\JPs I23 POLIcy RESEARCH WORKING PAPER 1623 Environmental Degradation Because of the importantrole that children play in coilection and the Demand for activities, the dernanc for Children children may increase as local environmental resources are depleted, setting up a viJious Searching for the Vicious Circle circle between resource depletion and population growth. Analysis of Deon Filmer household data frorm Pakistan Lant Pritchett yields some support for tiis hypothesis, although the effect may be small and dependent on endogenous local property rights. The World Bank Policy Research Department Poverty and Human Resources Division July 1996 I POI.IC RESEARCH WVORKING PAPER 1623 Summary findings Filmer and Pritchett explore the hypothesis that - * Collection absorbs a quarter of the time of children. because of the impportant role children play in collection * Women benefit when there are older children in the activities (firewood, water, grazinig) - the demand for household. They work 2.6 hours a week less in childre n may increase as local environmental resources household activities for each child aged 10 to 15, and 3.2 are depleted, setting tip a vicious circle betweeni resource hours less for each child over 15. depletioni and populatioll growth. * There seems to be a relationship between fertility U'sing a large-scale houselhold data set from Pakistan, and the availability of firewood. Even after controlling with detailed inforniation on fertility and the allocation for other determinants of fertility in reduced form of time (of womeni, childreni, and adults) to collection regressions, they show that households that live some activities, they find that: distance from firewood have more children whereas Collection activities absorb a substantial part of households that live where firewood is more expensive houselhold resources. Firewood collection accounts for have fewer children. 6.2 perccnt of houselhold expenditures, valued in collection time. T-his paper- a produrct of thie Poverty and Human Resources Division, Policy Research Department- is part of a larger effort in the department to investigate the social and environmental consequences of growth-oriented policies. Copies of rhe paper are availahle free from the World Bank, 1818 H Street NW, Washington, DC20433. Please contactSheilaFallon, roomn N8-030, telephone 202-473-8009, fax 202-522-1IS., Internet address sfallon@worldbank.org. July 1996. (5U pages) Ihi, Policy Researck Working IPaper Series disseminates the findings of work in progress to encourage the exchange of ideas about etCel,)plient iss4tcs. An obje,ti!e of the series is to get the findings out quickly', even if the presentations are less than fullypolished. The pipers carry tht natnes of the authors and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors'owun and sl'nuld not be attributed to the Wor/d Bank, its Executive Board of Directors, or any of its member countries. Prodtucced by the Policy Research Dissemination Center Environmental Degradation and the Demand for Children: Searching for the Vicious Circle Deon Filmer World Bank Lant Pritchett World Bank I E-nvironmental Degradation and the Demand for Children: Searching for the Vicious Circle' As environmental degradation and population growth have come to the fore of the development agenda, so too have beliefs and theories that suggest that problems of poverty, population, and the enviroment are inter-linked. In partcular, some have proposed that environmental degradation and population growth may exacerbate one another in a vicious circle in which greater population leads to a worsening environment and a worsening environment leads to more rapid population growth (Dasgupta, 1995, Cleaver and Schreiber, 1995). One half of the vicious circle may seem obvious, that greater population in a given area leads to pressures on, and ultimately degradation of, the natural resource base: soil quality, water, grazing areas, wood availability2. The explanation of the second half to the vicious circle, that greater envirommental degradation leads to increased population growth, is less straightforward. Most conventional We would like to thank Kenneth Chomitz, Lawrence Goulder, the seminar participants of the 1nvironmenta Policy Forum at Stanford University and the Advisory Committee for the research project, 'Social and Environmental Consequences of Growth-Oriented Economic Policies' for helpful comments. We would like to gratefully acknowledge funding from the research project. I This proposition is not at as obvious as it may seem: some elements of the natural resource base, pardcularly the soils, may be under a regime of property rights such that an increase in population density, given the right policy, leads to higher prices and increased investments such that soil qualky increases with greater populadon density. A recent study of the Machakos district of Kenya from 1930 to 1990 shows that a five-fold increase in population lead to no degradation (and possibly improvement) of soil quality, as firmers invested in terracing and other agricultural improvements (English, Tiffen and Mortimore, 1994). 1 theories suggest that greater natural resource scarcity should lead to a lower demand for children and slower population growth, both dtrough the direct effect of lower income in reducig demmd and because the scarcity of resources reduces the productivity of children as household produes. In this case the fertility response to environmental degradation mitigates, rather than exacerbas, environmental problems3. What is the behavioral mechanism that would produce a vicious circle? Greater environmental degradation could lead to increased population growth if increased scarcity of natural resource goods leads to a higher relative value of children. This is possible if children's comparative advantage as household producers is the acquisition of natural resources for which the family does not incur the full cost (for example, the collection of firewood, fetching of water, or grazing of livestock on open access land). It is possible this increase in the relative value of children from increased scarcity of available environmental goods could outweigh the income and productivity effects from privately owned resources and lead to a higher demand for children. This paper empirically examines the links between children and the environment using data from the 1991 Pakistan Integrated Household Survey (PIHS)4. This survey is particularly valuable in that it contains four types of data relevant to the vicious cycle which are rarely found together; a) detailed information about the use of firewood and other energy sources (including their 3 Although this is not to say the fertility response will be sufficiently strong to avoid either a Malthusian low level equilibrium or worse, an environmental crisis, in which population pushes the natural resource base beyond some 'breakdown" point, or especially a degradation of specific parts of environment resources such as biodiversity. Negative (or mitigating) feedback only implies the likelihood of these outcomes is decreased, not increased, by individuals' fertility responses. ' A detailed description of the survey methodology for this data is in World Bank (1995) and of this and other similar Living Standard Measurement Study surveys in Grosh and Glewwe (1995). 2 sources), b) dme allocation data for children, c) fertlity data, d) household characteristics, particularly including total expenditures. Pakisn is chosen for study beas the existence of this rich data set5. The paper first briefly reviews the theoretical work on the existence of a vicious circle. The second section examines five facts relevant to vicious circle theories with the Pakistani data and shows that while some elements of the story hold, others are more problematic. The third section estimates the association between various measures of firewood scarcity and the fertility behavior of rural Pakistani women. 1) A vicious cirle: A theoretical story There are two strands to the theoretical literature on a vicious circle between population and environmental degradation. The first strand examines the consequences a vicious circle, if one were to exist. The second presents household models of fertility determination that suggest there might indeed be a vicious circle. Nerlove (1991, 1992) examines the possible consequences of the existence of a vicious circle, defined as a two equation dynamic system relating environmental degadation (ED) and population (N): ' The reader should feel free to insert here the joke about the drunk looking under the streetlight for his keys that he lost elsewhere because there was light. We admit Paistan is the country being examined only because Pakstan is the country for which we have household income, fertliy, and child time use data along with extensive data about firewood use and coliection. Future research plans to examine the same issues in Nepal. 3 ED - g(N) , where ci( YaMNO N = AED), where aA Y8/ED>O This merely stat mathematically that an increase in population (N) causes the environment to deterorate (ED) while a deerioration in the environment causes population to increase. Nerlove woris out the implications of this simple dynamic system. As with any non-linear system without stabilizing (and in this case with destabilizing) feedback mechanisms, the dynamic implications are both complicated and far from pretty. In the Nerlove papers, however, the existence of a vicious circle is only asserted as a possibility. Nerlove's justification for this assertion is:6 For example, as forests recede up the mountain sides, parents may perceive a greater benefit of having an additional child to gather firewood. More reaisfically, in a poor agricultural setting lower quality environments may be associated with a greater livestock component in total production. [...] Arguably, children have a comparative advantage over adults in tending livestock in contrast to the heavier labor of planting, tilling and harvesdng crops. Thus, enviromnental deterioration may well enhance the marginal productivity of children, at least relative to family productivity. Dasgupta (1993, 1995) and Dasgupta and Maler (1995) also develop theories with similar reasoning: that children are devoted to that part of family income which is derived from the exploitation of naural resources for which the primary cost is the time required to coUect the good. Hence, as the implicit price of those goods goes up, the marginal value product of children relative to adults rises. Parents then may have the incentive to have more children in spite of the 6Nerlove also gives a ratinlition of enromental impacts on fertility through increased mortality. However, the rationale in this case is much less clear cut as while increased mortality may plausibly lead to increased fertility, whether this leads to increased popultion depends on a greater than one for one fertility response, which is empirically debatable. 4 worsening environmental conditions, and in spite of the fact that an additional child might furter worsen these condicons for all other families. There are two distinctive features of the vicious cycle story versus the more conventional theories of fertility. First, they emphasize that a significant part of household income, properly measured, is generated by activities that use natural resources for which the cost is not fully internalized by the household7. Second, they emphasize that children as workers are not like adults, but are relatively specialized in these natural resource related activities. This paper wiU not elaborate a fulUy specified model of mral household behavior which simultaneously endogenizes all the relevant household decisions (which would have to inchlde the allocation of household members' time across activities, the decision to purchase or collect goods, and fertlity behavior). Our much less ambitious objective is to explore various empirical elements of the vicious circle story. Before attempting to estimate the association between firewood scarcity (as one indicator of the environment) and fertility, the next section examines five factual questions relevant to the plausibility of the vicious circle versus more conventional stories. II) The vicious circle facts The answers to the following five questions will help determine whether the interaction with environmental factors is likely to be important for ferdlity decisions: * How uiportant are collected goods in household resources? ' This awkward circumlocution is the result of avoiding the familiar terms like 'private property' or "commons" as in general the pattern of use and transfer rights over particular local natural resources are very much more complex than either of these paradigmatic cases and are endogenously determined, not given. 5 * is firewood collected from open access or common property sources? * Are children relatively devoted to collection activities? * Does the presence of children alter adult's time use? * Do rural families alter their firewood use and collection activities in response to the presence of children? How important are colkcted goods relative to household resources? The collected good about which we have the most information, and which is relevant to environmental degradation, is firewood. Firewood is by the far the predominant fuel of households in Pakistan, especially in rural areas. Biofuels account for 86 percent and woodfuels alone account for 54 percent of all fuel use by households.8 Ninety percent of rural households use some firewood, with some regional variations, less in the more developed Punjab and more in the more remote regions (table A1.1). The primary uses of energy by households are for cooking, 78.5 percent of use, and heating, 13.2 percent (table Al.2). Households use significant amounts of firewood on a daily basis. Among those households that use firewood the mean (median) level of firewood use is 5.8 (5) kg per day per household in urban areas, and 7.4 (6) kg per day per household in rural areas (table Al.4)9. These values vary somewhat across regions, with much higher consumption in I As measured in the Pakistan Tonne of Oil Equivalent, which differs slightly from the international standard, Ouerghi and Heaps (1993) 9 These figures are consistent with results other from smaller surveys cited in Ouerghi and Heaps (1993). These found for example, 6.9 kg in a survey of 197 rural households in northern Pothowar Plateau and 7.8 kg from 119 households in rain fed areas. Figures based on actual consumption are significantly higher than other estimates of consumption based on "needs" requirements, such as a commonly cited 2.8 kg per day estimate based on cooking and heating requirements. 6 Baluhhisan, where rural household consumption is 10 kg per day versus only 5 kgs per day for rural Punjab. Households obtain firewood either through collection or purchase. The vast majority of urban households exclusively purchase firewood (80 percent), so for the remainder of the paper we will focus on rural households. Most rural households either only collect (64 percent) or combine purchasing and collecting (11 percent), which does leave a significant fraction of rural households (25 percent) that only purchase firewood (table A1.5). These shares vary significantly across regions, with the highest share of collection only being in the Sindh region (76 percent) and the lowest in Punjab region (50 percent). The degree of collection varies across expenditure groups: the poorest 50 percent of households in the Singh region has a share of collection of 82 percent, and that in the Punjab region has a share of collection only of 59 percentl. As part of their study of household energy consumption in Pakistan, Ouerghi and Heaps (1993) calculated the value of collected firewood both at market prices, and by valuing the time that entered into the collection activity (table 1). When labor time devoted to collection activities is valued by setting the male wage rate at 4 rs per hour, and female and child time at half that, the value of collected firewood is quite close to its value if purchased in the market. I This should '1 That is, the poorest 50th percentile as measured by household expenditures per person. This number is calculated for all rural areas, and then for each province separately. " The median reported daily male wage in the PHiS sample is 40 rs per day while the median female wage X 25 and the median child wage is 20. Since employment opportunities for females are quite limited and female and child labor cannot be separated in this exercise, Ouerghi and Heaps' assumption of half the male wage undervalues female time but is a good first approximation. 7 not be surprising as many households both purchase and collect firewood and hence are likely to be near the margin of indifference between the two. Table 1: Value of collected firewood in rural areas. Number of Time Average Labor cost of collection Value of Labor cost of collectors required, distance (rs/year) coUected collection person to source (at various valuaions of wood at (valued hours per of time) market difiremly for - - year firewood pries males, Adult Other All labor Male, (tslyear) me and male valued at female aMn c as made child labor fraction of wage valuedmeinH l _______ wage___ ______ differently exp e s All Pakisan .91 1.36 699 1.78 2797 1958 2099 6.2% Punjab .95 1.28 523 1.64 2092 1491 1752 5.2% Simdh .79 1.60 1044 1.76 4176 2778 2233 8.8% NWFP .99 1.26 760 1.75 3040 2188 2920 5.4% BaluchLstan .75 1.41 1091 3.79 4364 2939 2819 7.9% Source: Adapted from Ouerghi and Heaps (1993). Based on consumption of about 6 kg per household per day. The assuned wage rate for male time is 4 rs per hour. For female and child time the wage is assumed to be 2 rs per hour. Collected wood is valued at .80 rs/kg (as opposed to .98 rs/kg the average market price) to account for lower value of collected versus purchased wood. The median total value of expenditures per household (including imputations for housing and firewood and other non-marketed production) in rural areas in Pakistan was 31,427 rs per year"2. This implies that the imputed value of firewood is 6.2 percent of total typical household expenditures. This aggregate figure of course masks great variations, as in cases where firewood is scarcer or more distant it obviously consumes a larger fraction of income. For instance in Baluchistan, where distance to firewood is more than twice the national average, the value of 12 Since the average exchange rate was 23.8 rs/$ in 1991 this is US$1,320 per household and since average household size is about 7, the median per person expenditure in rural areas in dollars would be US$189. 8 collected firewood is almost 8 percent of total housold expnditures. Obvudly her are also large fluctations within thee large regions. Is firewood coUectedfrom open access or common property sources? A crucial element of the vicious circle hypothesis is the divergence of private and public costs of the collection activity, that is, parents are able to push some of the cost of their children's activities onto the local environment. Table 2 shows the reported sources from which wood was collected by nual residents by province and then separates these out, for each geographic area, the sources of collection for the poorest 50 percent of households. Most firewood is collected from households' own land and from "other private' land. Overall in Pakistan, state/forest land and common/village land account for only a quarter of firewood co11ection. This is higher in NWFP and much higher in remote and sparsely populated Baluchistan. This prevalence of firewood collection on private land highlights several points. First, only a small fraction of household firewood collection consists of felling trees. While purchased firewood tends to be large branches and stems, collected firewood is more likely to be small twigs, shrubs, leaves and roots (Ouerghi and Heaps, 1993). This composition of the types of firewood and the prevalence of collection of firewood from 'own land" reveals the multiple uses of individually owned trees, and trees and shnrbs that may be present on the land for other reasons (e.g. erosion control, shade provision, boundary demarcation) have the secondary benefit of providing firewood. 9 Table 2: Source of collected wood among rural households which collect wood, by province. __________ j ~~~~~~~~~~~~~~~~~~~(3)4(4) OwnLad Other Stae Waste Common / Oder Sub-Da: private forest land village land non-private __ __ ~~~~~~land_______ All mral households Pakistan 26.0 34.1 12.3 O.S 13.3 13.7 39.8 Punjab 35.8 33.7 7.7 0.8 8.4 13.8 30.7 Sindh 22.7 46.5 17.9 0.4 9.6 13.5 41.4 NWFP 35.9 19.7 15.3 0.0 18.2 10.8 44.3 BaluchLMs 3.3 6.6 10.9 0.0 56.0 23.1 90.0 Households in pB b 50 percntie of ex es per person Pakistin 21.2 37.5 12.1 0.5 13.6 15.2 41.4 26.6 41.8 7.6 0.9 9.1 14.0 31.6 Sindh 9.8 45.3 18.8 0.4 13.3 12.4 44.9 NWVFP 31.9 24.1 14.7 0.0 14.7 14.7 44.1 Baluchistan 2.0 3.9 15.7 0.0 41.2 37.3 94.2 Source: PIHS Second, a significant fraction of firewood collected off of "other private land" suggests that, at least in many areas of Pakistan, the value of firewood has not reached the point where exclusion or charges for collection activity, even on private land, are profitable. This brings us back to an issue discussed in the opening section: soil, forests, water, or grazing land are not intrinsically private property (at one extreme) or open access (at the other). The elements of ownership of resources (use, transfer rights and residual claims) depend not only on background social and legal conventions but also on the ratio of the value of the commodity under one regime or another. Even with constant prices for enforcement, as a resource becomes more valuable 10 there may be an endogenous switch from a regime of open access to more limited access (for example, restricted common property or purely private property). One important determining factor of this regime change will be population density itself. All else equal, as population density (and the density of economic activity) rises, the value of land rises which increases the likelihood of a switch to private property. As discussed above, however, increasing scarcity of private property natural resources can be expected to have a very different impact on fertility decisions than open access natural resources: the household cannot escape the full natural resource consequences of its fertility decisions13. This combination of differential household behavior with respect to common versus private property treatment of natural resources, and the endogeneity of the socially enforced property rights over certain resources will make for interesting dynamic behavior of a fully articulated system. It is possible that at stages of low population density initial increases in population density will lead to greater demand for children. This greater population density will in turn accelerate environmental degradation and raise the scarcity value of the open access resource. As long as the resource remains open access this will (or empirically may) in turn increase the demand for children. After a certain point, the increased population density may change the value of the resource such that it switches from open access to more limited access, such as private property. Once a natural resource is privatized, the negative externality to child bearing disappears and is internalized by the household. From that point on, households would respond to furter deterioration in availability by reducing their demand for children (unless children have 13 The differing views on the impact of populaion density on agricultural institutions and practices are well described in Turner, Hayden, and Kates (1993). 11 a comparative advantage over adults in the market supply from private prope which is unlikely). Recent studies (Ouerghi 1993, Leach 1993) on tie market for firewood in Paktn argue that something very much like the switch to a private property regime may have already occurred or is about to occur in some, and perhaps most, regions of Pakistan. There is now a large market for firewood and most of the new tree plantings are on large plantation style lots. These plantation markets supply firewood indirectly from twigs and branches cut (as the main market for plantation trees is construction poles or other uses). Figure 1 suggests something of the time path of the market for firewood in Pakistan with an ongoing, and perhaps well advanced in many regions, evolution from plentiful open access forest resources to scarce, market provided firewood. 12 F1%1 From foet ngemen to tree planting - otosdictated bypoutonadeari Forst or tree cover through time Vberable, open. > e~~~~ccess KreeorCe. 1aIrnastbiaa ean speed or slowa shift fro 3 to S. Piaty of forest\ few people so little or no coafilet over the resource. - Commrnity management systems arc found whee there is same resource scarcity, but wbere the resource is still seen as defensible externaly Once there are too mayn people and ausmgeable internly, or too many interests for the aim mnch open-access laud using local authority of the commrunal resource. coats becomes &rm-land if structures, f fmanagement tenure Is ofhed. wil begin to outweigh bneftts. Trees are plated and a Rukrs for and active local management wll noew resource ceated. owneenhip and use grdually cease. Reserves. If any, are found only near wiUl have to be protected the the boundary; by the state. or relate only to a Rapid deforestation will then take few high-valued phce gathered products Resources may remain in balance for many years. Increase in numbers of people through time Stage I Stage 2 | Stsge 3 Stage 4 Stage 5 Source: Chart devised by Gil Shepberd, ODI, autumn 1992/spring 1993. in OnerghL 1993 Are children relatvely devoted to collection activities? In the PIHS data there are two types of time allocation data. All individuals aged 10 and over were asked about their "labor' time devoted to a variety of tasks, either wage labor or the production of (potentially) marketable commodities. In addition, all females (but only females) aged 10 and over were asked about their participation in thirteen household production tasks. 13 Tabb 3: Allocato of tmin housold activiies in hours pr week (ad by fracton of total m) m slcted bas for women ad giuls, by urbbAn-na resWosuc Urban Girls Urban Rural Girls Rrl Women Ae: 10 Ae: 25 Women Fetcing waor 0.3 0.4 1.6 1.8 1.5 2.7 (1.9%) (1.3%) (7.7%) (5.1%) (11.4%) (4.4%) FirwOod 0.0 0.1 1.2 1.3 1.5 1.4 collction (0.2%) (0.2%) (5.9%) (3.7%) (11.2%) (2.4%) Annal care, 0.2 0.6 1.9 3.8 1.6 3.9 fodder (1.6%) (1.9%) (9.0%) (10.3%) (12.1%) (8.2%) Dungcake 0.1 0.2 0.5 1.1 0.4 1.1 prepaation (0.4%) (0.6%) (2.5%) (3.0%) (3.3%) (2.2%) Sub-total: 0.6 1.3 5.2 8.1 5.0 9.1 Colaecon tks (4.1%) (3.9%) (25.1%) (22.2%) (38.1%) (17.2%) Meals to wores 0.0 0.0 0.2 0.5 0.2 0.4 (0.0%) (0.0%) (0.9%) (1.4%) (1.4%) (0.8%) Grindw flour 0.0 0.1 0.1 0.4 0.0 0.6 (0.2%) (0.2%) (0.5%) (1.0%) (0.2%) (0.8%) Going to market 0.2 1.0 0.1 0.4 0.1 0.3 (1.3%) (3.2%) (0.5%) (1.1%) (0.8%) (0.6%) Stitchin 0.6 1.1 0.7 0.9 0.2 1.4 (4.1%) (3.3%) (3.6%) (2.6%) (1.4%) (3.0%) Milk, ghee 0.0 0.1 0.2 1.3 0.0 1.0 PIearadon (0. I %) (0.4%) (1.0%) (3.6%) (0.2%) (2.2%) Sub total: Other 0.9 2.3 1.3 3.5 0.5 3.6 tasks (5.6%) (7.2%) (6.4%) (9.7%) (4.1%) (7.4%) Cooking 6.9 14.2 7.3 12.3 2.6 15.9 (45.3%) (44.4%) (35.1%) (33.8%) (20.2%) (33-6%) Caesning 5.9 8.5 5.8 7.4 3.8 10.4 (38.6%) (26.4%) (28.1%) (20.2%) (29.1%) (22.0%) Cbild care 1.0 5.8 1.1 5.1 1.1 9.9 (6.3%) (18.8%) (5.3%) (14.1%) (8.5%) (19.8%) Total 15.3 32.0 20.7 36.5 13.0 48.9 (100%) (100%) (100%) (100%) (100%) (100%) Notes: Figures in paretes amre percent of ime devoted to activity. 'Girls" in this table refers to unmauried female household members less than 18 years old. Source: PIHS 14 Table 3 shows the allocation of time to each of the thirteen household tasks for women and for girls'4. On average, women in Pakistan spend 35 hours per week in household activities, while girls spend 18 hours per week on these tasks. The three most time consuming tasks for women are cooking (38 percent), cleaning (23 percent) and child care (16 percent) which together account for 77 percent of time devoted to household tasks. We classify fetching water, collecting firewood, collecting fodder for animals, and dung cake preparation as those 'collection" activities which involve natural resources. Collection activities account for 14 percent of time for women on average and 17 percent for girls. Not surprisingly, these activities are much more important both absolutely and as a fraction of time for rural women. Urban women and girls spend only 1.3 and 0.6 hours per week on collection activities, compared to 8.1 hours and 5.2 hours per week for rural women and girls. Rural women work 4 hours more per week on household tasks, hence the difference in collection activities is by itself greater than the total amount of labor time difference. What types of work do children do? As suggested as part of the vicious circle hypotheses, the percentage of time allocated to collection activities is much higher at young ages. The final two columns of table 3 time allocation of rural females at ages 10 and 25. Whereas 10 year olds devote 38 percent of their time to collection activities which falls to 17 percent by age 25. Figure 2 shows the steady decline in this percentage of time in collection activities by age. 4 In this discussion, "girls' refers to unmarried females less than 18 years old. 15 Figure 2 Share oftime in household activities devoted to "collection activities" Rural females apd 10 to 25 40 30- 1 5 - - - - - - - - - - - - - - - - - - - - - - - 10 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Age This relatively high frction of child tme devoted to collection activities is consistent with a number of studies of time allocation. Table 4 shows the results of four other studies of time allocation in Asian countries for both the total time devoted to household activities and the fraction of that time devoted to collection activity (appendix 2 reports more detail of these other studies). 16 Tabe 4: Afocadon of time by children to collection activities from otber studies Country Total tume to HH acdvides Fracuon of time to collection advities (hours per week) (percent) Woman Child Woman Child This study Average 36.5 20.7 22.2 25. 1 This Study 75th percentile 52.5 29.5 34.5 41.9 Palstan (Alderman and Cbishti) 32.0 16.4 Javanese village 32.8 20.9 (E) 3.5 22.9 (F) Nepalese village (A) 44.3 33.1 (F) 36.1 56.3 (F) Nepalese villages (B) 45.6 20.5 (all) 26.5 68 (all) Nepalese villages (C) 56.0 13.5 (all) 38.9 68.6 (all) Source and notes: Pakistan. based on Alderman and Chishti (1991) table 2 based on time allocation for all rural females over 10, with same classification of collection activities as our study. Javanese village and Nepalese village (A) are based on Nag, White and Peet (1980) table 10.2 and 10.4, based on femde time use. Children are ages 6 to 14. Collection activities are firewood coUection and animal care. Nepalese Villages (B) is based on Acharya ad Bennett (1981) based on five hill and one tarai village, coildren are 6 to 15, collection activities are fuelwood collection, water collection. leaf fodder colection. Nepalese villages (C) is based on Kumar and Hotchkiss (1988) based on three hill districts, children are 6-15, collection activities are collection of fuelwood, water, grass, and leaf fodder and the assunption that half of grazing is done by children. Nepalese B and C are based on timne use of male and female children. Time use per week based on six day work week. Does the presence of children alter adult's time use? Although the share of time devoted to collection activities is decreasing with age, we now turn to how much total time is spent on collection activities by children. How important is collection done by children as a fraction of total household time spent on various household tasks? Table 3 showed that the share of time devoted to collection activities falls from 38 percent to 17 percent between the ages of 10 and 25. Figure 3 shows the hours of work in household tasks per week by age for females from age 10 (the lowest age with this information in the survey) to age 65. This rises sharply from 10 hours per week at age ten to 24 hours per week by age fifteen, to 44 hours per week by age twenty-five. Total time devoted to household activities rises sufficiently with age that the absolute amount of 17 time in collection activities rises substanially, from 5.0 hours for a ten year old to 9.1 hours for a women aged 25. Figre 3 Hours per week spent by females in household activities, by age so - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - 20 - - co - - - -o - - - - - - - - - - ,, ,, - - - - -5 50 Age In table 5 we shift from the individual level, how much work each person does, to the household level. How much total time is spent by all members of the household, what is its allocation across tasks, and how much of each task is done by children? The total female time per household in household tasks in rural areas is 78 hours a week. Since from figure 3 the average time per wornan peaks at around 50 hours per week (at ages 30 to 35) it is obvious that major time contributions are made by other household members. Of the total time devoted to household tasks in mr-al areas, 18 hours a week, or 2.5 hours per day per household, are devoted 18 to collecdon acdvities, although this total varies substantially within Pakistan. Only about 17 percent of female time devoted to household tasks is contributed by children. Table 5: Total female time devoted to household tasks in Pakistan in rural areas, by type of activity, and the fraction which is done by (female) children. Total female household Share of total time whuch is performed by hours devoted to task devoted to task chidren Fetching water 4.4 5.6% 20.3% Firewood Collection 3.2 4.1% 19.5% Animal care, fodder 7.9 10.1% 13.7% Dung Cake Preparation 2.3 2.9% 13.9% S-t(tal cnilectign 17 a 22 -8 a _ __E Meals to Workers 1 1.3% 13.4% Grinding Flour 0.7 0.9% 10.9% Going to Market 0.8 1.0% 7.5% Stitching 2.2 2.8% 20.3% Milk, Ghee preparation 2.4 3.1% 5.2% = Sub-total other _ 7,1 9.1 % _,______ _ Cooking 26.6 34.1% 17.2% Cleaning 16.9 21.6% 21.6% Child Care 9.7 12.4% 7.1% Total 78.1 100.0% 16.6% N I2254 2254 2254 Note: Fraction of time in task which is performed by children is only for households in which the task is performned. Source: PIHS Some activities, particularly fetching water (20.3 percent), collecting firewood (19.5 percent), and cleaning (21.6 percent) appear to be "child" tasks. Others, such as animal fodder collection, are proportionately less performed by female children. Perhaps surprisingly, given other work which suggests a large degree of substitutability of mother's and older daughters' time 19 in caring for young children, child care appears not to be a child activity. Children do contribute a significant fraction of all hours spent in household tasks. The impact of this on mother's time and total hours by age is examined below. Bringing the various pieces of information together, we find that table 1 reports that roughly 60 percent of the time devoted to firewood collection comes from women and children, while table 5 suggests that only 20 percent of female time to firewood collection comes from children. The composition of members of trips to collect firewood in table 6 suggests that children account for about 15 percent of the people who go to collect firewood. Depending on the region, there are nearly equal numbers of adult males and females. The number of children varies from only 9 percent in Punjab and Baluchistan to 25 percent for the poorer households in Sindh and NWFP. Together the evidence suggests that an important, but not overwhelming part of all firewood collection is done by children. 20 Table 6: Composition of most recent trip for wood collection, in rurl areas among houeholds dmt ue firwood _____________ Percent which were: Number of Quliren (male servants and people Adult Males Adult Females and female) odhers All Pakistan 2.24 47.41 34.85 14.23 3.51 Punjab 2.16 50.54 37.09 9.83 2.54 Sindh 2.31 42.76 33.18 20.54 3.52 NWFP 2.33 49.63 28.34 16.50 5.53 Baluchistan 2.13 43.89 41.54 8.69 5.88 Poorest 50th household per person expenditure percentil _ All Pakistan 2.23 44.38 37.60 16.57 1.45 Punjab 2.26 47.03 40.36 12.01 0.60 Sindh 2.22 36.26 36.33 25.16 2.2S NWFP 2.35 45.26 32.55 17.97 4.22 Baluchistan 2.29 52.65 37.76 8.49 1.10 Source: PIHS Do rural families alter their firewood use and collection activities in response to the presence of children? Table 7 presents the association of household composition and the number of hours devoted by mothers to various tasks. 5 The fact that time use cannot be negative implies the impact of children has a particular non-linear structure, that is, time in each task could be higher or lower either because it equaled zero and now positive or because it was positive and 15 It should perhaps be emphasized here that these estimates-and those subsequent equations of the relationship between household composition and time devoted to tasks-do not correspond to the experinent of exogenously adding an individual of a particular age to the household. Such an interpretation is inconsistent with the endogenous demand for children (taken up in subsequent sections) being determined by their value from carrying out household tasks. 21 now has changed."6 Because these estimates are a combination of changes at the limit (from zero to positive hours), and changes above the limit, the proportion of the average total change due to the response above the limit is also reported in the table (as derived in McDonald and Moffitt, 1980). These results indicate that the presence of children under 10 is associated with increases in a rural mother's time in household activities, whereas each child over ten is associated with less time in household activities. The point estimates from table 7 indicate that by age 16 a female child has a been a net contributor to hours in household activities, with the positive contributions of labor from 10 to 15 offsetting the additions to mother's time from years 0 to 10.17 Table 7 also presents estimates of the relationship between household composition variables and total time devoted female time to each of several activities. These results suggest that an additional child in the household between 10 and 15 is associated with an additional 5 hours per week devoted to all household activities by all members of the household. But females 10 to 15 years old devote on average 17 hours to household activities. The net contribution to the household is 12 hours on average. Mother's time is reduced only 2.6 hours. This difference is 16 Tobit estimation is used because a significant number of women devote zero hours to some of the tasks (i.e. the dependent variable is censored at zero). If the underlying (non-censored) model is H*=Xp+e, then the change in observed (censored) H due to a change in one of the Xs, 6H1 bXi, is equal to PiF(z), where fPi is the Tobit parameter estimate, F() is the standard normal cumulative distribution function, and z is equal to Xp/a where a is the standard deviation of e. McDonald and Moffitt (1980) recommend evaluating this at the mean of the Xs. These are the average derivatives implied by tobit estimation of the relationship between mothers time devoted to various household tasks and household composition in rural areas " With discounting, this does not necessarily make a child a positive payoff, even in strictly time terms. Assuming a discount rate of 3 percent, (which is low for rural households in Pakistan) a female child staying until age 18 would pay off in mother's time terms. 22 Table 7: Mean change in expected hours per week per woman in household activities, rural areas Mothers | Sum for all household females over 10 years old Dependent Total Child care Cooking Collec- Toua Child care Cooking Collection variable in hours and tion and activities per week cleaning activities _ cleaning Household composition variables: Number of household members above 15 -3.219 -0.701 -2.103 -0.412 5.973 -0.037 2.738 1.139 (<0.001)* (<0.001)* (<0.001)* (0.001)* (<0.001)* (0.731) (<0.001)* (<0.001)* between 10 and -2.611 -0.392 -1.652 -0.463 5.249 -0.325 4.165 0.912 15 (<0.001)* (0.023)* (<0.001)# (0.059) (<0.001)* (0.161) ('<0.001)* (0.020)* between 5 and 10 0.974 0.646 0.147 0.122 5.146 1.325 1.707 1.432 (0.038)* (<0.001)* (0.635) (0.533) (<0.001)* (<0.001)* (<0.001)* (