growth in new set of policy considerations. Not only the poverty group. per capita income is unequally distributed, but Several factors point toward a children- calorie consumption, life expectancy and infant oriented approach to basic needs. First, basic- mortality. They reflect inequality in the con- needs deficits are associated with the sumption of particular services, such as distribution of income, and poor families have sewerage, vv;iper, education and health. Poli- more children; the incidence of children with cies to inci ae their consumption above levels basic-needs deficits is larger than adults. Sec- resulting from private demand and supply of ond, life expectancy at birth - basically deter- these services define what has become known mined by infant mortality - is perhaps the as a 'basic-needs' strategy. It implies interfering prime basic need. Infant nutrition, lowering of with resource allocation to reach these critical infectious diseases due to better shelter, and consumption levels at early stages of develop- childcare are crucial inputs in this process. ment, despite strategies to accelerate the Third, evidenice shows that the marginal pro- growth in income of the poor. This strategy pensity to spend on children (by definiing a becomes more important the smaller the broad concept of such spending such as trickle-down effect of high growth rate policies mothers' time, quality of parent-children on the income of the poor, the fewer the interaction) i. low. Additional rural-urban possibilities of increasing the earning capacity migration, a distinctive feature of the future, of the poor by high-yielding investments in will compound the problenm. All these pro- their assets, the smaller the marginal propensity pensities are lower in urban areas. In sum, of the poor to spend on basic needs, and the normal income growth will not solve the basic- lesser the possibility of the private sector needs deficits of children. INCOML DISTRIBUTION, BASIC NEEDS AND TRADE-OFUS WITH I GROWTH 85 Table 8. C'ost of tie caloric deficit as a frac-tiont of' amount spent on preventive imedicine as a per- (;NP (M) centage of C,NP has remained constant. Mn cThe expansion in the supply of rural, pre- Mean calorie availability over ventive-orieilted health care will not, per se, guarantee increases in the consumption of 0.9 1.0 1.1 these services. Intervenitions inecrea.sing the --_ .__ -_._ - utilization of health institutions anid in those 600 1.95 0.70 0.0034 factors affecting the dei.Lnn(d for healtli ser- 800 1.46 0.52 0.0025 vices must also hbe ietitified. It was found in 1000 1.17 0.42 0.0020 urban C(olombia that, controlling for the 1100 1.06 0.38 0-0018 income of the houselhold, the level of family education had an important imipact on the (ii) The calorie dteficit number of visits to doctors. In fact, doubling the level of schooling of the household head dtountrywide cLories in thep pultini increased per capita visits by 18p n Thus, any dstrbuto Oa caloroes n the populaton Isi tiosess extremely scanty. Anl alternative is to allocate schooling probably being one of them, becomes kniown aggregate (or country mean) consumip- a point of intervention, tion across income groups by using available HLow expensive is a prevenitive-cre-oriented income distribution data and asiptiois hlealth system aimed at a full coverage of the about the relationship between consumption and income.2' population'? Preliminary estimates for Brazil Consumption by income groups was derived show that such a system could he financed by for alternative values of mean countryper capita maintaining the present 2.5`' share of health consumption: 0.9; 1.0 and 1.1 times the mean c in in GNi.22 e rt Te *n c - d i Ihese estimiates are based on a stralegy that calore requirement. The ;esulting calorie deficit would change the present 'urban-based- was priced using the 'cost per calorie' b iypClcit individual- cuLrative system' trend. The alterna- in the cost of the food basket used by 1LA tin tive is a stronger emiiptiasis on primary and defining poverty lines. Table 8 shows the aggre- collective health care in rural areas and poor gate cost of the calorie deficit as a percentage regions based 0n the PIASS plan (Program for of GNP. The results are shown for countries of Grass Roots liealth and Sanitation Actions different GNP per capita and dlifferent mtieanl in the Northeast). The principle belhind PIASS calorie availability over mean requirement. is that 'a large variety of health problems may tUnder the worst of circumstances, this cost is be successfully prevented and/or treated at the between 1.5 and 2%3.e of GNP for countries of coinunily level, without recoure to expensive US$600---800 per capita income and abotut hospitalization in urban areas, and that pre- 1P'; for cotuntries above UJS$1000. ventive and simuple curative services should be integratcd with more complex curative services (iii) Options irn the hiealth sector: thle change of throuighi well-defined institutional channels'.23 present biascs This system is based on three levels of medi- Present public healtli systems are cliaracter- cal facilities of inicreasing complexity: first, ized by an expensive urban-curative orien- rural health posts in comnmunities of 500-.2000 tation. Social security systems have become inhahitants staffe(d by local auixiliaries, whose incrCdsilgly im7portant in providing healtli ser- funetion is to prevenit infectic;is (liseases and vices, hut because tlheir benieficiaries are miainly detect iuorC complex diseases. Seconid, healtli middle-income urban eniployees. belonging centres servicing tlhree healtlh posts and staffed either to the public sector or to the modern by a nurse, a sanitation auxiliary, and a day-per- private sector, they fail to reach non-affiliated week visit by a doctor based in a larger leDalthi urban workers and ilidividuals in rural areas. centre. This larger health centre is attaclhed to a In Colombia 11ousCholds in the poorest (quin- r.egi0nal hospital, whicli represents the third tile lixing in large cities receive a subsidy from level of health facilities and also serves the urban the public health sector 5 times larger than population. those in rural areas. In Brazil the percentage of By eosting oUt the expansion of such insti- public healtlh expenditures devoted to preventive tutions (operatinig an(d capital costs) and by medicine declined from 87.1 to 29.7 from projectilIg [Moplulation growth, two alternative 1949 to 1975. The effect is that, in spite of an proJections are made and shown in Table 9. increase in total public expendituires in health The first aims at full coverage by the year 2000 as a percentage of GNP from 1.0 to 2.5, the and assumes a 7% GDP growth rate in the 86 WORLD DEVELOPMENT Table 9. Brazil: coverage and costs of expanding a restructured health system Percentage coverage Urban Rural Northeast* Cost per cent of GDP Projection 1 Projection 2 Projection 1 Projection 2 Projection 1 Projection 2 1975 79.0 79.0 0(50) 0(50) 1980 100.0 92.8 30(60) 15(60) 2.3 2.2 1985 100.0 96.4 65(80) 30(65) 2.4 2.4 1990 100.0 100.0 95(100) 45(70) 2.3 2.5 2000 100.0 100.0 100(100) 70(80) 2.0 2.5 * This coverage reflects the expansion of the first level of health posts. Values in parentheses refer to the capa- city of the second and third levels, wlhich are highly influenced by the present structure of the system. period 1980-2000. The second involves a more remain in schlool. Subsidies to school attend- moderate increase in coverage and assumes a ance, such as school breakfasts and lunches, GDP growth of 4%. In both cases the total free provision of textbooks and clothing, and health expenditure as a fraction of GDP remains transportation, are relevant policy options. In in the 2-2.5%Yr range. the state of Goias, for example, experiments with This exercise shows that 2.5% of GDP can school meal programmes increased effective finance a proper health system in a $1000 per school attendance from 67-80 to 84-95%.25 capita income country with reasonably good Anotller constraint is that teachers in rural growth, if the system is restructured to a less primary schlools, 1larticular-ly in the more back- expensive primary-care-oriented system. Most ward regions, have extreimely low levels of of the other countries discussed here spend pedagogical training. Ninety per cent of the between I and 1.5% of GNP on health; there- teachers in Northeast Brazil lack pedagogical fore, it is likely that an additional percentage training altogether and most have little primary poilnt of spending, plus a restructuring of their schooling. In 1972, 31%' of the Northeast systems along the lines described previously, teaching force had not completed primary edu- could have considerable impact on the health cation, and in 1974, 77%0 of these teachers were coverage of the population. reported to have had only primary schooling. The inability to attract better teachers restilts (iv) Issues in delivering educationial services from the low salaries paid by the rural munici- The first priorities in expanding educational palities in Northeast Brazil. Teachers with comn- services are to expand rural enrolment rates by plete primary educatioll received between increasing incentives to attend school, and to US$10 and $51/month, compared with the increase the quality of teachers and teaching legal minimum salary of US$57. They usually materials in rural areas. Neither goal can auto- work without contracts, and their salaries fluc- matically be attained by simply expanding the tuate according to the budget of the rural muni- physical supply of schools or school places. cipio.26 Expansion of rural enrolment implies From the previous data it is clear that a drawing an increasing nurnber of children from larger fraction of a country's primary educationF families of lower socio-economic status and less budget must be used, first, to increase incentives densely populated areas. Their low enrolnment is to attend rural .schools, basically school meals the result of factors affectinig the demj.:i for (this is quite related to the nuitrition objectives schooling as well as of the availability of school discussed earlier), and second, to irnprove, the places. Low e(lucational levels of parents and working conditions and increase the salaries of the cost of sclhooling foregone earnings and teachers in rural areas. transportation costs - are crucial factors in Prclimlinary estiniates show that, for Latin influenciCng the demand foradditional schooling. America, providing 6 yr of schooling to the An illustrative case is Brazil. In 1974, enrol- overall population would increase the edu- ment rates of children aged 7-14 were 72.5, cational budget by 6 17 %.27 Because the share 55.3 and 44.4% for the total country, total of expendittures in education in (NP is approx, rural areas, and the rural Northeast, respect- 3 -4'S, this increase amounts to 1/5F-2/3 of one ively.24 The constratint is not the lack of school per cent of GNP. This is the cost of expan(ling places hut rather an insufficient incentive to the suLipply of primary school places: an ad- INCOME DISTRIBUTION, BASIC NEEDS AND TRADE-OFFS WITH GROWTHl 87 100 , Figure 1, for Colombia, shows a Lorenz curve E for the distribution of personal income and the E I / . distribution of the subsidy to primary and 80 higher education; it is clear that the subsidy to Subsidy to / higher education is substantially more unequally 0 primary educc¸tlon 0 °/ l distributed than personal income. Equity ancd efficiency within the educational ine nt of free higher education by a system where / Personl tuition covers the true cost of edlucation and / where students are given loans - to finance tui- _____ " 20 /, _ tion plus foregone earnings - to be repaid after 20 0/ to graduation. higher ecmucaton (v) Housinig anzd associated services in urban 20 40 60 80 I00 areas: a cost simlulation exercise Percentage distribution of households, % Locationi and typical dwelling standards Figure 1. Distribution of t/e sutbsidv to educatiotn anid limit the access of the poor to new housing and of persotnal income. a lack of capital miarkets compounds the prob- lem. Table 10 shows thiat in Bogota and Mexico ditional fiscal effort is required if better teacher City the cheapest housing unit currently avail- salaries and schlool mleals in rural primary able cannot be afforded by 47 and 55'7 of the schools become also part of this package. What inhabitants of these cities, respectively. Alterna- additional sources of finance can be cornecivi2d tive lhousitng standards and locations are identi- within the educational sector? fied to analyse the sensitivity of costs and the The miiost obvious is to gradually eliminate suibsidy required to finance the dlifferenci the subsidy to higher education and to transfer between mortgage playllWnts and hlousehold these funds to primary rural education. This is c\pendictures for rents. Practically all house- an important source of financing because in holds in Mexico City couild affordl ;nulti-frallily most countries the subsidy to higher edoi- dwelings with basic services located on the cation is half the subsidy to primary education. periphery of the city. In Bogota, only 15" of Subsidies to higher education are perhaps one households would be unable to finance these of the most regressive transfers in the economy. types of housing: to afford these dwellings, a Table 10. Cost of houising utnits of various standards and lot;ationIS, pcrcentagc of houzseholds iunaablc to afford them antd required subsidy oni their rents* (in 1976 US$) Bogota Mexico City Location Cost Houiselholds Subsidy Cost Households Subsidy (G0) (%) (%) (%) Periphery Single family, indi idual services 1256 17 21.7 2063 14 9.2 Multi-family, individual services 1809 36 21.9 1960 12 3.4 Multi-family, basic services 1134 15 12.9 1274 4 Intermediate zones Single family, individual services 4787 73 71.1 31,184 95+ 92.6 Multi-family, individual services 2509 50 44.1 7777 72 70.1 Multi-family, basic services 1837 36 23.5 7096 69 67.1 Present cheapest housing unit 2477 47 4988 55 Per capita income country US$635 US$1090 Source: Orville Grimes, Housing for Low-Income Urban Farnilies (John Hopkins University Press, 1976). * Subsidy is defined as the percentage by which monthly income available for housing falls short of required montly payment. Based on repayment period of 25 yr, 10% interest rate and a share of income devoted to rent equal to 15'7. 88 WORLD DEVELOPMENT subsidy of only 12.9% on the required mortage sive segments of subsidies within each sector. would be required. The issue is how to imple- Without this qualitative change, those ad- ment these new standards of housing. ditional resources will lose a large part of their Because these examples are quite specific to poverty-oriented objective. The share of GNP a particular metropolis (for example, the results spent on higher education must be kept constant are very sensitive to location in the case of or decline as a consequence of a policy of self- Mexico City), a sensitivity analysis for a typical financing. In health, the share of urban-based country of US$800 was undertaken with the curative medicine must be kept constant, and objective of computing the subsidy that would all increments devoted to primary-preventive allow urban households in the poorest 40% (of rural health care. total households) to afford: (a) a dwelling For countries like Brazil, which have already equal to twice the per capita GNP, $1600, reached 2.5% public spending in health, ad- (b) a dwelling 5 times the per capita GNP, ditional funds amount to only 3% of GNP. For $4000. other countries the increment reaches 5%, a If urban households in the poorest 40% doubling of public spending in these sectors. devote 15% of their income to rent, the required Table 12 presents some options for financing fiscal subsidy on housing would have to be such a programme. In the oil-producing coun- 0.21% of GNP for the first type of dwelling and tries -- Colombia, Mexico and Ecuador - 1.28% for the second type. The covesage would domestic oil prices are kept substantially below be all urban households in the bottom 40% of world prices. It is estimated that lifting them to households, accounting for 30% of all urban world price levels would increase government households. revenues by 3, 2.5 and 4% of GNP in these countries, respectively. Half of this increase could be devoted to finance the basic-needs 3. BALANCING THE OPTIONS: TRADE- package, the other half to finance investment in OFFS BETWEEN BASIC NEEDS AND the oil sectdr. In the case of Mexico. additional GROWTH oil production will increase governmLtnt saving, even by maintaining low domestic pi_es, by an additional 4.3% of GNP (from 2.2 to 6.5%') by (a) Financing the basic-needs package the year 1982.28 Half of this increase, 2.15% of GNP, is another source of finance. Based on the previous considerations, our If lifting the price of oil (row I in Table 12) recommendation is to increase public subsidies is considered politically unfeasible or having a to basic-needs-oriented sectors, from their regressive impact (which we doubt, an important present level of 5-7% of GNP to a new level of issue to be researched), additional taxation 10% of GNP (Table 11). This must be accom- would be the alternative option. panied by changing the nature of delivery Not a single report fails to emphasize that systems in each sector and transferring regres- there is still room for additional fiscal revenues Table 11. Resource requirements Resources per sector (as % of GNP) Sectors Actual Target Reallocations within sectors Education 3-4 5 Toward self-financing of higher education: food programmes and teacher's quality in rural areas Health 1-2.5 2.5 Pro collective-preventive rural health care systems Nutrition <1 1.0 Target-oriented instead of general food subsidies Housing < 1 1.5 Eliminating subsidies to middle- class housing. New standards for lower-income groups Total 5-7 10.0 INCOME DISTRIBUTION, BASIC NEEDS AND TRADE-OFFS WITH GROWTH 89 Table 12. Sources of finance (percentages of GNP) Brazil Peru Colombia Mexico Ecuador Hfalf the increase in net revenues out of increasing the price of domestically sold (and produced) oil to world prices* 1.5 1.25 2 Half the additional savings out of increased oil production (at present domestic prices) 2.15 Additional taxation (in parentheses, taxation from eliminating exemption on imputed rent on owner- 3 4 1.2 2 4.8 occupied dwelling) (1.2) (1.2) (1.2) (1.2) (1.2) Eliminating the subsidy to the electricity sector 1 Additional government royalties from coal and uranium exploitation 1 Total 3 4 4.7 5.4 6.8 * The additional oil revenues obtained by charging domestic consumers thc world price is, as a fraction of GNP, 3, 2.5 and 4% for Colombia, Mexico and Ecuador, respectively. If only 0.5% of GNP can be raised by manipulating this option, the difference will have to be obtained by additional taxation. by increasing (effective) direct taxation at the been contemplated here, except for Colombia, top of the income distribution and also by where l% of GNP could be made available by increasing the tax base. In several oil-producing eliminating the subsidy to the electricity sec- countries, additional oil revenues have been tor. Defense represents at least one-fourth of accompanied by declines in the tax ratio (taxes the fiscal budget in several countries. In Peru, over GNP) in the rest of the economy. In the sevcral of the present subsidies to electricity case of Ecuador, this ratio declined from 0.186 and consumer goods -benefitting consumers to 0.138 between 1972 and 1977.29 Main- of all income groups - are not considered here taining the 1972 figure yields the additional a source of finance because their elimination is share of 0.048 uised in Table 12. part of their present anti-inflationary pro- It is estimated that eliminating tax exemp- gramme. In the case of Brazil, the investment tions on imputed rent on owner-occupied budget of the government is held constant in dwellings -- one of the most regressive subsi- this exercise. dies - could yield a tax equal to 1.2% of GNP. If the upper 10% of the population receives a (b) Trade-offs with growth: keeping share of GNP equal to 0.4 and their imputed raffthgrotvn rent is 0.2 of their income, their tax base will uncer/aint down increase by 0,08 of GNP. If their income tax is % it yields an additional 1.2% N wth rate induced y this aditioal of In several countries, particularly Mexico and 'NP devoted to basic needs? Let us assume the Bra7il. either inflation or high levels of deduc- woP devo tions erode the tax base. In Brazil tax de(uc- worst scenario: tions exempt families with gross inco ies up to (a) This additiona! basic-needs package has 4 or 5 times the per capita GNP. In C'olomibia, .ero productive or asset creation effect on adjusting property prices at comnmercial values the poor. We conceive of it as a pure con- could yield an additional 1 c of GNP, and tax sunmption transfer (tlhough this would not reforms now being contepnnlated in Peru could be the case, Piarticularly in the case of edu- yield adiditional taxation equal to 4%,r of GNP. cation). Sources of finance from reallocating subsi- (b) Oil rcveiiues are not feasible sources of dies presently given to othier sectors lhave not financing; tlherefore, the 5%>^ is financed 90 WORLD DEVFLOPMENT Table 13. Declines in the couintry's investment rate Closed-economy case Open-economy case Interest Marginal Interest Interest elasticity elasticity propensity rate Total of demand for Total of savings effect effect effect investment effect e = 0.5 --0.1 0.007 --0.017 17 0.5 -0.0078 e = 1.0 --0.01 - 0.014 --0.024 1.0 0.0"56 1.5 -- 0.0234 entirely by an additional 12.5% income tax, initially: (a) this income group investmenlt ac- levied on the income of the richest 10% of counts for 10% of GNP, a rather high figure; the population accounting for 0.4 of GNP. (b) this group pays a 20%h (t,) income tax; In terms of investment incentives, this tax, the additional 12.5% tax (t2) is equivalent, which affects the income of new investment, therefore, to a decline of 1 5.6% in the yield of is clearly worse than the tax options discussed new investment (t2 / - t1 ). In the closed- previously, such as the elimination of deduc- economy case, the results are a function of the tions and exemptions on imputed rent and marginal propensity to save and the interest increases in property taxes (which tend to elasticity of the supply of savings, e. We assume tax a larger share of iriframarginal income). a rather high figure of 0.2 for the propensity and 0.5 and 1.0 for the elasticity. In the open- What is the maximum effect of this tax oni economy case only the interest elasticity of the the investment behaviour of the richest 10%. demand for investment, 71, is relevant; a range and its impact on the overall country investment between 0.5 and 1.5 is used in the analysis. rate? Two capital market scenarios are pre- Table 13 shows the results. In the worst of sented to estimate this effect. Within each, circumstances, the decline in the country's extreme parameter values and assumptions are investment rate is 2.5 percentage points; that is, used to derive an upper bound for the decline if the rate was initially 20%7, the new rate will in investment.30 be 1 7.5'S. If the social return to capital in the One scenario is a closed-economy capital economy is 0.2 (gross of taxes, depreciation market, that is, no capital flight takes place. and excess market wage over the opportunity The decline in investment results from both a cost of labour), the decline in the growth rate smaller supply of savings (a function of the w%ill be at most one-half of 1%, (0.2 x 0.025). marginal propensity to save by this income If a country's historical growth rate was 6%,/yr, group) and a decline in the net yield of new the maximum effect will be to reduce that investment opportunities. The decline is largest rate to 5.5%: at the tnost 'the trade-off ratio' by assuming a perfectly elastic demand for between redistribution and growth is 10:1: in investment, that is, the new equilibrium net other words, a 5% transfer of GNP toward 'con yield goes down by the amount of the ad- sumption'results in a half a per cent lower GNP ditional tax. growth. In the open-economy case we assume that Crucial in the implementation of such a domestic investors can invest abroad and that policy is to keep uncertainty down. Investment the return abroad is not affected by the new behaviour is much more sensitive to uncertainty tax. This assumes that either the proceeds of than to net yields; if the tax reform takes place this investment can only be consumed abroad in a milieu where other 'rules of the game' - or the proceeds can be brought back into the concerning other property incomes or property country in a way such that the new tax can be rights - are being question:ed, the negative evaded. effects on investment could be substantially To derive maximum values we assume that larger. NOTES 1. Vinod Thomas, The measurement of spatial de la pobreza: aspectos conceptuales y metodologicos', differences in poverty: the case of Peru', World Bank mimeo (Proyecto Interinstitucional de Pobreza Critica Staff Working Paper No. 273 (January 1978). en America Latina, L:CLA, May 1978). 2. Sebastian Piniera, 'Definicion, medicidn y analisis 3. Richard C. Webb, Governmenzt Policy and the INCOME DISTRIBUTION, BASIC NEEDS AND TRADE-OFFS WITH GROWTHI 91 Distribution of Income in Peru, 1963-1973 (Cam- 15. Mareelo Selowsky, WhoBen efitsjfrom G(,o,erntment bridge, MA: Harvard University Press, 1977), p. 13. Ecvpenditures: A Case Stud)' in Colombia (London: Oxford University Press, 1980). 4. Urrutia and Berry, Itcoene Distributioni in Colom- bia (New Hlaven, CT: Yale Unvcrsity Press, 1967). 16. Marcelo Selowsky, 'El cfecto del desempleo y el crecimniento sobre la rentabilidad de la inversion edu- 5. G. Kutcher and P. Scandizzo, The Agricultural cacional', Revista de Planeacion y Desarollo (Bogota, Economny of Northeast Brazil, forthcoming. Colombia), No. 2 (July 1969). 6. S. Pifiera, 'Cuantificaci6n, Analisis y Descripcion 17. If any change in a worker's occupation (associated de la Pobreza en Algunos Paises Latinoamericanos' with additional schooling) is the result of education (Proyecto Interinstitucional de Pobreza Critica, ECLA, acting as a rationing device in higlhly paid sectors or 1978). occupations, the previous results would be the rele- vant ones. 7. S. Pixnera, 'Se i3enefician Los Pobres del Cre- cimiento Economico?', mimeo (ECLA, 1978). 18. Phillip Musgrove, Consumer Behavior in ,atin Amierica: AnZ ECILY, Study (Washington, D.C.: The 8. For a general taxonomy of policy and non-policy Brookings Institution, 1978). -induced distortions, see J. N. Bhagwati, 'The gencral- ized theory of distortions and welfare', in J. N. 19. The mcan rates of return for Brazil, Colombia and Bhagwati et al., Trade, Balance of Payments and Chile are: Growth (Oxford: North Holland, 1971). primary education 25.1%, 9. This is less true for social security contributions; secondary education 17.0%, to the extent that workers perceive as benefits only a hig fraction of these contributions, this legislation becomes From M. Selowsky, 'Balancing trickle down and basic partly a tax on the use of labour. needs strategies', op, cit. 10. C. Dougherty and M. Selowsky, 'Measuring tlle 20. G. Kutcher and P. Scandizzo, op. cit. effects of misallocation of labor', Review of lEcon- omnies and Statistics (August 1973). 21. This follows a teclhnique used in S. Reutlinger 11. J. de Melo, 'Distortions in the factor market: and M. Selowsky, 'Mialnluitriiion and poverty', World some general equilibrium estimates', Review of I,'con- Bank Occasional Paper No. 24 (1976). A semi-log omnics and Statistics (November 1977). calorie income relationship is specified with an elasti- city equal to 0.15 at a level of consumption equal to 12. Assume a Cobb-Douglas function X - .L e requirements. where A is the growth rate of all other comnplementary factors of production, capital and technical progress. 22. Peter Knight and Dennis Mahar, 'Brazil human Differentiating logarithmically the first-order con- resources special report, anne- III: health and nutri- ditions yields: tioii', World Bank, mimeo (25 October 1978). diovgw - (a- 1)dlogL + A, diogw= (c - 1)logL+ X,23. Knight anid Mahar, op. cit. where w is the wage rate. Assuming a = 0.5, the value of N consistent with Table 4 is 0.055. Given X, we canl 24. Yves Tencalla, World Bank, Brazil Education derive the trade-off between dlogw, and dlogl. Sector Memorandum, mimeo. 13. Simulation exercises, based on human capital 25. Federal University of Goias-INEP -Sl C-CNAE mTodels witlh complementarities between preschool Project (1976). abilities and education. yielded a figure of 1% of GNP as the amount to be invested in preschool progranmnes 26 Tencalla op it for children in the poorest quintile of the population. See Marcelo Selowvsky, 'Nutrition, health and edu- 27. Sebastian Piiiera, 'Preliminary estimates from the cation: the economic significance of comiiplementarities Critical Poverty in Latin America Project' (ECLA). at early ages', Presented at Hrwrnan Resources, Enpiplo'- ment and Dereluptnent, 6th World Congress of the International Economic ikssociation, Mexico City, 28. World Bank economic mission estimate. 4-9 August 1980. 29. WNorldI Bank econo)mic mission estimate. 14. A. Berg, The Nutritiotn Factor: Its Role itn National Development (Washington, D.C.: The Brookings Institution, 1973). 30. See Appendix for details. 92 WORLD DEVELOPMENT APPENDIX Let us denote: r = gross rate of return to investment, T7 = elasticity of investment demand with t, = initial income tax paid by the richest 10% respect to the interest rate, of the population = 0.2, l = initial investment of the richest 10% over t, = additional income tax = 0.125, GNP = 0.1, m = marginal propensity to save by the richest Al = change in the investment of the richest 10% = 0.2, 10% as a fraction of GNP, e = supply elasticity of savings with respect to - = + A,, A1 = propensity to save effect, the interest rate by the richest 10%, A, - lower interest rate effect. S savings S sovrngs r r abroad . r dormestkc r(i-f) L A2 A A L I 'r-I t2) '0 10 Investment by the richest il. Irwestment by the richest 10% asa froction ot GNP as a fraction of GNP Figure Al. Closed-economy case: no capital flight (maximum effect: T? = 00) Figure A2. Open-economy capital market (maximnurn effect: returns abroad not lr = A1 + A, = -[°'.4t2 n ml-(I )(il-§)e affected by taxation) =-0.4t,ml -tI-0.4t,in)(l2)e 1 THE WORLD BANK Headqzrnrtcrs: 1818 H Street, N.W. U Washington, D.C. 20433, U.S.A. European Office: 66, avenue d'Iena 75116 Paris, France Tokyo Office: Kokusai Building, 1-1 Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan The full range of World Bank publications, both free and for sale, is described in the World Bank Catalofg of Publications, and of the continuing research program of the World Bank, in World Bank Research Progranrt: Ab- stracts of CUrrLirt Studies. The most recent edition of each is available with- out charge from: PUBLICATIONS UNIT THIE WORLI) BANK 1818 1-1 STREET, N.W. WASH INC,TON, D.C. 20433 US.A.