- Email: [email protected]
Household strategies to cope with the economic costs of illness
Pergmon
0277-9536(95)000375-4
Soc. Sci. Med. Vol.43, No. 3, pp. 291-301, 1996 Copyright © 1996ElsevierS~ace Ltd Printedin Great Britmn.All rights reserved 0277-9536/96$15.00+ 0.00
HOUSEHOLD STRATEGIES TO COPE WITH THE ECONOMIC COSTS OF ILLNESS R. SAUERBORN)* A. ADAMS 2 and M. HIEN 3 ~Harvard Institute for International Development, 1 Eliot Street, Cambridge, MA 02138, U.S.A., 2Harvard Center for Population and Development Studies, and 3Ministry of Health, Ouadougou, Burkina Faso Ab~'aet--The authors examine the strategies rural households in Burkina Faso used to cope with the costs of illnessin order to avert negative effects for household production and assets. They use information from 51 qualitative interviews, a household time allocation study and a household survey. Both surveys use the same sample of n = 566 households. The authors analyze these strategies along four dimensions: the type of behavior, the sequence in which strategies employed, the level at which strategies are negotiated, i.e. the household level,the non-household extended kin level or the community level, and finally the success of strategies in protecting household production and assets. A taxonomy of I 1 distinct types of coping behavior is developed which have the effect of either avoiding costs by 'ignoring'disease, or of minimizing the impact of costs on the household once illness is perceived. Intra-household labor substitution was the main strategy to compensate for any labor lost to illness. However, labor substitution did not eliminate production losses in the majority of households struck with severe illness of a productive member. Only wealthy household were able to fully compensate labor losses by hiring labor or by investing in equipment to enhance productivity. Sales of livestock was the main strategy to cope with the financial costs of health care. None of the households studied fell into calamity. However, the households' ability to avert the loss of production and/or assets was very varied and depended on household size, composition and assets, on the type and duration of illness and on clustering of crises (e.g. several repetitive or simultaneous illnessesor concurrent seasonal stress). Coping with the costs of illness largely occurred at the level of the household. Inter-household transfers of resources played only a small role. The authors develop the concept of risk households and suggest several policies with the potential to strengthen the ability of households to cope with the economic costs of illness. Copyright © 1996Elsevier Science Ltd. Key words--illness, costs, household, coping, production loss, utilization of health care
l. INTRODUCTION 1.1. The problem
The household production of health has recently received much attention from both the academic and policy communities [1]. With it, a new interest in the costs of illness for the household has emerged. A recent paper [2] reviewed the growing number of studies on the economic costs of illness. Only four of the 12 studies reviewed dealt with the time costs of illness and only two estimated intra-household transfers. Using the case of Burkina Faso, the authors concluded that the household, rather than the individual ill person, was the appropriate unit of analysis, given that the large majority of time and financial costs of illness were borne by healthy household members. Study findings indicated that average annual time costs of caring for the sick were considerably greater than the financial costs of care (21,700 and 8700 F CFA respeetivelyt) which on their own represent 6.2% of total household expenditures. *Author for correspondence tConversion rate at the time of the field study: I U.S.$ ffi 265 F CFA.
These attempts to describe the household costs of illness, however, beg several questions. With respect to financial costs, does relating reported health care expenditures to total household expenditures reflect the real burden to the household? Do other households transfer resources---both in cash and in kind--to support the household struck by illness7 Therefore, isn't the comparison of health care expenditures with total household expenditures overestimating the actual burden of illness costs for the household? Furthermore, what can we learn from risk-sharing arrangements within the community? If indeed a system of mutual transfers exists to cover health care expenditures, what implications does this have for policy makers struggling to introduce prepayment or insurance schemes? With respect to time costs, Nur [3] and Weisbrod [4] have argued that intra- and inter-household labor substitution occurs to counter potential production losses due to illness. If this observation held for the Burkina study population, the economic impact of illness of productive household members on the household economy would be much less than customary economic analysis indicates. This finding might
291
292
Rainer Sauerborn et al.
also undermine the economic argument for investing in health as a means of preventing production loss in a labor intensive economy. The research questions guiding the present paper are as follows: (i) What strategies do households use to cope with the financial and time costs of illness? (ii) Are these coping strategies effective in preventing or managing illness-related costs borne by households? (iii) Are there differences in the ability of households to cope with the time costs compared to the financial costs of illness? (iv) What policy conclusions can we draw from these micro-level observations? I.Z Conceptual framework
Coping is defined as a short-term strategy adopted within the prevailing value system to avert a negative effect on the actor [5, 6]. In the context of this paper, the actor is the household defined as a group of individuals--most commonly but not necessarily linked by kinship ties--who live together and share functions of production and consumption as well as of reproduction [7]. In the context of this paper, the negative effect coping strategies seek to avert is the breakdown of the household as an economic and social entity (referred to in this paper as 'calamity'). As in famine research [8], the criteria for calamity are household destitution and disintegration. In this paper, coping strategies to deal with the economic costs of illness are analyzed along four dimensions following the approach used by Chen [9]: • The type of the coping behavior is described. • The assessment of the level at which support is negotiated. Coping may be confined to the household in 7qhich illness occurs, or it may involve extended kin or non-kin related households. Finally, the community as a whole or groups within it may be a source of support. • The sequence of different strategies. Here we are interested in whether and how strategies are pursued concomitantly or sequentially. • The success of coping behavior. Three economic criteria are used to judge whether household coping is successful: First, did coping avert calamity--i.e, the social and economic breakdown of the household? Second, did it reduce or avert any negative effect of illness on household production? Third, did coping strategies preserve household assets? These four dimensions will be explored separately for strategies to cope with financial costs and those to cope with time costs.
2. METHODS
This particular analysis is part of a larger study carried out in the Kossi province of Burkina Faso which has been described elsewhere [10]. Both qualitative and quantitative methods were used in order to enhance the validity of results through triangulation [11, 12]. 2.1. Qualitative methods
Fifty-one qualitative interviews were carried out in two villages, S~riba and Bourasso. The households were selected using a theoretical sampling approach which sought to include households representing a wide variety of characteristics deemed to influence coping behavior [13] which comprise • Household size and composition. Here it is assumed that coping strategies are more effective in large households with many productive members who can substitute or compensate for work lost to illness. • Household wealth. It is hypothesized that wealthy households have a greater arsenal of coping strategies than poorer households. • Illnesses of differing degrees of perceived severity. Here, coping responses to mild/moderate and severe/fatal illnesses are compared. • Sick individuals of different age and gender. Comparing coping behavior in response to illness among productive and unproductive age groups as well as between genders, it is assumed that the more an individual contributes to household production, the more resources the household allocates to treatment [14]. • Seasonality. Finally, the sample should capture illness episodes occurring both in the dry and rainy seasons. Here it is assumed that the wide variations in the availability of household resources, in terms of time, cash and assets [15] shapes both the type and the effectiveness of coping strategies. Thirty illness episodes were analyzed through interviews with the household head, the sick individual or, in case of sick children, with their main care-giver. The households in which these 30 illness episodes occurred were selected from the sample of the household survey (n = 566 households, see below). This enabled us to select fatal illness episodes and nonfatal ones which had been perceived as severe or mild in the household surveys. Since the same households were revisited for the survey in a monthly rhythm throughout the year, we were able to select households who recorded illness episodes that had occurred in the rainy and those reporting dry season episodes. In addition, 21 'key informants' comprising village midwives, village heads, religious leaders and members of village organizations, such as the Young Farmer Association of Bourasso, were interviewed to further explore the determinants of household coping strategies. Interviews were translated from three local
293
Strategies to cope with illness costs Table 1. Seasonal variations in the economic costs of illness by type of costs Household cost Household cost Household cost per month: per month: per year: dry season rainy season all seasons Type of cost F CFA % F CFA % F CFA % Time costs of illness 1395 34.9 604 56.7 11,998 39.5 Time costs of caretaker 1356 33.9 263 24.7 9717 32.0 Financial costs of care 1250 31.2 197 18.5 8686 28.6 Total economic costs 4002 100.0 1065 100.0 30,401 100.0 Source: Household surveys no. 1 (dry season) and no. 3 (rainy season). Costs per year are calculated on the basis of a length of 6 months for both the dry (December through May) and rainy seasons (June through November). Table 2. Type and frequency of strategies to cope with the household costs of illness and death Time costs Financial costs Coping strategy No. of cases Coping strategy No. of cases Intra-household substitution 24/27 Mobilize cash/savings 16/25 Change in capital-labor mix 2/27 Asset sales (animals) 13/25 Hire labor 3/27 Loan 4/25 Free labor 2/27 Income diversification 5/25 Wage labor 1/25 Free care 2/25 Gift
15/25
Change in illness perception Change in health care choice languages into French, and coded using the software package " E t h n o g r a p h " . The coded interview transcripts were the basis o f conceptual analysis [16, 17]. In addition, seven household case studies permitted contextual analysis of factors influencing the decision to incur costs and how to cope with them [17]. A m o r e detailed description o f the qualitative method used in this study is provided by Sanerborn [18].
2.2. Quantitative methods In addition, this paper uses survey data on a sample of 566 households. The same households were visited six times between March and October 1992, spanning both dry and rainy seasons. Information was gathered on past perceived illnesses (recall: 1 month), on healer choice, on health care expenditures as well as on days lost to illness both by the sick and by healthy household members tending to the sick and/or accompanying them to a distant treatment site. Time costs were calculated by converting days o f f w o r k * due to illness into monetary terms using the current wage rate for day laborers [15] to provide a proxy for the opportunity cost o f time. In addition, two time allocation studies based on the recall of activities o f the previous day were carried out on the
*The monetary value of a day lost was equated with the average wage a household was stated to be willing to pay for day laborers during the month of the interview [15]. No attempt was made to value leisure time and economically active time separately. tThe terms "rainy" and "dry" season are not used in the climatological sense as having "rains" and "no rains", but rather in economical terms: the rainy season is the time of agricultural production, whereas the dry season is slack time, used for repairs, house building, marketing, crafts, and ceremonies. :~Details of the cost analysis have been presented elsewhere [2, 15].
same household sample during the dry and rainyt seasons respectively. Further details on the sample, the source population, the questionnaires and field procedures are described in a previous paper [2]. 3. RESULTS Table 1 summarizes the main components of household costs o f illness.:[: In the context o f the present analysis, the key findings are threefold: (i) time costs represent more than two thirds o f the total household costs o f illness; (ii) time costs of healthy household members are about as large as those incurred by the sick member, and (iii) the average annual financial costs amount to 6.2% of total annual household expenditures. This chapter describes how households coped with these costs. The analysis will evolve along the four dimensions of coping strategies described above: type, sequence, level and success.
3.1. Types of coping strategies Our taxonomy of coping behavior comprises eleven strategies: seven addressing the financial costs and four addressing the time costs of illness (see Table 2). 3.1.1. Coping with Financial Costs. In 25 of the 30 illness episodes studied in the qualitative sample, households incurred financial costs. The following analysis uses these 25 cases as the denominator.
(i) Using cash and mobilizing savings As a first step, most households used any available cash or savings to pay for health care expenditures. The amount, however, was generally insufficient: in only four out of 25 cases were illness expenditures completely covered. Cash sources included agricul-
Rainer Sauerborn et al.
294
tural and craft production and migrant remittances. In addition, two households derived cash from millet beer sales, and three from the pension income of members who were veterans of the French Army
(anciens combattants). (iO Sale of assets The sale of assets was the second most common way to meet health care expenditures. For households who did not possess sufficient cash, the sale of assets in the form of livestock was a widespread response to crises of many kinds [8, 19-21]. Indeed, in this part of Africa, animals were perceived as 'ambulatory savings-banks'. The sale of cereals, on the other hand, was considered a last resort You know you raise animals with the aim in mind to face this kind of expenditure, but as to the millet, that's very different because even if you sell the millet to pay for health care, after the cure, what are you going to eat? Villagers emphasized the dangers of selling cereal to overall food security, and stressed the need to avoid adverse seasonal fluctuations in food prices. During the dry season in which the majority of health expenditures occur, cereal prices are substantially lower than in the rainy season. As one farmer stated, selling millet in this season to cover health care costs was foolhardy as it might necessitate having to buy millet in the rainy season at a much higher price But if you have animals, that's always better to sell than millet, because if you sell millet now, you will be forced to buy back the millet at double the price later. For an expense in the order of the price of an animal will leave a big hole in your granary. Only one household head, who had neither cash nor animals, was forced to sell millet as a last resort in order to feed his newborn with formula milk after the mother had died in childbirth.
(iiO Loans In both villages of the qualitative sample it was customary to take loans. In S~riba no interest was claimed, while in Bourasso the going rate was 10% (except for loans to kin, which were interest free). The option to take loans, however, was only available to wealthy households with collateral (generally in the form of animals) Not everybody can get a loan. You have to have a guarantY or you must have someone who can guarantee for you .... By all means, the poor can't get a credit. In general, loans were perceived as a buffer between the time of need for cash and the time when the
household was in a position to pay back. Given the widespread belief that selling livestock under pressure would lead to a bad price, it was considered more advantageous, even for the wealthy, to take a loan, and delay repayment until livestock prices were more favorable. Another less common way to receive small loans was for the loanee to offer his labor to work on the fields of the creditor.
(iv) Income diversification Men and women from poor households used slack time in the dry season to generate additional income. Men typically enjoyed more leisure time during the dry season (3.4 hours per day compared to 2.0 hours per day for women*). This leisure time could be used to generate additional revenue to cover extraordinary expenditures, such as for health care. Household members engaged in a wide variety of income generating activities ranging from fetching firewood for millet beer breweries, building fences, weaving straw mats and honeycombs, and tailoring.
(v) Wage-labor In this subsistence farming environment, working as a wage-laborer was considered a last resort for those lacking assets, access to credits and kin/community support. In the rainy season, the trade-off between working one's own field to ensure household livelihood and selling one's labor to generate money for health care expenditures was especially stark ... one is obliged to leave one's own field in order to go to cultivate for someone else in order to be able to meet the needs of one's own family, this is the difference.
(iv) Free care Both the formal health services and traditional healers provided services at no cost (in 3 out of 25 cases). However, free care was not received by the needy, but rather by wealthy and influential households.
(rio tiffs In 14 out of 25 illness events, households received gifts from extended kint. The large majority of households who received gifts were wealthy. Only one poor household received a 5000 F CFA gift by an expatriate health worker. 3.1.2. Coping with Time Costs. In 27 out of 30 illness episodes, the households concerned lost time* due to illness. The analysis of strategies to cope with time costs was based on these 27 cases.
(i) Intra-household labor substitution *Recall covered the time from 7 a.m. to 9 p.m. tWealth was defined by the villagers themselves based on Grandin's [22] technique of wealth ranking among smallholder communities. ~Only full days lost for work due to illness were considered. §Household SER:I8 lost production of five doors to a competitor worth 12,500 F. Household SER:23 was not able to produce eight mats for a loss of 2400 F. Both craftsmen felt it was impossible to recover these losses.
The reallocation of tasks among household members was the most frequently chosen strategy to cope with anticipated production losses. In 24 out of 27 illness episodes, households sought to reallocate the home production tasks of the ill member among the healthy members. In the two cases§ where craftsmen suffered from
295
Strategies to cope with illness costs illness that prevented them from working, the household was not able to substitute their specialized skills. In the case of lost field production, household members who had not participated in agriculture before the illness event were mobilized. Children less than 10 years old, those who had retired from field work or who participated in other activities were called to the fields. Another strategy was to switch the sick person's tasks from physically demanding field work to craft production. In household SER=01*, the sick person took to weaving, whereas in BO_U-06, the chronically ill household head started to sew clothes when he could no longer work in the fields. It is intuitively obvious that the ability to reallocate the tasks of ill or deceased members depends on household size and composition. The monogamous nuclear family household BOU_-05 in Bourasso, for example, had no capacity to 'recruit' a new member into the field labor force. The polygamous 23-member household SER-03, on the other hand, was able to compensate for the production of a deceased household member by mobilizing the labor of three individuals.
(i0 Changing the capital-labor mix of production Two households responded to imminent longterm production loss due to illness by investing in capital, thus changing the capital/labor mix. One such household introduced oxen-plowing on its millet and sorghum fields to enhance productivity. Another way to change the capital-labor mix of agricultural production was to change the crop mix from a less capital-intensive one, namely millet, to one that requires more capital input (pesticide, fertilizer) and less labor for each unit of output. After the death of one of its main field workers, household BO_U-12 shifted half of its sown area from millet to cotton. The household head was able to do the necessary investments (seeds, fertilizer, pesticides, etc.), since he drew a pension from the French army as a veteran of the Indochina war. We are now growing cotton, which we have not done before, in order to fill the gap. Instead of selling the millet as cash crop, cotton is ideal for this. In only 3 out of 27 illness episodes with production loss, households chose this strategy. This comes as no surprise since additional financial resources, on top of those used for health care expenditures, were needed to invest in additional capital. Both of the above
*The interview ID is composed of the three first digits of the village and the two-digit interview number. 1" Work was not done for credit in the study villages.
households belonged to the upper wealth quartile in the village.
(iii) Hiring labor Some households hired laborers when they anticipated a short-term illness-related production loss during the peak agricultural season, when farmers had little flexibility to substitute or postpone work. However, the expenditures the household had to incur to pay laborerst in addition to paying for health care, were considerable which limited this strategy to households with the ability to generate cash quickly. Only two households in the sample hired labor. In this ease, the household head was a veteran who received a monthly pension of 15,000 CFA, the equivalent of about 60 days of hired labor. He hired 27 young men to work for one day as a substitute for one month of his wife's foregone labor. She had to be operated on for an incarcerated umbilical hernia during harvest time.
(iv) Free community labor In 2 out of 27 illness episodes with production loss, free labor was supplied to the household, in one instance by a community organization (the Young Farmers' Association in Bourasso), in the other instance by a neighboring household in the village. According to accounts of his wives and brother and to hospital records, Philippe Tiawa, a household head from Bourasso, had a bicycle accident with an open head trauma with a cerebrospinal fistula and subsequent meningitis. He was treated from March to June 1991, when he died. During the course of his illness, the household lost his work input as well as the field work of one of his three wives who stayed with him at the hospital. However, the resulting harvest equaled the previous one to which Philippe had contributed. This was due to the fact that the young men from the quarter went several times to cultivate his fields. In summary, intra-household labor substitution was by far the most frequently chosen strategy (in 24 out of 27 events with production loss). Two additional strategies, namely changing the capital-labor mix, and hiring labor required cash resources, which most households did not have. Free labor was only given to two wealthy households. Therefore, only four households in the upper wealth quartile were able to employ strategies other than intra-bousehoid substitution. In addition to poverty as a factor limiting the ability to cope with production loss, household size and composition compromised the availability and effectiveness of intra-household labor substitution. Poor small households are therefore at greatest risk of experiencing illness-related production loss.
296
Rainer Sauerborn et al.
no
I
I use uvlncs
no
~
yes
SUCCESS yes rio
11o
~
"<5.
take loan
n° 1 wage labor
sell animals
( ~.oo-, )
)
.o _ ~
no
yes
accept gift
( CAL~TY ~
( .~:.pt,,t
Fig. 1. Sequence of strategies to cope with financial costs. 3.2. Sequence 3.2.1. Strategies to cope with financial costs. With regard to household responses to financial costs, in 19 out of 25 illness episodes, households chose more than one strategy, and in 10 episodes more than two strategies. When several strategies were used, a clear sequence of coping behavior was observed which is depicted in Fig. 1. Using savings was in all cases the first strategy to cope with financial costs of illness. Yet only in four cases, savings covered illness related costs completely. The key factor influencing the choice of subsequent coping strategies was whether the household possessed any livestock.
*The possession of livestock was perceived as the main source of wealth by villagers [18].
Households possessing livestock had two options for the next strategy: they could either take a loan using their animals as collateral or they could sell their animals, pending on their evaluation of market prices. Loans were generally not available to households without livestock*. For these poor households (27% of the sample), the next coping strategy consisted of wage labor and/or income diversification through crafts. None of the wealthy households had to engage in income diversification or in selling labor. Rather, gifts were negotiated: In 15 cases, households received gifts when struck with illness or death. This option was available to only one poor household. 3.2.2. Strategies to cope with time costs. Household strategies to cope with time costs varied much less than strategies to cope with financial costs. In 23 out
Strategies to cope with illness costs
297
Table3. Levelof support by typeof economiccostand by typeof event(illnessor death) Production Financial loss
Level of support lntra-household Extended kin Nonkin and communitysupport
Illness 19/19 1/19 2/19
loss
Death 6/8 2/8 2/8
Illness Death 16/16 9/9 4/16 9/9 4/16 6/9
Table 4. Differencesin treatmentchoice and expenditureby age group and season Health care utilization(%) Average costs By age group By season Treatment choice (CFA) 0-9 yr />10yr Rainy Dry Total Home care 76 67.2 45.3 64.9 49.0 52 Village health worker 221 5.7 4.4 3.5 4.6 5 Traditionalhealer 656 II.l 19.6 11.7 15.4 17 Nurse, outpatientcare 1432 8.4 12.2 8.5 10.1 11 Hospital care 6939 3.1 10.6 7.4 2.1 8 Other n/a 1.6 8.0 6.7 6.6 7
of 27 illness episodes, intra-household labor substitution was the first strategy used to counter illness related production loss*. In only 5 of these 23 cases, households employed more than one strategy. As mentioned previously, only wealthy households could undertake alternative strategies to labor substitution. No clear sequence of further coping behavior was manifest. Rather, the main principle guiding the choice of coping strategies was whether the anticipated production loss was long-term (change in capital-labor mix) or short-term (hiring labor, free labor).
3.3. Level In-depth qualitative interview data revealed the strong perception that ultimately it was up to the household itself to deal with the financial and time costs of illness. The means to cope are first and foremost those of the household itself. It is only after having exhausted your resources that you can ask other parents to help you. All interviewed households stressed the imperative to explore their own resources to solve their economic problems before requesting or receiving economic support from outside (Table 3). It was only when intra-household support proved insufficient that support from kin was invoked. Indeed, respondents emphasized the advantage of having strong kinship networks on which they could depend in times of need. Kin support--both in cash and kind--was especially strong in cases of catastrophic illness of a productive household member.
*Four households did not use any strategy to counter production loss. tit is difficultto quantify how much production would have been lost in the absence of any coping strategy.
For the 1500F and the 2000 F, it was the entire big family who collected the money. As to the sheep it was me alone. Now, in order to compensate this hole, my brothers, big brothers and younger brothers, sent their children to help me in the fields. Contrary to the perception of the 'tightly knit' African community, support from village groups or non-kin households was rare in the case of ordinary illnesses. Data from the quantitative household survey indicate that only 15.4% of households with a sick person received any transactions from other households. Rather, outside help was highly concentrated in a few cases of catastrophic illness: 38.5% of the total value of illness related transactions took place in only 1% of illnesses.
3.4. Success As to the first criterion laid out above, namely calamity aversion, coping strategies were highly successful: none of the households disintegrated or migrated as a consequence of illness costs. The second criterion concerns the degree to which coping was successful in minimizing adverse effects on household production. Only a small proportion of wealthy households (4 out of 27) were able to maintain household production by substituting labor lost due to illness. Large households were at a clear advantage in replacing labor lost to illness. The majority of households, however, lost production in spite of attempts to reallocate labor internallyt. As to the third success criterion, whether coping behavior jeopardized the household's asset base, the majority (16 out of 25) of households which purchased health care or labor did so by selling livestock, thus reducing their asset base. They emerged from the illness episode poorer and more vulnerable, i.e. with a reduced ability to cope with further stress. Since the same asset buffer was needed for coping with a variety of stresses including
Rainer Sauerbom et al.
298
~
...............~
I
. . . . . . . . . . . . . . . . . . . . . ~"
m
4. DISCUSSION
~
/'/OUS~'~OL~
4.1. The role of the household in coping
II
Fig. 2. Model of household coping with illness costs. Strategies 1-3 aim at cost prevention and strategies 4--5 at cost management. For explanations, see text.
seasonal food insecurity* and other social and economic crises, the seeds for future calamity were sown. Gifts from other households were rarely given to cover treatment costs (5 out of 25), and in only two illness episodes did gifts cover costs sufficiently, such that the household concerned did not have to sell livestock. So far we have looked at household behavior to cope with illness related costs after they had been incurred by the household. In the previous paper, we presented evidence that household members change their perception of illness during the rainy season [15]. In spite of evidence for higher disease transmission and increased case fatality of key diseases, such as malaria during this season, people perceived fewer illnesses, and perceived them as less severe. We developed the hypothesis that the change of the household's economic parameters in the rainy season (i.e. increased opportunity costs of time, lower availability of cash, diminished assets) influenced illness perception. We consider this a strategy (albeit unconscious) of households to avoid costs (see Fig. 2, cost prevention strategies). For the same reasons, households are less inclined to seek any treatment at all, and even those who perceive themselves to be ill tend to keep working. Finally, during the rainy season healthy household members dedicate much less time to their sick kin than during the dry season. As we laid out elsewhere [15], these behavioral changes significantly reduced the economic costs of illness for the household.
*e.g. Animal sales represented the largest source of revenue to buy food during the pre-harvest season, tChayanov [27] pointed out that households at the beginning and the end of their life-cycle have a consumer-toproducer ratio ofl. In the mid-life of the household this ratio changes more and more in favor of consumers. With children gradually entering productive age the dependency ratio goes down again.
While illness occurs in individuals, its costs do not fall on ill individuals alone (see Fig. 2). Indeed, substantial amounts of costs would not have been captured in the analysis, nor could coping behavior have been assessed, without treating the household as the unit of analysis. It was the household as the "therapy managing group" [23, 24] that made decisions regarding health care choice and the allocation of time and financial resources to treatment. Sickness in one member affects the time use of healthy members and influences household decisions regarding the allocation of financial resources. Both the costs of care and strategies to cope with them, therefore, can only be understood in a household framework [1]. Study findings indicate that coping with the costs of illness largely occurred at the level of the household itself, and that inter-household transfers of financial or time resources played a smaller role. Chen's study [9] of rural Indian farmers coping with drought and seasonality, and Adam's [19,20, 25] work on household coping with food insecurity in Mali also emphasized the central role of the household in coping. However, evidence suggests that inter-household transfers are more acceptable in the context of food shortage than in the context of health care [19]. In the present study, both kin and community support (loans, gifts) were generally not available to poor households. The exclusion "~f poor households from inter-household networks of support has been observed in other African settings [19]. In rural Mali, Adams [20] noted that poor households had neither the time, status nor resources to invest in the maintenance of exchange relations, or their cultivation through marriage and reciprocal gift giving. This is in contrast to Das Gupta's [26] observation of strong community support for the poor households in rural India. Not surprisingly, other household characteristics such as size, dependency ratio and wealth had a strong influence on coping behavior. Larger households were able to substitute labor much more easily than small ones. In addition to household size, household age (i.e. the years elapsed since the founding of the household) and life cycle stage may influence its capacity to substitute labor. As the household matures, and children move from being net consumers to net producers, dependency ratios become more favorable, labor more easily substituted and coping strategies more diverse1". The relationship would work through changes in dependency ratios during the lifetime of the household. In spite of the importance of the household in coping with illness, policy has customarily targeted individuala at risk by characteristics such as age or gender. One lesson from the current study is that
Strategies to cope with illness costs there are households at risk of being pushed into poverty and calamity as a resuR of catastrophic illness. The most vulnerable and least able to cope are small and/or asset deprived households. It is conceivable that health centers establish household lists for their catchment areas which target small and poor households for special care [28]. Home visits, for example, might reduce the time access costs of care, waiving user fees for these households might reduce their financial access costs. 4.2. Policy and household coping strategies What policy options might strengthen the capacity of rural households to cope with the financial costs of illness? One suggestion is to develop programs to increase the asset buffer of households, and enable households to generate resources to cover health care costs. Expanding opportunities for animal husbandry, especially of sheep and goats, to the 29% of households who do not own any of these domestic animals, and protecting existing herds through good veterinary care would contribute to this end. A second way of strengthening household coping capacity is through the facilitation of income diversification. Promoting craft production which takes advantage of leisure time in the dry season represents a promising way of reinforcing household coping capacity. The provision of wage employment which is a very popular policy in India [9] may be another strategy to provide households with additional income. Developing rural credit markets and enhancing access to credit for poor households would give them the option to take loans to cover health care. Improving the capacity of households to effectively deal with frequent ailments such as malaria, on their own, and without recourse to expensive professional health care providers, might also lower both time and financial cost of care. For example, households could be encouraged to buy a stock of chloroquine tablets before the rainy season and be appropriately instructed about their use [15]. Schemes to finance health services might be designed to ease the difficulties households have to pay for health care in the rainy season [15]. Drugs and services could be priced seasonally such that higher prices in the dry season would subsidize fees and drugs in the rainy season. Another option would be to introduce prepayment schemes such that payments are collected after the harvest season to cover essential preventive and curative care throughout the year. For a detailed discussion of policies to help households reduce the costs of illness specifically in the rainy season, we refer the reader to our previous paper [15]. *This assumes, of course, that the quality and quantity of the work will remain the same.
299
4.3. Towards a model of household coping While illness occurs in individuals, its costs were largely borne by healthy household members, and to a much smaller extent, by extended kin or the community at large. Households demonstrated an impressive array of coping strategies which had two broad effects: to prevent costs from arising, and to manage costs so that the negative impact on household livelihood was minimized. Figure 2 models the three pathways through which costs can be avoided: (l) by modifying illness perception (the phenomenon of ignoring disease [15]); (2) by continuing to work irrespective of perceived illness (i.e. not assuming the sick role), thereby avoiding potential production loss due to illness*; (3) by allowing illness to go untreated which avoids financial costs. We call these three pathways 'cost prevention'. Should all these mechanisms fail, the household faces time and/or financial costs. We described two 'cost management pathways' the household used to minimize costs: (4) by summoning a range of strategies to minimize production loss; (5) by resorting to coping strategies to cover health care expenditures. 4.4. Does coping matter? Most households in our qualitative sample did lose production due to illness, as they were unable to fully substitute the labor of the sick individual. More importantly, the ability to cope with labor loss decreased with household size and wealth. By reducing the incidence, severity, and duration of disease, effective health services are a major means of maintaining and/or increasing production in settings where labor intensive subsistence agriculture prevails. A population-based intervention trial is currently underway in Burkina to test and quantify this conclusion. However, even when successful, internal labor reallocation is not without costs, as Over et al. [29] have demonstrated. For example, the household may shift labor away from health-maintaining activities such as washing, collecting water, cleaning the compound and cooking. Though this study did not attempt to measure the consequences of reallocating labor for child care, food preparation, leisure and education, it is safe to assume that the 'neglect' of these activities has a negative effect on household welfare in the long run. In contrast to Nur [3] and Weisbrod [4] who reported that illness related labor loss was easily substituted both from within the household and from the community, we found that community support played only a marginal role in compensating for household production lost to illness. With respect to health care expenditures, households showed considerable flexibility and inventiveness in garnering resources. However, even when
300
Rainer Sauerbornet aL
coping was successful in the short run, it was likely to increase vulnerability in the long run by reducing the household's ability to cope with future adverse events [8, 30]. While selling livestock enabled households to cover health care expenditures, it also made them poorer and less prepared to cope with future economic or social crises. Unfortunately, the time frame of the present study did not allow us to analyze the effect of repetitive stress sufficiently. A carefully designed, prospective study is needed to assess simultaneous illness events, food insecurity, seasonal stress as well as other adverse economic or social events (Fig. 2) over several years, and to analyze the priorities and coping behavior of households facing concomitant or repetitive stress. 5. CONCLUSIONS AS to conclusions for future research, the study underlines the need to focus on the entire household, rather than on the sick individual, when studying health seeking behavior and the economic costs of illness. Further research should illuminate the hypothesis that high opportunity costs of time and low cash availability during the rainy season impede illness perception. We would also welcome a longitudinal study of the inter-relationship between different types of economic stress on the household-seasonal food shortages, price fluctuations, expenditures for ceremonies and for health care. As to policy conclusions, it would be helpful to use the concept of 'risk households' in the delivery of health care. A household that is small and poor having experienced recent severe illness or death would be defined as 'at risk' of either ignoring illness (and thus allowing it to exacerbate) or of being unable to seek health care. Health services could target such households for special care, such as regular home visits, accelerated immunizations, and exemptions from fees. Since the household priority is to maintain production and assets rather than to maximize health, policies to increase the ability of households to produce and generate resources would seem appropriate in light of the study results. Such strategies include: improving, diversifying income generating activities, enhancing access to loans for poor households and improving animal husbandry to improve households' asset buffers. Finally, it does not seem realistic to count on community support for health care expenditures. Risk and cost-sharing schemes to finance health care would seem to be promising tools to promote equity in shouldering the economic burden of illness. Acknowledgements--We would like to thank Peter Berman, Allan Hill, Mohammed Maiga and Joseph Maxwell, with whom we had many stimulating discussions on the issue of household coping from which this paper benefited a great deal. We are deeply indebted to the households we interviewed, for their hospitality, and the patience with which
they answered the many questions we posed. This study was part of the 'Burkina Health Care Intervention Study' financed by the European Commission, Directorate General XII (Science, Research and Development) under contract no. TS3"C'I92"0078 and carried out by the Ministry of Health of Burkina Faso (as principal investigator), the Medical Faculty of Ouagadougou University, ORSTOM (Organisme de la Recherche et la Technologie pour le Developpement), Paris, and the Institute for Tropical Medicine and Hygiene of Heidelberg University. In addition, the Gesellschaft ffir TechnischeZusammenarbeit (GTZ) and the Deutscher Entwicklungsdienst (DED) took active part in the study. Finally, we wish to thank the entire study team for their untiring work. REFERENCES
1. Berman P., Kindle C. and Bhattarcharya K. The household production of health: integrating social science perspectives on micro-level health determinants. Soc. Sci. Med. 38, 205, 1994. 2. Sauerborn R., Ibrangho I., Nougtara A., Borchert M., Koob E., Hien M. and Diesfeld H. J. The economic costs of illness for rural households in Burkina Faso. Tropical Med. & Parasitol. 42, 219, 1995. 3. Nur E. The impact of malaria on labour use efficiency in the Sudan. Soc. Sci. Med. 37, 1115, 1993. 4. Weisbrod B. Disease and Economic Development: the Impact of Parasitic Disease in St Lucia. University of Wisconsin Press, WI, 1973. 5. Davies S. Are coping strategies a cop out? IDS Bull. 24, 60, 1993. 6. Gore C. Entitlement Relations and Unruly Social Politics. A Comment on the Work on Amartya Sen. mimeo, IDS, Brighton, 1992. 7. BenderD. R. A refinement of the concept of household: families, co-residence, and domestic function.A m . Anthropol. 69, 493, 1967. 8. Swift J. Why are rural people vulnerable to famine?IDS Bull. 20, 8, 1989. 9. Chen M. A. Coping with Seasonality and Drought. Sage, New Dehli, 1991. 10. Sanerborn R., Nongtara A. and Diesfeld H. J. Action research on health services in Burkina Faso: concept and methodology 1988-1992. Proc. Conf. Methodology and Relevance of Health Systems Research International Children's Centre, Paris, April 1992. 11. Maxwell J. A. Understanding and validity in qualitative research. Harvard Educ. Rev. 62, 279, 1992. 12. Webb E., Campbell D., Schwartz R. and Scchrest L. Unobtrusive Measures. Rand McNally, Chicago, 1965. 13. Glaser B. and Strauss A. The Discovery of Grounded Theory. Aldine, Chicago, Mich. 1967. 14. Behrman J. Nutrition, health, birth order and seasonality: intra-household allocation among children in rural India. J. Devel. Econ. 2g, 99, 1988. 15. Sauerborn R., Nougtara A., Hien M. and Diesfeld H. J. Seasonal variations of household costs of illness in Burkina Faso. Soc. Sci. Med. 43, 281, 1996. 16. Miles M. B. and Huberman A. M. Qualitative Data Analysis, pp. 53-72. Sage, New York, 1984. 17. Strauss A. L. Qualitative Analysis for Social Scientists, pp. 55-128, 215-240. Cambridge University Press, Cambridge, 1990. 18. Sauerborn R., Nougtara A., Benzler J., Borchert M., Diesfeld J., Hien M., Ibrango I. and Mal'ga M. Strategies d'Adaptation des M~nages aux Co,~ts Economiques de la Maladie. Lang, Verlag Frankfurt, 1995. 19. Adams A. M. Seasonal food insecurity in the Sahel: nutritional, social and economic risk among Bamana agriculturalists in Mall Doctoral Thesis. University of London, 1992.
Strategies to cope with illness costs 20. Adams A. M. Food insecurity in Mali. Exploring the role of the moral economy. IDS Bull. 74(4), 41, 1993. 21. Reardon T., Malton P. and Delgado C. Coping with household level food insecurity in droughtaffected areas of Burkina Faso. World Develop. 16, 65, 1988. 22. Grandin B. Wealth Ranking in Smallholder Communities: A FieM Manual. Russell Press, Nottingham, 1988. 23. Janzen J. M. Therapy management: concept, reality, process. Med. Anthropol. Q 1, 68, 1987. 24. Janzen J. M. The Quest for Therapy in Lower Zaire. University of California Press, Berkeley, 1978. 25. Adams E. and Sauerborn R. A theory of household coping strategies. J. Peasant Stud. submitted for publication (May 1995).
301
26. Das Gupta M. Informal security systems and population retention in rural India. Econ. Devel. Cult. Change 35(2), 99, 1987. 27. Chayanov A. V. Organizatsiya krest 'yanskogo khozaistva. Translated: The Theory of Peasant Economy (Edited by Thorner D., Smith R. E. F. and Kerblay B.). Irwin, Illinois, 1966. 28. Daveloose P. L'organisation de la zone rurale de sant6 de Kasongo, Zaire. Ann. Soc. Beige Med. Trop. 59, 127, 1979. 29. Over M., Ellis R. P., Huber J. H. and Solon O. The consequences of adult ill health. In The Health of Adults in Developing Countries. (Edited by Feachem R. G. A. et al.). Oxford University Press, Oxford, 1992. 30. Chambers R., Longhurst R. and Pacey A. Seasonal Dimensions to Poverty, pp. 1-7. Pinter, London, 1981.
0277-9536(95)000375-4
Soc. Sci. Med. Vol.43, No. 3, pp. 291-301, 1996 Copyright © 1996ElsevierS~ace Ltd Printedin Great Britmn.All rights reserved 0277-9536/96$15.00+ 0.00
HOUSEHOLD STRATEGIES TO COPE WITH THE ECONOMIC COSTS OF ILLNESS R. SAUERBORN)* A. ADAMS 2 and M. HIEN 3 ~Harvard Institute for International Development, 1 Eliot Street, Cambridge, MA 02138, U.S.A., 2Harvard Center for Population and Development Studies, and 3Ministry of Health, Ouadougou, Burkina Faso Ab~'aet--The authors examine the strategies rural households in Burkina Faso used to cope with the costs of illnessin order to avert negative effects for household production and assets. They use information from 51 qualitative interviews, a household time allocation study and a household survey. Both surveys use the same sample of n = 566 households. The authors analyze these strategies along four dimensions: the type of behavior, the sequence in which strategies employed, the level at which strategies are negotiated, i.e. the household level,the non-household extended kin level or the community level, and finally the success of strategies in protecting household production and assets. A taxonomy of I 1 distinct types of coping behavior is developed which have the effect of either avoiding costs by 'ignoring'disease, or of minimizing the impact of costs on the household once illness is perceived. Intra-household labor substitution was the main strategy to compensate for any labor lost to illness. However, labor substitution did not eliminate production losses in the majority of households struck with severe illness of a productive member. Only wealthy household were able to fully compensate labor losses by hiring labor or by investing in equipment to enhance productivity. Sales of livestock was the main strategy to cope with the financial costs of health care. None of the households studied fell into calamity. However, the households' ability to avert the loss of production and/or assets was very varied and depended on household size, composition and assets, on the type and duration of illness and on clustering of crises (e.g. several repetitive or simultaneous illnessesor concurrent seasonal stress). Coping with the costs of illness largely occurred at the level of the household. Inter-household transfers of resources played only a small role. The authors develop the concept of risk households and suggest several policies with the potential to strengthen the ability of households to cope with the economic costs of illness. Copyright © 1996Elsevier Science Ltd. Key words--illness, costs, household, coping, production loss, utilization of health care
l. INTRODUCTION 1.1. The problem
The household production of health has recently received much attention from both the academic and policy communities [1]. With it, a new interest in the costs of illness for the household has emerged. A recent paper [2] reviewed the growing number of studies on the economic costs of illness. Only four of the 12 studies reviewed dealt with the time costs of illness and only two estimated intra-household transfers. Using the case of Burkina Faso, the authors concluded that the household, rather than the individual ill person, was the appropriate unit of analysis, given that the large majority of time and financial costs of illness were borne by healthy household members. Study findings indicated that average annual time costs of caring for the sick were considerably greater than the financial costs of care (21,700 and 8700 F CFA respeetivelyt) which on their own represent 6.2% of total household expenditures. *Author for correspondence tConversion rate at the time of the field study: I U.S.$ ffi 265 F CFA.
These attempts to describe the household costs of illness, however, beg several questions. With respect to financial costs, does relating reported health care expenditures to total household expenditures reflect the real burden to the household? Do other households transfer resources---both in cash and in kind--to support the household struck by illness7 Therefore, isn't the comparison of health care expenditures with total household expenditures overestimating the actual burden of illness costs for the household? Furthermore, what can we learn from risk-sharing arrangements within the community? If indeed a system of mutual transfers exists to cover health care expenditures, what implications does this have for policy makers struggling to introduce prepayment or insurance schemes? With respect to time costs, Nur [3] and Weisbrod [4] have argued that intra- and inter-household labor substitution occurs to counter potential production losses due to illness. If this observation held for the Burkina study population, the economic impact of illness of productive household members on the household economy would be much less than customary economic analysis indicates. This finding might
291
292
Rainer Sauerborn et al.
also undermine the economic argument for investing in health as a means of preventing production loss in a labor intensive economy. The research questions guiding the present paper are as follows: (i) What strategies do households use to cope with the financial and time costs of illness? (ii) Are these coping strategies effective in preventing or managing illness-related costs borne by households? (iii) Are there differences in the ability of households to cope with the time costs compared to the financial costs of illness? (iv) What policy conclusions can we draw from these micro-level observations? I.Z Conceptual framework
Coping is defined as a short-term strategy adopted within the prevailing value system to avert a negative effect on the actor [5, 6]. In the context of this paper, the actor is the household defined as a group of individuals--most commonly but not necessarily linked by kinship ties--who live together and share functions of production and consumption as well as of reproduction [7]. In the context of this paper, the negative effect coping strategies seek to avert is the breakdown of the household as an economic and social entity (referred to in this paper as 'calamity'). As in famine research [8], the criteria for calamity are household destitution and disintegration. In this paper, coping strategies to deal with the economic costs of illness are analyzed along four dimensions following the approach used by Chen [9]: • The type of the coping behavior is described. • The assessment of the level at which support is negotiated. Coping may be confined to the household in 7qhich illness occurs, or it may involve extended kin or non-kin related households. Finally, the community as a whole or groups within it may be a source of support. • The sequence of different strategies. Here we are interested in whether and how strategies are pursued concomitantly or sequentially. • The success of coping behavior. Three economic criteria are used to judge whether household coping is successful: First, did coping avert calamity--i.e, the social and economic breakdown of the household? Second, did it reduce or avert any negative effect of illness on household production? Third, did coping strategies preserve household assets? These four dimensions will be explored separately for strategies to cope with financial costs and those to cope with time costs.
2. METHODS
This particular analysis is part of a larger study carried out in the Kossi province of Burkina Faso which has been described elsewhere [10]. Both qualitative and quantitative methods were used in order to enhance the validity of results through triangulation [11, 12]. 2.1. Qualitative methods
Fifty-one qualitative interviews were carried out in two villages, S~riba and Bourasso. The households were selected using a theoretical sampling approach which sought to include households representing a wide variety of characteristics deemed to influence coping behavior [13] which comprise • Household size and composition. Here it is assumed that coping strategies are more effective in large households with many productive members who can substitute or compensate for work lost to illness. • Household wealth. It is hypothesized that wealthy households have a greater arsenal of coping strategies than poorer households. • Illnesses of differing degrees of perceived severity. Here, coping responses to mild/moderate and severe/fatal illnesses are compared. • Sick individuals of different age and gender. Comparing coping behavior in response to illness among productive and unproductive age groups as well as between genders, it is assumed that the more an individual contributes to household production, the more resources the household allocates to treatment [14]. • Seasonality. Finally, the sample should capture illness episodes occurring both in the dry and rainy seasons. Here it is assumed that the wide variations in the availability of household resources, in terms of time, cash and assets [15] shapes both the type and the effectiveness of coping strategies. Thirty illness episodes were analyzed through interviews with the household head, the sick individual or, in case of sick children, with their main care-giver. The households in which these 30 illness episodes occurred were selected from the sample of the household survey (n = 566 households, see below). This enabled us to select fatal illness episodes and nonfatal ones which had been perceived as severe or mild in the household surveys. Since the same households were revisited for the survey in a monthly rhythm throughout the year, we were able to select households who recorded illness episodes that had occurred in the rainy and those reporting dry season episodes. In addition, 21 'key informants' comprising village midwives, village heads, religious leaders and members of village organizations, such as the Young Farmer Association of Bourasso, were interviewed to further explore the determinants of household coping strategies. Interviews were translated from three local
293
Strategies to cope with illness costs Table 1. Seasonal variations in the economic costs of illness by type of costs Household cost Household cost Household cost per month: per month: per year: dry season rainy season all seasons Type of cost F CFA % F CFA % F CFA % Time costs of illness 1395 34.9 604 56.7 11,998 39.5 Time costs of caretaker 1356 33.9 263 24.7 9717 32.0 Financial costs of care 1250 31.2 197 18.5 8686 28.6 Total economic costs 4002 100.0 1065 100.0 30,401 100.0 Source: Household surveys no. 1 (dry season) and no. 3 (rainy season). Costs per year are calculated on the basis of a length of 6 months for both the dry (December through May) and rainy seasons (June through November). Table 2. Type and frequency of strategies to cope with the household costs of illness and death Time costs Financial costs Coping strategy No. of cases Coping strategy No. of cases Intra-household substitution 24/27 Mobilize cash/savings 16/25 Change in capital-labor mix 2/27 Asset sales (animals) 13/25 Hire labor 3/27 Loan 4/25 Free labor 2/27 Income diversification 5/25 Wage labor 1/25 Free care 2/25 Gift
15/25
Change in illness perception Change in health care choice languages into French, and coded using the software package " E t h n o g r a p h " . The coded interview transcripts were the basis o f conceptual analysis [16, 17]. In addition, seven household case studies permitted contextual analysis of factors influencing the decision to incur costs and how to cope with them [17]. A m o r e detailed description o f the qualitative method used in this study is provided by Sanerborn [18].
2.2. Quantitative methods In addition, this paper uses survey data on a sample of 566 households. The same households were visited six times between March and October 1992, spanning both dry and rainy seasons. Information was gathered on past perceived illnesses (recall: 1 month), on healer choice, on health care expenditures as well as on days lost to illness both by the sick and by healthy household members tending to the sick and/or accompanying them to a distant treatment site. Time costs were calculated by converting days o f f w o r k * due to illness into monetary terms using the current wage rate for day laborers [15] to provide a proxy for the opportunity cost o f time. In addition, two time allocation studies based on the recall of activities o f the previous day were carried out on the
*The monetary value of a day lost was equated with the average wage a household was stated to be willing to pay for day laborers during the month of the interview [15]. No attempt was made to value leisure time and economically active time separately. tThe terms "rainy" and "dry" season are not used in the climatological sense as having "rains" and "no rains", but rather in economical terms: the rainy season is the time of agricultural production, whereas the dry season is slack time, used for repairs, house building, marketing, crafts, and ceremonies. :~Details of the cost analysis have been presented elsewhere [2, 15].
same household sample during the dry and rainyt seasons respectively. Further details on the sample, the source population, the questionnaires and field procedures are described in a previous paper [2]. 3. RESULTS Table 1 summarizes the main components of household costs o f illness.:[: In the context o f the present analysis, the key findings are threefold: (i) time costs represent more than two thirds o f the total household costs o f illness; (ii) time costs of healthy household members are about as large as those incurred by the sick member, and (iii) the average annual financial costs amount to 6.2% of total annual household expenditures. This chapter describes how households coped with these costs. The analysis will evolve along the four dimensions of coping strategies described above: type, sequence, level and success.
3.1. Types of coping strategies Our taxonomy of coping behavior comprises eleven strategies: seven addressing the financial costs and four addressing the time costs of illness (see Table 2). 3.1.1. Coping with Financial Costs. In 25 of the 30 illness episodes studied in the qualitative sample, households incurred financial costs. The following analysis uses these 25 cases as the denominator.
(i) Using cash and mobilizing savings As a first step, most households used any available cash or savings to pay for health care expenditures. The amount, however, was generally insufficient: in only four out of 25 cases were illness expenditures completely covered. Cash sources included agricul-
Rainer Sauerborn et al.
294
tural and craft production and migrant remittances. In addition, two households derived cash from millet beer sales, and three from the pension income of members who were veterans of the French Army
(anciens combattants). (iO Sale of assets The sale of assets was the second most common way to meet health care expenditures. For households who did not possess sufficient cash, the sale of assets in the form of livestock was a widespread response to crises of many kinds [8, 19-21]. Indeed, in this part of Africa, animals were perceived as 'ambulatory savings-banks'. The sale of cereals, on the other hand, was considered a last resort You know you raise animals with the aim in mind to face this kind of expenditure, but as to the millet, that's very different because even if you sell the millet to pay for health care, after the cure, what are you going to eat? Villagers emphasized the dangers of selling cereal to overall food security, and stressed the need to avoid adverse seasonal fluctuations in food prices. During the dry season in which the majority of health expenditures occur, cereal prices are substantially lower than in the rainy season. As one farmer stated, selling millet in this season to cover health care costs was foolhardy as it might necessitate having to buy millet in the rainy season at a much higher price But if you have animals, that's always better to sell than millet, because if you sell millet now, you will be forced to buy back the millet at double the price later. For an expense in the order of the price of an animal will leave a big hole in your granary. Only one household head, who had neither cash nor animals, was forced to sell millet as a last resort in order to feed his newborn with formula milk after the mother had died in childbirth.
(iiO Loans In both villages of the qualitative sample it was customary to take loans. In S~riba no interest was claimed, while in Bourasso the going rate was 10% (except for loans to kin, which were interest free). The option to take loans, however, was only available to wealthy households with collateral (generally in the form of animals) Not everybody can get a loan. You have to have a guarantY or you must have someone who can guarantee for you .... By all means, the poor can't get a credit. In general, loans were perceived as a buffer between the time of need for cash and the time when the
household was in a position to pay back. Given the widespread belief that selling livestock under pressure would lead to a bad price, it was considered more advantageous, even for the wealthy, to take a loan, and delay repayment until livestock prices were more favorable. Another less common way to receive small loans was for the loanee to offer his labor to work on the fields of the creditor.
(iv) Income diversification Men and women from poor households used slack time in the dry season to generate additional income. Men typically enjoyed more leisure time during the dry season (3.4 hours per day compared to 2.0 hours per day for women*). This leisure time could be used to generate additional revenue to cover extraordinary expenditures, such as for health care. Household members engaged in a wide variety of income generating activities ranging from fetching firewood for millet beer breweries, building fences, weaving straw mats and honeycombs, and tailoring.
(v) Wage-labor In this subsistence farming environment, working as a wage-laborer was considered a last resort for those lacking assets, access to credits and kin/community support. In the rainy season, the trade-off between working one's own field to ensure household livelihood and selling one's labor to generate money for health care expenditures was especially stark ... one is obliged to leave one's own field in order to go to cultivate for someone else in order to be able to meet the needs of one's own family, this is the difference.
(iv) Free care Both the formal health services and traditional healers provided services at no cost (in 3 out of 25 cases). However, free care was not received by the needy, but rather by wealthy and influential households.
(rio tiffs In 14 out of 25 illness events, households received gifts from extended kint. The large majority of households who received gifts were wealthy. Only one poor household received a 5000 F CFA gift by an expatriate health worker. 3.1.2. Coping with Time Costs. In 27 out of 30 illness episodes, the households concerned lost time* due to illness. The analysis of strategies to cope with time costs was based on these 27 cases.
(i) Intra-household labor substitution *Recall covered the time from 7 a.m. to 9 p.m. tWealth was defined by the villagers themselves based on Grandin's [22] technique of wealth ranking among smallholder communities. ~Only full days lost for work due to illness were considered. §Household SER:I8 lost production of five doors to a competitor worth 12,500 F. Household SER:23 was not able to produce eight mats for a loss of 2400 F. Both craftsmen felt it was impossible to recover these losses.
The reallocation of tasks among household members was the most frequently chosen strategy to cope with anticipated production losses. In 24 out of 27 illness episodes, households sought to reallocate the home production tasks of the ill member among the healthy members. In the two cases§ where craftsmen suffered from
295
Strategies to cope with illness costs illness that prevented them from working, the household was not able to substitute their specialized skills. In the case of lost field production, household members who had not participated in agriculture before the illness event were mobilized. Children less than 10 years old, those who had retired from field work or who participated in other activities were called to the fields. Another strategy was to switch the sick person's tasks from physically demanding field work to craft production. In household SER=01*, the sick person took to weaving, whereas in BO_U-06, the chronically ill household head started to sew clothes when he could no longer work in the fields. It is intuitively obvious that the ability to reallocate the tasks of ill or deceased members depends on household size and composition. The monogamous nuclear family household BOU_-05 in Bourasso, for example, had no capacity to 'recruit' a new member into the field labor force. The polygamous 23-member household SER-03, on the other hand, was able to compensate for the production of a deceased household member by mobilizing the labor of three individuals.
(i0 Changing the capital-labor mix of production Two households responded to imminent longterm production loss due to illness by investing in capital, thus changing the capital/labor mix. One such household introduced oxen-plowing on its millet and sorghum fields to enhance productivity. Another way to change the capital-labor mix of agricultural production was to change the crop mix from a less capital-intensive one, namely millet, to one that requires more capital input (pesticide, fertilizer) and less labor for each unit of output. After the death of one of its main field workers, household BO_U-12 shifted half of its sown area from millet to cotton. The household head was able to do the necessary investments (seeds, fertilizer, pesticides, etc.), since he drew a pension from the French army as a veteran of the Indochina war. We are now growing cotton, which we have not done before, in order to fill the gap. Instead of selling the millet as cash crop, cotton is ideal for this. In only 3 out of 27 illness episodes with production loss, households chose this strategy. This comes as no surprise since additional financial resources, on top of those used for health care expenditures, were needed to invest in additional capital. Both of the above
*The interview ID is composed of the three first digits of the village and the two-digit interview number. 1" Work was not done for credit in the study villages.
households belonged to the upper wealth quartile in the village.
(iii) Hiring labor Some households hired laborers when they anticipated a short-term illness-related production loss during the peak agricultural season, when farmers had little flexibility to substitute or postpone work. However, the expenditures the household had to incur to pay laborerst in addition to paying for health care, were considerable which limited this strategy to households with the ability to generate cash quickly. Only two households in the sample hired labor. In this ease, the household head was a veteran who received a monthly pension of 15,000 CFA, the equivalent of about 60 days of hired labor. He hired 27 young men to work for one day as a substitute for one month of his wife's foregone labor. She had to be operated on for an incarcerated umbilical hernia during harvest time.
(iv) Free community labor In 2 out of 27 illness episodes with production loss, free labor was supplied to the household, in one instance by a community organization (the Young Farmers' Association in Bourasso), in the other instance by a neighboring household in the village. According to accounts of his wives and brother and to hospital records, Philippe Tiawa, a household head from Bourasso, had a bicycle accident with an open head trauma with a cerebrospinal fistula and subsequent meningitis. He was treated from March to June 1991, when he died. During the course of his illness, the household lost his work input as well as the field work of one of his three wives who stayed with him at the hospital. However, the resulting harvest equaled the previous one to which Philippe had contributed. This was due to the fact that the young men from the quarter went several times to cultivate his fields. In summary, intra-household labor substitution was by far the most frequently chosen strategy (in 24 out of 27 events with production loss). Two additional strategies, namely changing the capital-labor mix, and hiring labor required cash resources, which most households did not have. Free labor was only given to two wealthy households. Therefore, only four households in the upper wealth quartile were able to employ strategies other than intra-bousehoid substitution. In addition to poverty as a factor limiting the ability to cope with production loss, household size and composition compromised the availability and effectiveness of intra-household labor substitution. Poor small households are therefore at greatest risk of experiencing illness-related production loss.
296
Rainer Sauerborn et al.
no
I
I use uvlncs
no
~
yes
SUCCESS yes rio
11o
~
"<5.
take loan
n° 1 wage labor
sell animals
( ~.oo-, )
)
.o _ ~
no
yes
accept gift
( CAL~TY ~
( .~:.pt,,t
Fig. 1. Sequence of strategies to cope with financial costs. 3.2. Sequence 3.2.1. Strategies to cope with financial costs. With regard to household responses to financial costs, in 19 out of 25 illness episodes, households chose more than one strategy, and in 10 episodes more than two strategies. When several strategies were used, a clear sequence of coping behavior was observed which is depicted in Fig. 1. Using savings was in all cases the first strategy to cope with financial costs of illness. Yet only in four cases, savings covered illness related costs completely. The key factor influencing the choice of subsequent coping strategies was whether the household possessed any livestock.
*The possession of livestock was perceived as the main source of wealth by villagers [18].
Households possessing livestock had two options for the next strategy: they could either take a loan using their animals as collateral or they could sell their animals, pending on their evaluation of market prices. Loans were generally not available to households without livestock*. For these poor households (27% of the sample), the next coping strategy consisted of wage labor and/or income diversification through crafts. None of the wealthy households had to engage in income diversification or in selling labor. Rather, gifts were negotiated: In 15 cases, households received gifts when struck with illness or death. This option was available to only one poor household. 3.2.2. Strategies to cope with time costs. Household strategies to cope with time costs varied much less than strategies to cope with financial costs. In 23 out
Strategies to cope with illness costs
297
Table3. Levelof support by typeof economiccostand by typeof event(illnessor death) Production Financial loss
Level of support lntra-household Extended kin Nonkin and communitysupport
Illness 19/19 1/19 2/19
loss
Death 6/8 2/8 2/8
Illness Death 16/16 9/9 4/16 9/9 4/16 6/9
Table 4. Differencesin treatmentchoice and expenditureby age group and season Health care utilization(%) Average costs By age group By season Treatment choice (CFA) 0-9 yr />10yr Rainy Dry Total Home care 76 67.2 45.3 64.9 49.0 52 Village health worker 221 5.7 4.4 3.5 4.6 5 Traditionalhealer 656 II.l 19.6 11.7 15.4 17 Nurse, outpatientcare 1432 8.4 12.2 8.5 10.1 11 Hospital care 6939 3.1 10.6 7.4 2.1 8 Other n/a 1.6 8.0 6.7 6.6 7
of 27 illness episodes, intra-household labor substitution was the first strategy used to counter illness related production loss*. In only 5 of these 23 cases, households employed more than one strategy. As mentioned previously, only wealthy households could undertake alternative strategies to labor substitution. No clear sequence of further coping behavior was manifest. Rather, the main principle guiding the choice of coping strategies was whether the anticipated production loss was long-term (change in capital-labor mix) or short-term (hiring labor, free labor).
3.3. Level In-depth qualitative interview data revealed the strong perception that ultimately it was up to the household itself to deal with the financial and time costs of illness. The means to cope are first and foremost those of the household itself. It is only after having exhausted your resources that you can ask other parents to help you. All interviewed households stressed the imperative to explore their own resources to solve their economic problems before requesting or receiving economic support from outside (Table 3). It was only when intra-household support proved insufficient that support from kin was invoked. Indeed, respondents emphasized the advantage of having strong kinship networks on which they could depend in times of need. Kin support--both in cash and kind--was especially strong in cases of catastrophic illness of a productive household member.
*Four households did not use any strategy to counter production loss. tit is difficultto quantify how much production would have been lost in the absence of any coping strategy.
For the 1500F and the 2000 F, it was the entire big family who collected the money. As to the sheep it was me alone. Now, in order to compensate this hole, my brothers, big brothers and younger brothers, sent their children to help me in the fields. Contrary to the perception of the 'tightly knit' African community, support from village groups or non-kin households was rare in the case of ordinary illnesses. Data from the quantitative household survey indicate that only 15.4% of households with a sick person received any transactions from other households. Rather, outside help was highly concentrated in a few cases of catastrophic illness: 38.5% of the total value of illness related transactions took place in only 1% of illnesses.
3.4. Success As to the first criterion laid out above, namely calamity aversion, coping strategies were highly successful: none of the households disintegrated or migrated as a consequence of illness costs. The second criterion concerns the degree to which coping was successful in minimizing adverse effects on household production. Only a small proportion of wealthy households (4 out of 27) were able to maintain household production by substituting labor lost due to illness. Large households were at a clear advantage in replacing labor lost to illness. The majority of households, however, lost production in spite of attempts to reallocate labor internallyt. As to the third success criterion, whether coping behavior jeopardized the household's asset base, the majority (16 out of 25) of households which purchased health care or labor did so by selling livestock, thus reducing their asset base. They emerged from the illness episode poorer and more vulnerable, i.e. with a reduced ability to cope with further stress. Since the same asset buffer was needed for coping with a variety of stresses including
Rainer Sauerbom et al.
298
~
...............~
I
. . . . . . . . . . . . . . . . . . . . . ~"
m
4. DISCUSSION
~
/'/OUS~'~OL~
4.1. The role of the household in coping
II
Fig. 2. Model of household coping with illness costs. Strategies 1-3 aim at cost prevention and strategies 4--5 at cost management. For explanations, see text.
seasonal food insecurity* and other social and economic crises, the seeds for future calamity were sown. Gifts from other households were rarely given to cover treatment costs (5 out of 25), and in only two illness episodes did gifts cover costs sufficiently, such that the household concerned did not have to sell livestock. So far we have looked at household behavior to cope with illness related costs after they had been incurred by the household. In the previous paper, we presented evidence that household members change their perception of illness during the rainy season [15]. In spite of evidence for higher disease transmission and increased case fatality of key diseases, such as malaria during this season, people perceived fewer illnesses, and perceived them as less severe. We developed the hypothesis that the change of the household's economic parameters in the rainy season (i.e. increased opportunity costs of time, lower availability of cash, diminished assets) influenced illness perception. We consider this a strategy (albeit unconscious) of households to avoid costs (see Fig. 2, cost prevention strategies). For the same reasons, households are less inclined to seek any treatment at all, and even those who perceive themselves to be ill tend to keep working. Finally, during the rainy season healthy household members dedicate much less time to their sick kin than during the dry season. As we laid out elsewhere [15], these behavioral changes significantly reduced the economic costs of illness for the household.
*e.g. Animal sales represented the largest source of revenue to buy food during the pre-harvest season, tChayanov [27] pointed out that households at the beginning and the end of their life-cycle have a consumer-toproducer ratio ofl. In the mid-life of the household this ratio changes more and more in favor of consumers. With children gradually entering productive age the dependency ratio goes down again.
While illness occurs in individuals, its costs do not fall on ill individuals alone (see Fig. 2). Indeed, substantial amounts of costs would not have been captured in the analysis, nor could coping behavior have been assessed, without treating the household as the unit of analysis. It was the household as the "therapy managing group" [23, 24] that made decisions regarding health care choice and the allocation of time and financial resources to treatment. Sickness in one member affects the time use of healthy members and influences household decisions regarding the allocation of financial resources. Both the costs of care and strategies to cope with them, therefore, can only be understood in a household framework [1]. Study findings indicate that coping with the costs of illness largely occurred at the level of the household itself, and that inter-household transfers of financial or time resources played a smaller role. Chen's study [9] of rural Indian farmers coping with drought and seasonality, and Adam's [19,20, 25] work on household coping with food insecurity in Mali also emphasized the central role of the household in coping. However, evidence suggests that inter-household transfers are more acceptable in the context of food shortage than in the context of health care [19]. In the present study, both kin and community support (loans, gifts) were generally not available to poor households. The exclusion "~f poor households from inter-household networks of support has been observed in other African settings [19]. In rural Mali, Adams [20] noted that poor households had neither the time, status nor resources to invest in the maintenance of exchange relations, or their cultivation through marriage and reciprocal gift giving. This is in contrast to Das Gupta's [26] observation of strong community support for the poor households in rural India. Not surprisingly, other household characteristics such as size, dependency ratio and wealth had a strong influence on coping behavior. Larger households were able to substitute labor much more easily than small ones. In addition to household size, household age (i.e. the years elapsed since the founding of the household) and life cycle stage may influence its capacity to substitute labor. As the household matures, and children move from being net consumers to net producers, dependency ratios become more favorable, labor more easily substituted and coping strategies more diverse1". The relationship would work through changes in dependency ratios during the lifetime of the household. In spite of the importance of the household in coping with illness, policy has customarily targeted individuala at risk by characteristics such as age or gender. One lesson from the current study is that
Strategies to cope with illness costs there are households at risk of being pushed into poverty and calamity as a resuR of catastrophic illness. The most vulnerable and least able to cope are small and/or asset deprived households. It is conceivable that health centers establish household lists for their catchment areas which target small and poor households for special care [28]. Home visits, for example, might reduce the time access costs of care, waiving user fees for these households might reduce their financial access costs. 4.2. Policy and household coping strategies What policy options might strengthen the capacity of rural households to cope with the financial costs of illness? One suggestion is to develop programs to increase the asset buffer of households, and enable households to generate resources to cover health care costs. Expanding opportunities for animal husbandry, especially of sheep and goats, to the 29% of households who do not own any of these domestic animals, and protecting existing herds through good veterinary care would contribute to this end. A second way of strengthening household coping capacity is through the facilitation of income diversification. Promoting craft production which takes advantage of leisure time in the dry season represents a promising way of reinforcing household coping capacity. The provision of wage employment which is a very popular policy in India [9] may be another strategy to provide households with additional income. Developing rural credit markets and enhancing access to credit for poor households would give them the option to take loans to cover health care. Improving the capacity of households to effectively deal with frequent ailments such as malaria, on their own, and without recourse to expensive professional health care providers, might also lower both time and financial cost of care. For example, households could be encouraged to buy a stock of chloroquine tablets before the rainy season and be appropriately instructed about their use [15]. Schemes to finance health services might be designed to ease the difficulties households have to pay for health care in the rainy season [15]. Drugs and services could be priced seasonally such that higher prices in the dry season would subsidize fees and drugs in the rainy season. Another option would be to introduce prepayment schemes such that payments are collected after the harvest season to cover essential preventive and curative care throughout the year. For a detailed discussion of policies to help households reduce the costs of illness specifically in the rainy season, we refer the reader to our previous paper [15]. *This assumes, of course, that the quality and quantity of the work will remain the same.
299
4.3. Towards a model of household coping While illness occurs in individuals, its costs were largely borne by healthy household members, and to a much smaller extent, by extended kin or the community at large. Households demonstrated an impressive array of coping strategies which had two broad effects: to prevent costs from arising, and to manage costs so that the negative impact on household livelihood was minimized. Figure 2 models the three pathways through which costs can be avoided: (l) by modifying illness perception (the phenomenon of ignoring disease [15]); (2) by continuing to work irrespective of perceived illness (i.e. not assuming the sick role), thereby avoiding potential production loss due to illness*; (3) by allowing illness to go untreated which avoids financial costs. We call these three pathways 'cost prevention'. Should all these mechanisms fail, the household faces time and/or financial costs. We described two 'cost management pathways' the household used to minimize costs: (4) by summoning a range of strategies to minimize production loss; (5) by resorting to coping strategies to cover health care expenditures. 4.4. Does coping matter? Most households in our qualitative sample did lose production due to illness, as they were unable to fully substitute the labor of the sick individual. More importantly, the ability to cope with labor loss decreased with household size and wealth. By reducing the incidence, severity, and duration of disease, effective health services are a major means of maintaining and/or increasing production in settings where labor intensive subsistence agriculture prevails. A population-based intervention trial is currently underway in Burkina to test and quantify this conclusion. However, even when successful, internal labor reallocation is not without costs, as Over et al. [29] have demonstrated. For example, the household may shift labor away from health-maintaining activities such as washing, collecting water, cleaning the compound and cooking. Though this study did not attempt to measure the consequences of reallocating labor for child care, food preparation, leisure and education, it is safe to assume that the 'neglect' of these activities has a negative effect on household welfare in the long run. In contrast to Nur [3] and Weisbrod [4] who reported that illness related labor loss was easily substituted both from within the household and from the community, we found that community support played only a marginal role in compensating for household production lost to illness. With respect to health care expenditures, households showed considerable flexibility and inventiveness in garnering resources. However, even when
300
Rainer Sauerbornet aL
coping was successful in the short run, it was likely to increase vulnerability in the long run by reducing the household's ability to cope with future adverse events [8, 30]. While selling livestock enabled households to cover health care expenditures, it also made them poorer and less prepared to cope with future economic or social crises. Unfortunately, the time frame of the present study did not allow us to analyze the effect of repetitive stress sufficiently. A carefully designed, prospective study is needed to assess simultaneous illness events, food insecurity, seasonal stress as well as other adverse economic or social events (Fig. 2) over several years, and to analyze the priorities and coping behavior of households facing concomitant or repetitive stress. 5. CONCLUSIONS AS to conclusions for future research, the study underlines the need to focus on the entire household, rather than on the sick individual, when studying health seeking behavior and the economic costs of illness. Further research should illuminate the hypothesis that high opportunity costs of time and low cash availability during the rainy season impede illness perception. We would also welcome a longitudinal study of the inter-relationship between different types of economic stress on the household-seasonal food shortages, price fluctuations, expenditures for ceremonies and for health care. As to policy conclusions, it would be helpful to use the concept of 'risk households' in the delivery of health care. A household that is small and poor having experienced recent severe illness or death would be defined as 'at risk' of either ignoring illness (and thus allowing it to exacerbate) or of being unable to seek health care. Health services could target such households for special care, such as regular home visits, accelerated immunizations, and exemptions from fees. Since the household priority is to maintain production and assets rather than to maximize health, policies to increase the ability of households to produce and generate resources would seem appropriate in light of the study results. Such strategies include: improving, diversifying income generating activities, enhancing access to loans for poor households and improving animal husbandry to improve households' asset buffers. Finally, it does not seem realistic to count on community support for health care expenditures. Risk and cost-sharing schemes to finance health care would seem to be promising tools to promote equity in shouldering the economic burden of illness. Acknowledgements--We would like to thank Peter Berman, Allan Hill, Mohammed Maiga and Joseph Maxwell, with whom we had many stimulating discussions on the issue of household coping from which this paper benefited a great deal. We are deeply indebted to the households we interviewed, for their hospitality, and the patience with which
they answered the many questions we posed. This study was part of the 'Burkina Health Care Intervention Study' financed by the European Commission, Directorate General XII (Science, Research and Development) under contract no. TS3"C'I92"0078 and carried out by the Ministry of Health of Burkina Faso (as principal investigator), the Medical Faculty of Ouagadougou University, ORSTOM (Organisme de la Recherche et la Technologie pour le Developpement), Paris, and the Institute for Tropical Medicine and Hygiene of Heidelberg University. In addition, the Gesellschaft ffir TechnischeZusammenarbeit (GTZ) and the Deutscher Entwicklungsdienst (DED) took active part in the study. Finally, we wish to thank the entire study team for their untiring work. REFERENCES
1. Berman P., Kindle C. and Bhattarcharya K. The household production of health: integrating social science perspectives on micro-level health determinants. Soc. Sci. Med. 38, 205, 1994. 2. Sauerborn R., Ibrangho I., Nougtara A., Borchert M., Koob E., Hien M. and Diesfeld H. J. The economic costs of illness for rural households in Burkina Faso. Tropical Med. & Parasitol. 42, 219, 1995. 3. Nur E. The impact of malaria on labour use efficiency in the Sudan. Soc. Sci. Med. 37, 1115, 1993. 4. Weisbrod B. Disease and Economic Development: the Impact of Parasitic Disease in St Lucia. University of Wisconsin Press, WI, 1973. 5. Davies S. Are coping strategies a cop out? IDS Bull. 24, 60, 1993. 6. Gore C. Entitlement Relations and Unruly Social Politics. A Comment on the Work on Amartya Sen. mimeo, IDS, Brighton, 1992. 7. BenderD. R. A refinement of the concept of household: families, co-residence, and domestic function.A m . Anthropol. 69, 493, 1967. 8. Swift J. Why are rural people vulnerable to famine?IDS Bull. 20, 8, 1989. 9. Chen M. A. Coping with Seasonality and Drought. Sage, New Dehli, 1991. 10. Sanerborn R., Nongtara A. and Diesfeld H. J. Action research on health services in Burkina Faso: concept and methodology 1988-1992. Proc. Conf. Methodology and Relevance of Health Systems Research International Children's Centre, Paris, April 1992. 11. Maxwell J. A. Understanding and validity in qualitative research. Harvard Educ. Rev. 62, 279, 1992. 12. Webb E., Campbell D., Schwartz R. and Scchrest L. Unobtrusive Measures. Rand McNally, Chicago, 1965. 13. Glaser B. and Strauss A. The Discovery of Grounded Theory. Aldine, Chicago, Mich. 1967. 14. Behrman J. Nutrition, health, birth order and seasonality: intra-household allocation among children in rural India. J. Devel. Econ. 2g, 99, 1988. 15. Sauerborn R., Nougtara A., Hien M. and Diesfeld H. J. Seasonal variations of household costs of illness in Burkina Faso. Soc. Sci. Med. 43, 281, 1996. 16. Miles M. B. and Huberman A. M. Qualitative Data Analysis, pp. 53-72. Sage, New York, 1984. 17. Strauss A. L. Qualitative Analysis for Social Scientists, pp. 55-128, 215-240. Cambridge University Press, Cambridge, 1990. 18. Sauerborn R., Nougtara A., Benzler J., Borchert M., Diesfeld J., Hien M., Ibrango I. and Mal'ga M. Strategies d'Adaptation des M~nages aux Co,~ts Economiques de la Maladie. Lang, Verlag Frankfurt, 1995. 19. Adams A. M. Seasonal food insecurity in the Sahel: nutritional, social and economic risk among Bamana agriculturalists in Mall Doctoral Thesis. University of London, 1992.
Strategies to cope with illness costs 20. Adams A. M. Food insecurity in Mali. Exploring the role of the moral economy. IDS Bull. 74(4), 41, 1993. 21. Reardon T., Malton P. and Delgado C. Coping with household level food insecurity in droughtaffected areas of Burkina Faso. World Develop. 16, 65, 1988. 22. Grandin B. Wealth Ranking in Smallholder Communities: A FieM Manual. Russell Press, Nottingham, 1988. 23. Janzen J. M. Therapy management: concept, reality, process. Med. Anthropol. Q 1, 68, 1987. 24. Janzen J. M. The Quest for Therapy in Lower Zaire. University of California Press, Berkeley, 1978. 25. Adams E. and Sauerborn R. A theory of household coping strategies. J. Peasant Stud. submitted for publication (May 1995).
301
26. Das Gupta M. Informal security systems and population retention in rural India. Econ. Devel. Cult. Change 35(2), 99, 1987. 27. Chayanov A. V. Organizatsiya krest 'yanskogo khozaistva. Translated: The Theory of Peasant Economy (Edited by Thorner D., Smith R. E. F. and Kerblay B.). Irwin, Illinois, 1966. 28. Daveloose P. L'organisation de la zone rurale de sant6 de Kasongo, Zaire. Ann. Soc. Beige Med. Trop. 59, 127, 1979. 29. Over M., Ellis R. P., Huber J. H. and Solon O. The consequences of adult ill health. In The Health of Adults in Developing Countries. (Edited by Feachem R. G. A. et al.). Oxford University Press, Oxford, 1992. 30. Chambers R., Longhurst R. and Pacey A. Seasonal Dimensions to Poverty, pp. 1-7. Pinter, London, 1981.