The Implications of Private Safety Nets for Public Policy - Case Studies of Mozambique and Zimbabwe

26705 Africa Region H u m a n Development W o r k i n g Papers Series The Implications of Private Safety Nets for Public Policy Case Studies of Mozambique and Zimbabwe H a r o l d Alderman The World Bank Copyright 0 April 2001. The International Bank for Reconstruction and Development/ THE WORLD B A N K Africa Region Human Development Department 1818 H Street N W Washington D C 20433 USA For further information about the Africa Region Human Development Working Paper Series or Africa Region Human Development Series, please send an e-mail message to the following address: afrhdseries@worldbank.org. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author. They do not necessarily represent the views of the World Bank, i t s Executive Directors, or the countries that they represent. Cover design done by Tomoko Hirata and Joyce Petruzzelli T h i s paper was written by Harold Alderman of the World Bank. The author wishes to thank Jeff Alwang, Stephen Devereux, Trina Haque, Ebrahim Jassat, and Leslye Obiora for various forms of advice and assistance. The paper was edited by Fiona Mackintosh. The paper was prepared as a background study for the Social Protection Strategy of the Africa Region of the World Bank: Dynamic Risk Management and the Poor. The Strategy paper has been published in March 2001 and i s available as part of the Africa Region Human Development Series. Table of Contents Chapter I Introduction ............................................................. 5 Chapter I 1 Social Safety Nets in Zimbabwe .................................. 10 Response to Droughts................................................ 10 Coping with Economic Stagnation................................ 13 Health Shocks........................................................... 15 Widowhood and Family Break-Up ................................ 16 Chapter I11 Social Safety Nets in Mozambique.............................. 21 Response to Agricultural Risk., .................................... 22 Widowhood and Family Break-Up ................................ 24 Chapter I V Conclusion.............................................................. 27 References .............................................................................. 30 4 DEVELOPMENT AFRICAREGIONHUMAN SERIES WORKINGPAPERS  5 OF PRIVATE IMPLICATIONS NETS SAFETY FOR PUBLICPOLICY Chapter I Introduction M o s t households would l i k e t o reduce the ex- private a n d public safety nets. A n u m b e r o f excel- tent to w h i c h their consumption varies as a result o f lent recent empirical studies f r o m A f r i c a have con- fluctuations in their income. This can be derived tributed to w h a t i s known about h o w low-income from simple assumptions in economic theory, a n d households insure themselves against risk.’ This confirmed both with econometric studies a n d by study uses the insights from these studies as a frame- participatory interactions with the poor. Govern- work within w h i c h t o r e v i e w social protection p o l i - ments respond t o this desire by providing a range cies inMozambique and Zimbabwe. The limitations of programs aimed at reducing the vulnerability o f o f such a r e v i e w are readily apparent. Despite the poor households t o these income fluctuations. How- significant differences between the t w o countries in ever, the state i s a comparatively recent player in terms of their economic resources a n d physical a n d the provision of social protection from shortfalls in managerial infrastructure a n d despite the existence household income. In the absence o f state support, of a complementary s t u d y o f Togo (Bendokat a n d households developed a w i d e range o f i n f o r m a l TOVO, 1999), these case studies d o not cover a w i d e mechanisms to h e l p t h e m to cope with adverse cir- enough range of institutions t o p e r m i t easy gener- cumstances. alization. Moreover, detailed p r i m a r y data are nec- Several recent papers (Morduch, 1999; K o z e l essary t o study any one such institution in detail, and Parker, 1999; Besley, 1995; and A l d e r m a n and whereas this review m u s t integrate from a range of Paxson, 1994) have reviewed h o w w e l l these infor- secondary sources. Nonetheless, this study adds t o m a l safety nets have protected the poor f r o m extreme the perspective offered by the existing reviews a n d fluctuations in consumption, who are excluded f r o m empirical studies by stressing the interplay between such safety nets, and under w h a t circumstances func- government a n d NGO programs on the one h a n d tioning systems break down. A related question is a n d the various i n f o r m a l insurance mechanisms o n whether public programs complement informal pro- the other. grams or c r o w d them out, perhaps leaving some poor Before reviewing the programs inMozambique households even more vulnerable. a n d Zimbabwe, i t i s useful t o discuss a f e w general This paper reviews social protection p o l i c y in observations f r o m the literature on the mechanisms t w o A f r i c a n countries t o highlight the interplay o f that household a n d communities use t o smooth con- 6 DEVELOPMENT AFRICAREGIONHUMAN WORKINGPAPERS SERIES sumption. I t i s h a r d t o begin any such discussion of household income lost as a result of shocks. Simi- without a brief taxonomy. Households can manage larly, Fafchamps, Udry, and Czukas (1998) f o u n d risk, that is, reduce the variability of their income, that households in Burkina Faso sold cattle t o buy by a variety of practices that, o n average, lower their grain during periods of drought but that these sales income but also reduce fluctuations in their earn- financed o n average only 30 percent of the house- ings. They can also cope with r i s k by a variety of holds' income shortfalls. This was measured dur- measures that moderate the impact of any income ing a period of repeated droughts. Evidence f r o m losses o n their consumption. The mechanisms t o Burkina Faso and elsewhere confirms the intuitive achieve t h i s include b o t h individual resources such proposition that t h e degree and duration of eco- as liquid savings and assets and group-based lend- nomic stress affect a household's ability t o smooth ing. In addition, households sometimes rely o n re- i t s consumption; households that are able t o main- ciprocal assistance and cash transfers. Other classi- tain their consumption in the face of a single ad- fications indicate the difference between r i s k shar- verse shock m a y fail to do so following sequential ing over time (as in borrowing money t o be repaid shocks (Webb and Reardon, 1992 and Alderman, later) a n d across space (receiving transfers f r o m 1996). This i s in keeping with the argument made other households including remittances f r o m fam- by Deaton (1989) w h o noted that, in principle, most ily members w h o have migrated elsewhere to find households can smooth their consumption with their work). Another important distinction i s between savings. However, a large shock or a series of shocks covariant risk (those shared by a community) and m a y deplete the household's entire stock of savings. idiosyncratic r i s k s (those experienced only by indi- Another general observation o n informal in- vidual households). Finally, there i s the distinction surance i s that it i s generally more costly and less between stochastic risks -- such as drought - w h i c h accessible to the poor than t o other households. For cannot be anticipated except in a probabilistic sense example, the poor often have less access to credit, and those risks that can largely be anticipated such although group-based lending schemes m a y help as o l d age and seasonal factors like rainfall. In some asset-poor households t o overcome some of addition, there is the r i s k of death, w h i c h i s certain the obstacles that they face. Further, various forms but which occurs at a n uncertain time. of coinsurance based o n reciprocity among house- With a few noteworthy exceptions, most em- holds m a y break d o w n in the face of a sufficiently pirical studies have indicated that none of these large economic setback, w h i c h i s exactly w h e n it i s mechanisms perfectly insures a household against most needed (Morduch, 1999). Even ways of sav- shortfalls in i t s consumption. N o r i s it clear that ings are comparatively inaccessible t o the poor. The they should be expected t o d o so. The unit cost of costs of physical assets often have declining costs many insurance instruments m a y increase as the with the scale of purchase or--as in the case of live- level of protection increases, even t o a point where stock-- cannot be divided into small u n i t s for pur- it becomes completely unavailable. Households chase by the poor. Indeed, some forms of invest- m a y shift f r o m one means of consumption smooth- ments-- s u c h as irrigation, livestock, and schooling ing t o another as the relative costs of each method --both reduce risk and increase profits and thus d o change. Moreover, w h i c h method i s chosen by a not require the household t o endure income vari- given household i s affected by whether the mecha- ability in order t o raise i t s average earnings. Yet, n i s m i s reversible or n o t (Corbett, 1988). Finally, as the poor often cannot take full advantage of these l o n g as there are some costs t o smoothing consump- possibilities (Dercon, 1998). tion, the optimal amount of fluctuation inconsump- Another stylised fact about i n f o r m a l safety t i o n will not be zero. nets i s that they are more useful in helping house- Thus, although Rosenzweig (1988) f o u n d that holds t o cope with idiosyncratic shocks t h a n with transfers change in response t o changes in house- covariant shocks. Drought, floods, or insect infes- h o l d income, transfers o n l y make up for 10 percent tation can b e localized, but w h e n s u c h conditions  7 OF PRIVATE IMPLICATIONS SAFETY FOR PUBLIC NETS POLICY affect w i d e areas, community-based support sys- deprives these institutions o f sources of capital that tems are o f t e n overwhelmed. Although c o m m u - benefit credit societies a n d s i m i l a r (often r u r a l ) nity i s a fluid concept a n d involves k i n s h i p a n d groups. other ties that go b e y o n d t h e confines o f the vil- T o summarize, reciprocal insurance a n d self- lage o r neighborhood (Grimard, 1997), the reciproc- insurance through savings a n d the use of credit ity a n d i n f o r m a t i o n f l o w s t h a t are k e y features of markets plays a measurable r o l e inprotecting house- i n f o r m a l insurance systems do n o t generally ex- holds f r o m swings in their consumption. Yet, these t e n d over w i d e areas. Moreover, m a n y shocks are mechanisms often p r o v i d e only modest support, not directly weather related, but reflect macroeco- especially for those groups who are excluded f r o m n o m i c conditions, w h i c h put pervasive demands full participation insocial and economic interactions. on i n f o r m a l systems. Thus, governments can justify getting i n v o l v e d in O f t e n rotating savings clubs (Roscas) are con- h e l p i n g households to smooth their consumption in sidered a n i n f o r m a l insurance mechanism used by the same w a y that they can justify introducing pro- both u r b a n a n d r u r a l h o u s e h o l d s . P r e v a l e n t grams that seek to raise t h e welfare of the poor throughout the world, such clubs differ inhow flex- through the transfer of assets or income--to allevi- ible they can be in response to economic shocks. ate a n d reduce poverty.* Another justification i s the W h i l e some a l l o w households to remove their funds possibility t h a t r e d u c i n g r i s k m a y p r o m o t e eco- earlier t h a n scheduled in t h e rotation, based o n n o m i c growth. This m a y b e the case if the poor un- m u t u a l consent or o n the basis of need determined der-invest inproductive assets inorder to keep their through a n auction, m a n y schedule the order o f assets in liquid form, o r in order to participate in access t o funds by lottery. Such a schedule addresses activities that are less r i s k y but also less productive the need for households t o accumulate cash f o r pre- (Binswanger a n d Rosenzweig, 1993), or if the obli- dictable l u m p - s u m expenditures (such as school gations of community-based coinsurance discour- fees) but does not serve a n insurance function (Ru- age t h e m from accumulating assets. therford, 2000). One c o m m o n feature of these clubs W h i l e government interventions t o reduce r i s k i s that they have l i m i t e d transaction costs. Another m a y y i e l d welfare gains, they m a y also c r o w d o u t i s that they tend t o dismantle themselves at a pre- private insurance initiatives. The fact that the in- determined time (often t o convene again). Indeed, troduction of public transfers leads t o a reduction Rutherford (2000) argues that this self-liquidation in private transfers has been documented in a num- keeps transaction costs l o w . However, i t also im- ber of different circumstances (Cox and Jimenez 1998 plies that these clubs are vulnerable to long-term a n d Jensen, 1999 ). This possibility i s less l i k e l y covariant shocks; during a n economic downturn, w h e n the m o t i v e for transfers i s strategic such as to fewer members with regular incomes t o contribute assure access t o a n inheritance (Lucas a n d Stark, are available in each successive period. 1985), but, in such case, the recipient o f the transfer Also of interest are i n f o r m a l loans a n d oppor- i s n o t l i k e l y t o be inneed of p u b l i c assistance. M o r e - tunities f o r p a w n i n g (Rutherford, 2000). These tend over, the reduction in private transfers to a house- t o be m o r e accessible in u r b a n areas t h a n in rural. hold in response to the household receiving a pub- Again, they are vulnerable to repeated or long-term l i c transfer m a y l e a d to higher consumption in the economic stress, with their interest a n d default rates donor household. I f both households were insimi- b e i n g higher in the case of covariant shocks t h a n of l a r straitened economic circumstances the second idiosyncratic shocks. Moreover, w h i l e N G O s a n d round effects of a well-targeted p r o g r a m m a y also governments seek t o assist savings clubs, often try- b e w e l l targeted. ing to transform their short-term financial interme- Public programs m a y preempt private action diation i n t o longer-term development, they gener- inother manners as well. Restrictions on the move- ally l o o k u p o n the terms o f l e n d i n g in p a w n shops m e n t of grain in times of drought, f o r example, not a n d informal consumption credit as anathema. This only can interfere with n o r m a l market channels a n d 8 DEVELOPMENT AFRICAREGIONHUMAN WORKINGPAPERS SERIES price integration but m a y also inhibit the develop- percent of GNP but targeting the ultra-poor (47.2 ment of private coinsurance schemes. Similarly, percent and 37.8 percent of the population in Zim- regulations of credit markets m a y prevent contin- babwe and Mozambique respectively) w o u l d pro- gency lending at the local level. I t i s also possible vide a transfer of nearly 1 1percent of the income of that short-term assistance programs m a y interfere t h e p o o r in Z i m b a b w e but o n l y 5 percent in with reciprocity within a community without pro- Mozambique. To take this one step further, imag- viding a sustainable alternative. Finally, establish- ine a program in Mozambique o n the scale of South ing a legal framework for economic transactions Africa's o l d age pension, that is, one that transfers m a y clash with local customs and norms including 1.5 percent of GNP t o roughly 20 percent of house- those aimed at protecting widows and dependent holds. Ina n economy as relatively well off as South children. Africa and with a skewed income distribution, t h i s This paper examines this interplay between transfer increases the incomes of recipient house- government programs and traditional coping strate- holds by 50 percent (Case and Deaton, 1998) while gies in Mozambique and Zimbabwe--two countries in Mozambique such a transfer w o u l d only add 15 that differ appreciably f r o m each other in terms of percent t o the incomes of i t s beneficiary households. their level of infrastructure and the capacity of the A final perspective o n the limitations of social government to deliver services. Yet the t w o coun- protection in the poorest countries of Africa comes tries d o have some ecological and cultural character- f r o m looking at the current cash transfer program istics incommon. The countries chosen also share the in u r b a n areas of Mozambique (Low, Garrett, and need for selectivity since neither government can Ginja, 1998). This i s the largest single social protec- devote appreciable resources t o social protection t i o n program run by the Mozambique government without compromising their macroeconomic stabil- and one that, after some recent reforms, appears t o ity or investments in human and physical capital. be w e l l targeted inthe sense that comparatively few To illustrate this point, consider a transfer pro- of the recipients are non-poor. Nevertheless, in1998 gram that allocates 1percent of GNP t o social pro- it provided fewer than 40,000 households (1.4 per- tection, a figure that was estimated t o bring all the cent of the poor in r u r a l areas and 8 percent of the 30.4 m i l l i o n poor people in Brazil in 1995 up t o the urban poor) with a n average transfer per individual level of the poverty line (World Bank, 1999b). I f a that was roughly 10 percent of the national poverty s i m i l a r share o f GNP w e r e t o b e t r a n s f e r r e d line (and a smaller share of the urban poverty line) costlessly and equally t o all individuals defined as at a cost of roughly 0.2 percent of GNP. f a l l i n g b e l o w t h e p o v e r t y l i n e in Z i m b a b w e This and similar calculations indicate that di- (Zimbabwe,l998a), then each poor person w o u l d rect government transfer programs are likely t o have receive $0.85 a m o n t h . Each p o o r p e r s o n in only a l i m i t e d impact o n poverty in the poorest Af- Mozambique (Mozambique, 1998) w o u l d receive rican countries. The scale of poverty and the lim- only $0.11 a m o n t h i f a similar proportion of GNP ited resources in these countries are such that tar- was devoted to income transfers. Inb o t h cases, this geted programs that distinguish the poor f r o m the w o u l d provide the recipient with a n increase in in- nearly poor using the indicators of assets or poten- come of less than 5 percent of the average consump- tial earning can only bring about modest changes tion of the poor.3 in poverty rates. Therefore, i t i s useful t o find out The small amount of these transfers b o t h in more about what private arrangements households terms of their absolute value a n d of their share of are currently making t o provide themselves with household income reflects the l o w incomes in the social protection. I t i s also helpful t o explore the countries in question as w e l l as the fact that b o t h question of w h i c h of these methods could b e pro- countries use poverty lines that define the majority vided largely by informal networks a n d NGOs if of the population as poor (75 percent in Zimbabwe they were given technical assistance or seed money and 69 percent in Mozambique). Using the same 1 f r o m the government a n d which methods w o u l d  9 IMPLICATIONS OF PRIVATE NETS SAFETY FOR PUBLICPOLICY require substantial support f r o m the government to sources t o those households that are n o t anticipated be effective and sustainable. t o have the resources t o support themselves in the nthe con- This paper explores both of these issues i future. Yet, the design of a n insurance program text of t w o A f r i c a n countries--Zimbabwe a n d should differ f r o m one that p u r e l y transfers in- Mozambique. Chapter I1of the paper examines the ex- come. isting formal and informal social safety nets i nZimba- A similar, m o r e detailed i l l u s t r a t i o n for bwe, while Chapter I11does the same for Mozambique. M a l a w i can be f o u n d in W o r l d Bank, 1999a. The Chapter I V summarizes our conclusions. transfer inthe illustration used here varies as a share of income for the poor depending o n where they live, since b o t h Zimbabwe a n d Mozambique have Notes to Chapter I poverty lines that vary by locale. Innominal terms, the p o v e r t y lines u s e d f o r different regions in A p a r t i a l l i s t i n c l u d e s Udry (1994), Mozambique differ by as much as 250 percent. The A i n s w o r t h (1996), a n d Fafchamps, Udry, a n d estimates also require some creative interpolation Czukas (1998). Others are cited in M o r d u c h (1999). of the W o r l d Tables (World Bank 1998a), including, The phrase "safety nets" often fails t o d r a w in the absence of sufficient data, a n assumption that a distinction between the insurance function of so- the distribution of income inMozambique i s among cial assistance a n d the objectives of transferring re- the least skewed in the region. 10 AFRICAREGION DEVELOPMENT HUMAN WORKING SERIES PAPERS Chapter I1 Social Safety Nets in Zimbabwe Until recently Zimbabwe has been a moder- Response to Droughts ately prosperous country by Sub-Saharan African standards, with a n infrastructure that enables t h e Using a unique panel of households followed delivery of health and other public services. Never- between 1983 and 1996, Kinsey, Burger, a n d Gun- theless, Zimbabwe's economy has marginalized ning (1998) provided a portrait of the strategies used many of the poor households w h o live in commu- by farmers resettled in three settlements in Zimba- n a l lands. The country's agricultural l a n d i s semi- b w e to cope with droughts. The data cover the 1991- arid, receiving only 650 millimeters of r a i n a year 1992 drought - the worst o n record - as w e l l as the o n average. However, this i s highly variable; twice 1994 -1995 drought. Inkeeping with what has been in this decade total annual rainfall has been below observed elsewhere in Africa, the respondents indi- two-thirds of the long-term average. Inrecent years, cated that their m a i n source of cash t o buy the food Zimbabwe has also experienced widespread eco- that they w o u l d have n o r m a l l y produced them- nomic disruption due t o a combination of factors selves was through the sale of cattle. In b o t h 1992 including economic restructuring, the loss of tariff and 1996,60 percent of the panel indicated that they advantages in trade with South Africa, and high in- sold cattle, accounting for nearly 40 percent of the flation and interest rates attributable to general mac- cash that they used to purchase food. Somewhat roeconomic mismanagement. less typical for the Africa region, the second most Thus, in addition t o discussing private a n d important source of cash was cash holdings or postal public responses t o droughts, t h i s case study re- savings, with wage employment being the t h i r d views those programs that were put in place as most important means of financing f o o d purchases. safety nets in the current economic crisis. Another The panel provided some indications that this pervasive type of economic shock i s the A I D S epi- particular sample of households was resilient even demic, w h i c h i s discussed in b o t h the third sub- in the face of repeated droughts. First, the average section in t h i s section (which deals with health value of livestock that they owned in1994was greater shocks and death) and the fourth subsection (which than in 1991, despite the high number of sales and of deals with the economic consequences of f a m i l y the even larger number of animals that were lost breakup). during the dr0ught.l Second, while the number of  1 1 OF PRIVATE IMPLICATIONS FOR PUBLIC NETS SAFETY POLICY households that reported having savings accounts This long-term nutritional consequence of the dropped f r o m 42 percent in 1992 to 27 percent the Zimbabwe drought i s a bit of a surprise given that following year, it had recovered to 47 percent by 1994. the sample studied by H o d d i n o t t and Kinsey drew These figures, however, are not necessarily represen- o n their assets less than other communities t o make tative of the wider population in Zimbabwe. N o r i s up for agricultural shortfalls duringthe drought. For the fact that cattle prices dropped only 10 percent example, Cavendish (1999) indicated that the rela- below the 1983-1996 mean during the 1991-1992 tively isolated community that h e studied lost 80 drought necessarily typical of all communities. Nev- percent of their cattle between 1991and 1993. While ertheless, the special circumstances s t i l l indicate the there was a partial recovery by 1997--the number potential for households t o accumulate assets even cattle owned per household increased to half the pre- when their earnings are highly variable. Moreover, 1991level despite the occurrence of another drought they confirm the important role played by private --few households that h a d lost their entire stock were markets (for livestock) and by r u r a l financial institu- able to acquire cattle in this post-drought period. tions in providing self-insurance. In part, this reflect the high price of cattle in the re- The communities in t h e p a n e l studied by gion due to the reduced stocks locally and t o the Kinsey et al (1998) also benefited f r o m government c o m m u n i t y ' s l i m i t e d i n t e g r a t i o n t o t h e larger drought relief programs. In three years during the economy. Thus, the f i r s t drought left m a n y house- period 1981-1995, over three-quarters of the sample holds with reduced asset bases to serve as insur- reported receiving food a i d or drought relief, with ance against subsequent shocks as w e l l as reduced over 98 percent of the sample reporting that they draft power for agriculture. received such aid following the 1991-1992 drought. The programs for drought relief examined in Indeed, this m a y be partly why households were Hoddinott and Kinsey's study are s t i l l at the core of able to recover so rapidly after the rains returned. t h e Government of Zimbabwe's drought relief strat- Nevertheless, many households receiving food aid egy. In1992, the government offered a food for w o r k in 1992 or 1993 also reported having t o reduce the scheme as w e l l as supplementary feeding for chil- frequency of their meals a n d to rely o n foraged dren. A third of Kinsey, Burger, a n d Gunning's foods, a c o m m o n response t o f o o d shortages sample participated inthe food for w o r k scheme and (Corbett, 1988 and Cavendish, 1997). Moreover, a virtually the entire sample received food aid. By further analysis of the panel data by H o d d i n o t t and 1995 food for w o r k was largely discontinued due to Kinsey (1998) unearthed another aspect of inad- a n insufficient number of suitable projects and the equate consumption smoothing. H o d d i n o t t a n d m a i n vehicle for assistance inr u r a l areas was a grain Kinsey (1998) found that c h i l d r e n w h o were in the loan scheme, although with only 44,000 tons (plus most nutritionally vulnerable age bracket (12 to 24 cash w o r t h roughly another 10,000 tons) actually months old) during the drought that was caused by repaid out of 376,000 tons loaned in 1995-1996 the the failure of the 1994-1995 rains were shorter than program functioned largely as a grant. * other children w h o were in this age bracket in years The grain l o a n program assisted 6.4 m i l l i o n of normal rainfall. The ownership of liquid assets people in 1995-1996, a year that followed a harvest by a household was associated with more normal incommunal areas that h a d been only one-third the growth rates for children in the household in these average of the previous t w o years. The cost of this years, whereas the household being a recipient of program, exclusive of donations f r o m a i d agencies, state assistance h a d n o measurable i m p a c t o n was 708 m i l l i o n Z$. The grain loan program as- children's g r o w t h rates. While the latter result m a y sisted over 4.5 t o 5 m i l l i o n people in 1998-1999, fol- be due to limited variance--most of the households l o w i n g a harvest that, w h i l e affected by erratic rain, inthe sample were recipients--the fact that the com- was 75 percent larger than the 1995 harvest for the bination of savings, asset sales, and state support nation as a whole. Preliminary projections for 1999- did not prevent stunting i s puzzling. 2000 prepared in April 1999 were of a similar order 12 AFRICAREGION DEVELOPMENT HUMAN WORKING SERIES PAPERS of magnitude, in anticipation of a harvest that was ing program relies heavily o n community partici- expected to yield less than the level of annual na- pation. This participation as w e l l as caregivers' will- tional consumption because of flooding and pest ingness t o bring children t o the daily meal can be infestation rather than drought. considered a self-screening mechanism and has been Distribution of grain loans has been o n l y a observed t o decline w h e n enough locally g r o w n small share of the number of applications for a i d food i s available. However, this community par- received by the government. According t o the Her- ticipation differs f r o m key informal safety nets, as ald newspaper (April 13,1999), over 12 m i l l i o n ap- there i s n o reciprocity and n o explicit h a r d budget plications for grain loans and another 1m i l l i o n for constraint at the community level. free food were received--out of a population of 12.5 Given the scale of these programs, the l i m i t e d million. The issue of possible wholesale fraud i s a amount of administrative screening involved, and different concern than incentives for individual de- the absence of any incentive for self-selection (as pendence but nevertheless, this i s a n indication of w o u l d be the case in a true loan program or in a a serious management problem. Indeed, this m a y public works program), the government's drought have proven t o be a fatal flaw in t h e program. In policy runs a high risk of crowding out informal late April 1999, the government suspended the pro- insurance. In recognition of t h i s and of the pres- gram, and n o clear successor program has been an- sure o n the public budget, the government has tried nounced as of October 1999. After the program was t o revive a traditional system of support called suspended, the estimate of grain production incom- Zunde ua Mambo (ZR). The name means "chief's field" m u n a l areas was revised upwards by 400,000 tons. in Shona (the phrase in Ndebele i s Isipkala se Nkosi) The Ministry of Public Service, Labour and So- denoting land devoted to producing crops for a com- cial Welfare (MPSLSW) has also provided grants of m u n a l reserve. However, the ZR i s currently run at food, with n o pretense that they are loans, to indi- the village level rather than at the higher level of a viduals w h o are deemed to be unable to meet their chief's administration. In some communities, the food needs. T h i s includes the elderly and the chroni- ZR collects grain for storage or even for cash f r o m cally i l l. This i s not strictly a drought program--it residents. In others, i t modifies a tradition of com- reached 336,000 individuals in 1997 following a n m u n a l time-sharing (ckisi) t o mobilize labor t o cul- ample harvest. However, this was half the number tivate designated fields. of recipients in the year following the 1995 crop fail- The revival of some of the powers and respon- ure. Similarly, the concurrent supplementary feed- sibilities of traditional local leaders and chiefs has ing program--originally runby the Ministry of Health created some tensions with b o t h the centralization and n o w largely run by NGOs--scales up or d o w n of authority that has until recently been a hallmark depending o n harvests (and donor funding) but op- of Zimbabwe politics and with the nascent trend t o erates every year in some of the poorer districts. decentralize authority t o elected village develop- Thus, the government a n d donors have at- ment committees. A related issue i s that a number tempted to address the covariant r i s k of drought and of areas--resettlement communities, commercial locally chronic food shortages as w e l l as some idio- farms, and urban areas--have n o traditional struc- syncratic risks stemming f r o m the inability to earn tures t o build upon. In any case, the government's enough t o purchase a n adequate amount of food as role in this revival i s mainly as a motivator; little part of the set of programs covered under drought direct financial assistance i s available, although com- and food security assistance. These programs nomi- m u n i t i e s in districts that participate in the Commu- nally involve the local community structures. For nity Action Program can propose projects t o create example, village leaders are involved with drafting storage capacity. the lists of individuals w h o are eligible for free f o o d There i s currently little information o n the per- and in forming loan groups for participation in the formance of the ZR. However, given the extent of grain loan program. Also, the supplementary feed- government and NGO drought relief programs, i t  13 IMPLICATIONS OF PRIVATE NETS SAFETY FOR PUBLICPOLICY i s likely that the availability of external resources to crisis-induced) livestock sales or with remittances w o u l d discourage households f r o m turning t o the and wages. Current calculations of the requirements Z R w h e n faced with a covariant shock. However, for grain loans or free food d o n o t take such poten- the ZR m a y assist households t o p o o l their resources tial in consideration. t o deal with idiosyncratic shocks. Even in s u c h Moreover, the government's drought policies cases, the program must rely o n community soli- m a y discourage private (and communal) investment darity since the chiefs do n o t the ability t o directly in storage. While many of the trade restrictions that tax the community. hindered private trade have been removed in the Existing empirical studies of the degree t o last decade, these reforms have been designed more which public programs crowd out private safety nets t o promote spatial than temporal arbitrage. One have generally shown that there i s a modest over- unintended consequence of t h i s can be f o u n d in the lap between the t w o kinds of provision. However, market for maize meal. When the Government of these studies focused only o n static crowding out. Zimbabwe removed subsidies o n roller maize meal There may be as much or more crowding out in a in 1993, i t also liberalized private milling and trade. dynamic context by which government policy in- The net r e s u l t was that the cost of meal for poor con- fluence the nature of investments by individuals or sumers was lowered since these consumers were communities. For example, government drought able t o s w i t c h t o cheaper h a m m e r - m i l l e d m e a l policy may indirectly affect r i s k management strat- (Jayne a n d Jones, 1997). However, since smaller egies in that they may discourage the planting of m i l l s do not store grain, t h i s type of meal becomes drought-resistant crops. While sorghum and millet scarcer as more time goes by since the harvest and are adapted to l o w rainfall a n d can also be stored as the amount of domestic production goes down. for longer periods than maize, the acreage devoted Thus, the weighted average cost of meal (from b o t h t o these crops has stagnated inthe last three decades roller and hammer mills) i s more volatile than the while smallholder plantings of maize have roughly cost of grain itself. tripled. However, i t i s difficult to determine whether This i s not due t o a stated policy--in principle, the decline of the relative share of p l a n t i n g of hammer mills and small private traders can draw drought-resistant crops i s d u e t o a shift in the rela- f r o m the stocks and imports of the G r a i n Marketing tive profitability of maize (reflecting b o t h yield in- Board. Nevertheless, a combination of scale econo- creases and marketing reforms) or the availability mies inthe transport sector and administrative slug- of low-cost (to the farmer) - risk-coping mechanisms. gishness puts a de facto restriction o n access t o these I f the latter i s the case, then, t o the degree that the stocks. This reduces the opportunity that house- government's drought policy l e d to a n increase in holds have to cope with higher prices by shifting to expected earnings, these policies have served the cheaper maize meal. Given that the government will function expected of insurance programs, albeit us- be importing grain again inthe 1999-2000crop year, ing a fairly blunt and expensive means of provid- debate about t h e government's policy of releasing ing that insurance. grain to roller m i l l s will again be embedded in the I t i s also possible that, by p r o v i d i n g insurance larger issues of grain pricing a n d distribution. t h a t attempts t o f i l l t h e g a p b e t w e e n a r u r a l household's production and i t s annual grain con- Coping with Economic Stagnation sumption, the gain loan or free food programs dis- courage private and public investment innon-grain Zimbabwe introduced a number of reforms in r u r a l enterprise. Some regions of Zimbabwe are n o t the 1990s aimed at restructuring the economy. These likely to be self-sufficient in grain but can be self reforms liberalized external a n d internal trade and reliant in the sense of being able t o purchase grain reformed or divested many parastatals. Aware that g r o w n elsewhere (domestically or in neighboring similar efforts in other countries l e d t o substantial countries) with the proceeds of regular (as opposed economic hardship for at least a portion of the popu- 14 DEVELOPMENT AFRICAREGIONHUMAN PAPERS WORKING SERIES lation, the government took concurrent measures t o lute value of transfers received rises as household mitigate the negative impact of these reforms. H o w - expenditures rise. However, i f one looks at net trans- ever, while the reforms achieved a fair share of their fers after subtracting gifts given, one s t i l l sees that objectives, they did n o t lead t o macroeconomic sta- net transfers for the first three deciles amounted t o bility; massive deficits in 1997 and 1998 l e d to a 20 percent of their expenditures, but after that, the major devaluation a n d cascading domestic debt share declines rapidly. Although the richest t w o with substantial consequences for economic growth deciles received far more p e r capita t h a n poorer and inflation. This contributed to a large covariant groups, they also transfered more t o others. For economic shock that was only tangentially due t o these groups, net transfers were negative. structural adjustment, yet which relied o n safety n e t Remittances appear t o play a different role in programs designed t o address a far less pervasive r u r a l communities than in urban ones. The average loss of purchasing power. r u r a l resident receives a net transfer into h i s or her While r u r a l households are frequently affected household. This was confirmed by Cavendish (1999) w h e n devaluation raises the price of grain relative w h o also showed that remittances increased be- to wages (whichi s detrimental i f the households are tween 1991 and 1997 in the r u r a l community that net purchasers of grain), many of the households h e studied, which was supported by remittances that have been negatively affected by recent eco- f r o m agricultural labor. In contrast to t h e pattern nomic conditions are urban. Less i s k n o w n in gen- in r u r a l areas, the Income and Consumption Sur- eral about h o w u r b a n households respond t o eco- vey showed that the f l o w of transfers for a n aver- nomic shocks than about r u r a l households; infor- age urban household was outward. Moreover, while m a l networks m a y be less stable and less reciprocal households in the third decile in urban areas did in urban areas than in r u r a l areas. However, Cox receive a net transfer that amounted to 9 percent of a n d Jimenez (1998) indicated that remittances t o t o t a l expenditures, o n average, the t w o poorer some low-income u r b a n communities increase as deciles received negligible amounts relative t o their unemployment and income instability increase. expenditures. Thus, while it i s difficult t o extrapo- Plausibly, such remittances constitute a simi- late f r o m the available cross-sectional evidence t o a lar safety net in Zimbabwe; i t i s likely that r u r a l situation of transitory shocks, i t i s unlikely that the households regularly supply their urban relatives urban poor rely o n private transfers in time of eco- with grain (Mutangadura and Makaudze, 1999). nomic d o w n t u r n to the degree that r u r a l households However, n o direct evidence o n the dynamic na- benefit f r o m s u c h transfers. ture of transfers (which w o u l d be necessary to dis- Another w a y in which households sometimes tinguish coinsurance f r o m a long-term transfer of deal with economic crises that has been reported income, say, between generations or migrant mem- in the literature i s by w i t h d r a w i n g their children bers of households) i s documented in the available f r o m school. The Sentinel Surveys that monitor the data. However, the 1995-1996 Household Income social dimensions of adjustment have yielded some Consumption and Expenditure Survey does indi- evidence for this. Inthose surveys, the share of chil- cate that transfers, including gifts in kind, play a dren w h o reported that the cost of schooling was substantial role inboosting household income. The t h e reason why they were n o t attending school rose survey indicated that households inthe poorest four f r o m 33 percent in 1992 to 48 percent in 1996 (Zim- deciles in terms of expenditure per capita received babwe MPSLSW, various years). However, t h i s gifts and private transfers w o r t h over 20 percent of should be viewed with some caution since t h e Sen- total expenditures. Two-thirds of the entire popu- tinel Surveys did not record a decline in enrollment lation received some private transfers during the rates during t h i s period. Other sources of data o n survey period. enrollment are similarly ambiguous. For example, This ratio of transfers t o total expenditures while p r i m a r y enrollments as reported in national declines slowly as income rises. Indeed, the abso- statistics did decline 3 percent f r o m 1997 t o 1998,  15 IMPLICATIONS OF PRIVATESAFETYNETS FOR PUBLICPOLICY a n analysis of household survey data f r o m 1990 and Health Shocks 1996 showed that enrollment in p r i m a r y education increased despite a clear reduction in overall in- To a significant degree, a health shock affects come ( A l w a n g a n d Ersado, 2000). Moreover, al- a family in the same w a y as a loss of employment. t h o u g h secondary school enrollments did decline In many situations, the cost t o a household of treat- in the same period, the decline was n o higher for ing a n illness or injury is significantly less than the poor households than for other households, sug- income lost as a result of the illness (Gertler and gesting that the decline might b e d u e t o other Guber, 1997). Only a small share of employees has causes rather than households n o t being able t o some degree of formal insurance b o t h for unemploy- afford school fees. ment and retrenchment a n d for health shocks (in- The government explicitly addressed the pos- cluding medical leave). However, with the possible sibility that economic hardship m a y affect school exception of AIDS, t h e impact of health o n employ- enrolment by introducing the Social Development ment i s only weakly correlated across a n entire com- F u n d in 1991. This fund has a n employment and munity and, thus, i s the type f shock that m a y be training component as w e l l as a social safety net assisted with family and other support networks. function. The safety n e t offers means-tested assis- Even in the case of AIDS, w h i c h typically re- tance with school tuition and exam fees as w e l l as quires more care a n d involves more expense than with fees at health centers (it also initially included many other adult illnesses, community support a free food program to offset the impact of the re- n e t w o r k s have p r o v i d e d assistance t o affected m o v a l of maize subsidies). Since there are n o fees households. In Zambia, for example, community- for service in r u r a l health centers and n o tuition i s based care based, in part, o n strong volunteer sup- charged in r u r a l primary schools, the safety net i s p o r t was found to be a cost-effective means of car- largely a n urban program. The program explicitly ing for patients (World Bank, 1998b). Similar pro- targets the families of retrenched workers, though grams have been initiated in Malawi. This type of any family with a n income below Z$400 a m o n t h care m a y also diminishthe stigma of the disease and, was eligible in 1998 inprinciple. However, the pro- i f so, m a y contribute t o the open f l o w of accurate gram i s beset with problems, includinga n inefficient information that i s necessary t o limit i t s spread. screening mechanism a n d chronic u n d e r f u n d i n g Current global evidence (summarized in W o r l d (Kaseke, Dhemba, and Gumbo, 1997). The funding Bank, 1998b, Chapter 4) also indicates that a range for 1997-1998, for example, was less than 10 percent of informal networks provide remittances to fami- of the non-salary budget of the MPSLSW, w h i l e the lies that have members w h o are i lla n d often send grain loan and free food programs are expected t o someone to live with that family t o help with gen- comprise around two-thirds of these expenditures erating income w e l l as with in-migration to assist for the same year. There are n o data that permit with labor. Families m a y also assist by p r o v i d i n g analysts to distinguish between the different effects fosterage t o children. In general, such private remit- of the program, particularly between those that sub- tances exceed the amount of support provided by stitute for private savings or transfers and those that the government and NGOs by a substantial amount. increase demand for health a n d education. If de- This support--and the change in the composi- m a n d for these services increases, then the safety t i o n of the household--often continues after a death. net m a y encourage rather than c r o w d out private Indeed, the evidence f r o m panel data in Tanzania sources of income support inthe sense that the avail- and Cote d'Ivoire shows that households tend t o be ability of support for tuition prompts households fairly resilient in the face of the loss of a n adult; al- t o keep their children in school, meaning that they though total household expenditures decline imme- will either use their private savings or need in- diately after the loss of a n adult member, they then creased remittances or b o t h in order t o f u n d the re- partially recover. Such studies indicate that, in the maining costs of their children's education. longer-term, the death of a n adult appears mainly 16 AFRICA DEVELOPMENT REGIONHUMAN SERIES WORKINGPAPERS t o affect the household's investment in education that i s partially refunded if total contributions to and in child nutrition. This may be due t o the real- the society exceed outlays in a given period. Such location of chores as w e l l as to changes in parental societies m a y come under pressure as the probabil- care (and possibly the unrecorded transmission of ity of adult deaths increases in those places where AIDS t o the children). AIDS i s widespread. However, in Zimbabwe, i t i s Such panel studies are difficult t o conduct, and more common for members t o make a monthly sub- there i s n o direct evidence of this kind f r o m Zimba- scription. In s u c h a case, the subscription can be bwe. Indeed, what evidence there is, based o n De- increased if reserves get low. Similarly, n e w mem- mographic and Health Survey (DHS)cross-sectional bers are often asked to pay a n entry fee based o n evidence, indicates that Zimbabwe i s a n exception the accumulated reserves of the society. This some- t o the general pattern of lower school enrollments what offsets the r i s k of adverse selection. for orphans (World Bank, 1998b, Figure 4.13). H o w - ever, the difficulty in extrapolating f r o m the exist- Widowhood and Family Break-Up ing global evidence comes less f r o m the variations between countries than f r o m the fact that the prob- A death in a family generally causes the per- l e m i s accelerating with the possibility that house- sonal and financial bonds within the household t o holds m a y change f r o m one coping mechanisms t o change. M a n y of the issues involved in this rear- another as the scale of the problem increases. For rangement of financial bonds within a household example, traditional sources of support t o invalids overlap with the issues involved in traditional in- and t o families that have lost adults m a y become surance. However, many reciprocal relations in co- overstretched as the AIDS pandemic claims the sib- insurance are not enforceable under l a w - indeed, a lings and cousins of the original victims. Indeed, key question in understanding these structures i s the traditional practice in many African societies of h o w they function in the absence of mutual binding a m a n marrying h i s brother's widow, which pro- agreements. In contrast, rights of inheritance are vides the w i d o w and her children with a f o r m of covered by the law. In fact, they are often covered insurance in the case of accidents and other forms by more than one legal system. InZimbabwe, rights of adult mortality, m a y be responsible for the trans- of inheritance differ whether a marriage i s c i v i l or mission of AIDS. I t i s too soon t o k n o w h o w effec- customary and, if the latter, whether i t i s registered tive alternative structures will be i f t h i s practice or not. Moreover, even i f the marriage i s not regis- wanes. tered, certain rights are, in principle, guaranteed Burial societies also serve as insurance in that under legislation that was enacted t o ensure that the they spare the relatives of someone w h o dies the family of the deceased receives basic maintenance substantial burden of funeral expenses - which often f r o m the estate (Stewart and Ncube, 1992 and Gopal exceed the costs of medical care prior t o death even and Salim, 1998). in the case of protracted illness. Such burial societies Customary l a w in most of Zimbabwe allows arose inZimbabwe out of a need for transport incom- for rights to the property of a deceased individual munities of migrants and mine workers. However, t o pass t o h i s brothers and male descendants. H o w - they are n o w widespread i nthe country, covering as ever, in principle, traditional l a w also allows wid- many of 80 percent of residents insome urban neigh- ows to use their deceased husbands' property and borhoods, including individuals w h o are also cov- l a n d for a year, after w h i c h the widow makes the ered by formal sector funeral insurance. decision whether t o r e m a i n with h e r deceased These societies, in effect, subsidize the fami- husband's family with the right t o continue using lies of individuals w h o die young since these indi- h i s property or t o return t o her birth family (Stewart viduals make fewer payments than the average par- and Ncube, 1992, emphasis added). ticipant does. Rutherford (1998) indicated that of- H o w free the widow's choice i s in practice i s a ten such societies are based o n a n annual payment matter of interpretation. The w i d o w ' s access t o  17 IMPLICATIONS OF PRIVATESAFETY FOR PUBLICPOLICY NETS p r o p e r t y i s couched in terms of the use of the land or house i s the recognition that what one does lineage's--not her own--resources. H e r rights are not to one's brother's w i d o w m a y be done t o ones o w n transferable nor does the family responsibility for (Ncube a n d Stewart, 1997). However, this clearly i s her or h e r children's maintenance remain unaltered less direct than the incentives to honor obligations should she move t o a n u r b a n area. Moreover, i f she that are inherent in the repeated interactions of co- seeks to marry again outside her first husband's, this insurance. was often construed as leading t o the erosion of the Also, the nature of the inheritance structure family's resource base through her expenditures o n creates asymmetric economic incentives for wives her n e w husband or h i s children (Zimbabwe, 1995) and husbands. According t o Ncube et al(1997), one and m a y be cause for h e r t o be expelled f r o m the prevalent v i e w of the underlying tenet of t h i s tradi- kraal. In other instances, a w i d o w ' s refusal t o ac- tional l a w i s that a married w o m a n works for her cept her dead husband's family's choice of a n e w husband. Under this view, property acquired dur- marriage partner for her could be construed as a ing a marriage generally becomes the property of breech of her original marriage contract. In s u c h a her husband, just as property acquired through the circumstances, the w o m a n and her natal family m a y efforts of a n employee becomes the property of the be required t o return the bride wealth that was given employer. Thus, t o the degree that t h i s prevails, tra- her at marriage (Holleman, 1975). ditional inheritance practices are a disincentive for The tradition in w h i c h a dead man's family women t o accumulate assets just as in other circum- takes responsibility for h i s w i d o w provides her with stances inheritance and residency practices m a y be some guarantee of support. However, where l a n d disincentive for households to invest in educating i s the m a i n basis of livelihood, i t also leaves the their female members. w i d o w in a dependent state. The extensive social Arguably, departures f r o m the protection af- and economic consequences of such relations are not forded by traditional inheritance or the issue of in- confined to issues of inheritance a n d have been ex- centives in that system are of declining importance. tensively commented u p o n in the literature. While The L a w of Succession was amended in1997 to make many of these issues are beyond the scope of this surviving spouses the major beneficiary of a n estate paper, what i s germane t o the topic i s the question and to guarantee that the matrimonial home remains of whether the traditional or informal safety net for with the survivor or survivors (Gopal a n d Salim, widows, even i f i t i s functioning as intended, serves 1998). It remains to be seen, however, h o w the l a w i s the wider social goals of equity and economic de- interpreted and enforced. Indeed, the Supreme Court velopment. Moreover, in common with the other ruled in April 1999 (inMagaya uersus Magaya) that forms of informal insurance discussed in this pa- the Legal Age of Majority Act of 1982 does not apply per, i t i s important t o consider under what condi- to customary l a w and that under customary l a w only tions the traditional system fails t o function as in- men c a n inherit. While this ruling does n o t directly tended a n d what fallback o r recourse i s needed affect the L a w of Succession, i t implies that the Su- when those conditions exist. Finally, any attempt preme Court may be willing t o erode the legal gains to expand the coverage of the l a w must take i n t o made by women in recent years. account whether this will r i s k undermining local A number of NGOs have been active in chal- norms at a rate faster than the judicial and regula- lenging the earlier laws governing succession and tory system can be expanded to fill the gap. maintenance and are undoubtedly going t o press One issue for consideration i s the fact that, as the Supreme Court t o review i t s concept of the im- with any system, some individuals m a y n o t honor mutability of customary law. Presuming that the the obligations that offset their property rights un- current legal framework guarantees rights t o prop- der the traditional systems of inheritance. One check erty for w o m e n (such rights have n o t been chal- against the grabbing of assets a n d against the invol- lenged i f they are under c i v i l law), these NGOs also untary expulsion of a w i d o w f r o m her husband's will continue t o provide resources t o enable women 18 DEVELOPMENT AFRICAREGIONHUMAN SERIES WORKINGPAPERS to seek redress w h e n they are denied their rights. of childbearing in Africa, i t will only be a relatively We will return t o the question of access t o the l a w short time before AIDS-related "grandorphans" (chil- in the concluding section of this paper. dren whose grandparents died comparatively young) As with the norms regarding the maintenance become a noticeable share of the overall population of widows, all societies have norms that guarantee of orphans. There i s also anecdotal evidence that that minors w h o are orphaned are provided for. the number of street children i s increasing as i s the However, there are some signs that these are being number of child-headed households. To address the strained under the onslaught of the AIDS epidemic. strain o n traditional sources of care for orphans, the I t i s estimated that 30 percent of children under 15 MPSLSW has recently drafted a policy for orphans m a y be orphaned by 2010 (Foster, 1998). Estimates that will include grants for grandparents and for com- put the current number of maternal orphans in the munity services. However, the scale of the initial country at 600,000 (and more than twice that for all proposals clearly assumes that brothers and others orphans - UNICEF, 1998), out of which only about of the parents' age cohort are generally in a position 3,000 are ininstitutions. At the current time, as many t o support orphans. There i s some concern that any of half of all orphans are cared for by a grandpar- increase in government provision of maintenance ent, with a number of these households becoming grants for orphans will crowd out this source of sup- children-headed households after the death of the port f r o m the orphan's family. As with drought re- grandparent (Foster, 1998). While the precision of lief, the Department of Welfare also expects that the such estimates can be questioned, the implications expanded ZR will assist communities in providing are obvious for any traditional system that evolved support t o orphans, encouraged, in part, by the ex- under vastly different conditions. amples of a few pilot projects. Similarly, a few NGOs The different conditions are partially a matter and community-based organizations have explored of sheer numbers but also an issue of the correlation community-based orphan care o n a small scale that of AIDS across family members. T h e particular cir- relies o n local volunteers w h o are often trained cumstances of paternal orphans are directly tied to through projects and are linked to the Zimbabwe the widows' rights that were discussed above. They AIDS N e t w o r k (UNICEF, 1998). Plausibly, one such differ in the case of AIDS orphans mainly in the de- private source of supportthat i s provided t o orphans gree to which the widow's chances of remarrying are of farm laborers will be eroded by the current unrest limited by the decimation of the male population by in commercial farm communities in Zimbabwe over AIDS as well as to the degree that practices that lead land reform. to AIDS are more s i m i l a r among brothers than among Divorce has some features in common with other males in the same age cohort.1° Maternal or- other idiosyncratic shocks discussed in this section. phans are traditionally looked after by the new wives Like widowhood, divorce (or abandonment) irre- of their father, w h o frequently comes f r o m the same versibly changes the economic and social relations family as the deceased woman. However, such a re- of a woman's life and, as in the case of inheritance, marriage may be affected if the previous wife was a the degree t o w h i c h a woman's economic rights are victim of AIDS and the survivinghusbands has, or i s protected is influenced by the f o r m of her marriage assumed t o have, AIDS himself. However, the m a i n as w e l l as by society's views of her economic contri- difference between orphan due t o AIDS and the more bution. However, the rights of women in the event general risk of parental death i s an increased prob- of divorce are far less certain than inthe case of wid- ability of children with neither a surviving mother owhood.ll Moreover, given the additional elements nor father or whose surviving parent i s too i llto look of blame a n d of competition involved in divorce, after a child. the transfer of rights i s far more complex than in While such children are still largely looked af- inheritance. ter by their parents' relatives, often the principal The economic basis of the property rights of caregiver i s elderly. Moreover, given the early age a w o m a n in divorce m a y i n v o l v e one o r m o r e of  19 IMPLICATIONS OF PRIVATE NETS SAFETY FOR PUBLICPOLICY t h e f o l l o w i n g criteria: (i) her r o l e as the guardian though in many cases these are more l i m i t e d than of minors; (ii) the property that she brought i n t o those recognized under c i v i l law. There i s b o t h a t h e marriage; (iii) a share of the p r o p e r t y accu- policy issue about whether the differences between m u l a t e d during the marriage; (iv) her legitimate these legal systems are in keeping with the social expectations in marriage, o f t e n based o n t h e and economic goals of the country and a practical family's standard of living w h i l e t h e f a m i l y was issue of h o w the rights implicit under either kind of intact; a n d (v) the o p p o r t u n i t y costs t o her of de- l a w can b e protected in fact. v o t i n g her time t o h e r marriage obligations, when, for example, a w o m a n gives up a career. W i t h the L a w of Maintenance possibly still valid Notes to Chapter I 1 even for customary marriages in Zimbabwe, the first of these criteria may be relevant in b o t h customary Kinsey, Burger, a n d Gunning (1998) discuss and c i v i l law. Legal rights advocates have under- other studies f r o m Zimbabwe that contrasted with taken an educational campaign, that has largely been their o w n findings. A n u m b e r of other s t u d i e s successful in spreading i n f o r m a t i o n about these showed far fewer livestock sales; other f o u n d more rights. However, in customary practice, a woman's reliance o n private remittances. economic rights in divorce are largely determined by * Unless otherwise indicated, all data on pro- her fairly circumscribed rights of custody; in most gram costs and number of recipients in this case cases, if lobola (a bride price) has been paid, children study come f r o m the MPSLSW. Annual costs and stay w i t h their father's family rather than with their similar figures vary a bit f r o m report t o report even mother. Thus, many w o m e n have n o traditional within the ministry. However, s u c h differences d o claim to resources under the first basis for property n o t affect the general picture. settlement outlined above. Juvenile courts can re- As i s often the case, the programs also have a solve disputes about custody and child maintenance m i x e d record of managerial a n d logistic capacity includingthe logistics of h o w the w o m a n c a n receive (Kaseke, Dhemba, and Gumbo, 1997). This influences regular child support payments w h e n her children's b o t h the targeting efficiency and the costs of opera- father does not have regular wages. Even when the tion, but these administrative problems reported are father has made a payment t o the court, a woman secondary t o the policy concerns of sustainability and sometimes fails to take up the payment because she incentives covered in t h i s case study. has not been informed the money i s available or be- Several caveats are important. A certain share cause she has n o access t o transport or time t o queue. of &ts reported are social functions that differ from The third criterion listed above (a share of the either insurance of income support functions as com- property accumulated during the marriage) i s rec- monly conceived although they represent part of the ognized under family l a w in Zimbabwe, but the social fabric. Over two-thirds of the population re- right i s not uniformly enforced by the judiciary. The ported receiving such &ts i nkind or cash. W h i l e gov- problem stems inpart f r o m the fact that, in c i v i l law, ernment transfers appear to have been small (received property in marriage i s considered out of community by 3 percent of the population and amounting to only (which implies that property i s held individually) 2 percent of the value of private @s and transfers re- unless there i s a specific agreement otherwise. In ceived), some government transfers i n kind (for ex- most case, t h i s discriminates against a woman's ample, tuition support) may have been coded as pri- claim to property. The courts do, however, have vate @s i nkind. The data imply a net inflow of funds. the right to award a w o m a n a share of property o n While this i s possible i f the family has migrant rela- the basis of her contribution t o the household, and tives living abroad, it i s also consistent with the classi- judges in urban areas sometimes u p h o l d this right. fication of some public transfers as private. Customary l a w also recognizes women's rights j The outflows of transfers f r o m urban com- under t h e second, third, a n d f o u r t h criteria, al- munities in 1995-1996 may be a direct investment in 20 DEVELOPMENT AFRICAREGIONHUMAN WORKING SERIES PAPERS social capital that could be drawn u p o n inother years. This p o i n t i s general. B r o w n i n g (1994), for However, t h i s i s more likely when an economic shock example, observed that differences in l i f e expect- i s confined to the urban area or when the shock i s ancies a n d in the age at w h i c h people receive a idiosyncratic. pension leads t o different preferences regarding The higher costs of AIDS were documented, the f o r m of a pension. for example, in r u r a l Tanzania (World Bank, 1998b) l o UNICEF (1998) cites a study f r o m M a l a w i and in Zimbabwe (Zimbabwe, 1998b). that indicates that orphans in patrilineal kinship ’ Inpublic meetings sponsored by the Masvingo groups were more likely t o be isolated f r o m their Department of Social Welfare (Zimbabwe, 1995), a mothers. The study also found that widows were number of local leaders stressed that the responsibil- often sent away and that land grabbing was more ity for the care of widows does not necessarily involve common in patrilineal groups than in matrilineal conjugal rights. However, f r o m the context of their groups. comments, it appears that they were attempting to l 1 The probability than a m a n will leave be- shift, rather tha n to reflect, prevailing norms. hind one or more w i d o w s i s quite different than * Louis Masuko, personal communication. t h e odds that a g i v e n w o m a n will survive her See also Mutangadura and Makaudze (1999). spouse.  21 OF PRIVATE IMPLICATIONS NETS SAFETY FOR PUBLICPOLICY Chapter I11 Social Safety Nets in Mozambique A l t h o u g h it shares a c o m m o n border with Zim- riorate. This again, reduced communities' ability babwe, Mozambique differs appreciably from its t o insure themselves against agricultural fluctua- neighbor in terms of its colonial a n d post-colonial tions. Third, m a n y households lost their herds, history. Its liberation struggle was f o l l o w e d by a w h i c h h a d served both as a productive assets a n d prolonged a n d devastating c i v i l war. W h i l e the rate as a traditionalmeans o f savings against drought o f economic g r o w t h since the e n d of this w a r has a n d o l d age. Finally, with a large share o f the popu- been impressive, the legacy of that conflict i s pro- l a t i o n h a v i n g recently been refugees o r migrants, found. Not o n l y were m a n y of the health posts a n d traditional systems o f co-insurance based o n mu- schools destroyed in the war, the generation that t u a l knowledge a n d on repeated interactions m a y s h o u l d h a v e benefited m o s t f r o m these services be under m o r e stress t h a n the degree o f economic spent its childhood as refugees or, not infrequently, transformation in the countryside might otherwise as conscripts. Another legacy of the w a r is the fact imply. It i s h a r d to m a k e a full assessment of this that, among the various idiosyncratic risks that con- final proposition. A l t h o u g h extensive data o n vul- front the Mozambique people, i s the pervasive r i s k nerability have been collected at the district level, o f the estimated 2 million r e m a i n i n g l a n d m i n e s a n y data o n trends are either based o n a short pe- (Mozambique, 1998). r i o d or have gaping holes. In a d d i t i o n t o i t s g e n e r a l i m p a c t o n t h e W h i l e this case s t u d y has m u c h in c o m m o n economy, Mozambique's recent history affects the with the previous one o n Zimbabwe, i t concentrates issues reviewed in this paper in four specific ways. o n t w o specific types of shocks, n a t u r a l disasters First, the conflict disrupted the already r u d i m e n - (drought or floods) a n d f a m i l y break-up. The dis- tary markets in the country. This has l i m i t e d the cussion of natural disasters also discusses some o f extent t o w h i c h trade a n d financial intermediation the programs designed t o respond to the wartime has been able t o mitigate the impact of idiosyncratic disruption, since any of the mechanisms to deal with a n d communal economic shocks. Second, m u c h of drought - both i n f o r m a l a n d f o r m a l - o w e m u c h o f the grain storage capacity of households a n d com- their current form t o strategies devised t o deal with munities was either destroyed or a l l o w e d to dete- the dislocation of the c i v i l w a r . LL AFRICAREGIONHUMAN WORKING DEVELOPMENT SERIES PAPERS Response to Agricultural Risk of the number of months per year that a household (or a n average household in a district) has to rely While Mozambique was seriously affected by o n i t s o w n production of crops for i t s consumption. the regional drought in 1992, there are several rea- This near self-sufficiency in many r u r a l com- sons why it i s unlikely ever to suffer in the aggre- munities i s reminiscent of the conditions that pre- gate as greatly as Zimbabwe did in either 1991-1992 vailed prior to the c i v i l war. During the war, h o w - o r 1994-1995. These i n c l u d e t h e fact t h a t ever, many households were unable to plant or to Mozambique has more and more regular rainfall in store grain and fled to refugee camps or towns where the central and northern part of the country and the food distribution centers were established. Food a i d fact that i t s resource base i s more ecologically di- reached as high as 70 percent of total food produc- versified. Moreover, crops such as millet and cas- t i o n in the country. While the country n o longer sava, which are relatively drought resistant, make receives food a i d f r o m international donors, one up a significant portion of the country's total pro- legacy of the distribution of food aid i s that the coun- duction and reduce i t s aggregate vulnerability to try has benefited f r o m various systematic mapping shocks. exercises that recorded chronic food insecurity and This said, the r u r a l economy in Mozambique r i s k of disaster at the district level. For example, i s characterized by regular localized drought, flood, the Ministry of Health has district profiles of nutri- and crop infestations a n d by subsistence-level stan- t i o n that i n c l u d e rates of malnutrition a n d price dards of living in many r u r a l households. Given the trends as w e l l as information o n production systems limited number of feeder roads into r u r a l areas and and markets. While t h i s information i s useful for the fact that there i s l i t t l e truck transport between nutrition planing, i t has also been analyzed along the n o r t h a n d s o u t h o f the c o u n t r y except v i a with extensive data f r o m the W o r l d Food Program Malawi, the local variability of production i s virtu- and the Famine Early Warning System in a vulner- ally synonymous with the local variability of sup- ability assessment (Mozambique, 1998). Inaddition, ply. Prices vary greatly b o t h over t i m e and among Medicins Sans Frontiers (MSF) collects price, pro- different communities. For example, maize prices duction, and food stock data every m o n t h f r o m 5,000 rose seasonally by 91 percent o n average in the households t o produce a bimonthly bulletin of food five months following t h e excellent 1996-1997 har- security at the district level. These bulletins alert vest (Mozambique, 1998 ). Moreover, in November the government and NGOs about local transitory 1997, maize prices in Maputo were virtually twice shocks. For example, t h e bulletin for December those in Nampula where there was a surplus. H o w - 1998/January 1999 pointed out flood damage that ever, i t i s n o t necessarily the case that traders fail t o was occurring in areas of j u s t a few hundred hect- respond t o price differences. Traders currently ex- ares. p o r t substantial quantities of maize f r o m This illustrates the checks and balances that are Mozambique to Malawi, and they handled food aid possible w h e n several repeated a n d overlapping during the war, albeit with some direct subsidies data collections exist. Estimates reported in the (and local monopolies). Ministry of Commerce's Boletim Mensa1 d e Com- Farmers can deal with the seasonal rise in mercia de Cereais (supported by the FAO) for the maize prices, t o a degree, with a shift t o cassava; m o n t h of February 1999 indicates that 400,000 there i s some econometric evidence that this i s what households were affected by floods. As indicated happens. Nevertheless, the seasonal price variabil- in contemporaneous MSF a n d WFP data, these ity i s consistent with the extensive evidence that a households were mainly in the south and center of "hungry season" occurs between harvests. This, of t h e country. This prompted a closer evaluation of course, i s n o t strictly a n issue of risk since the sea- the situation by the government and international sonal pattern i s fairly predictable. Instead, i t i s evi- NGOs. The March bulletin of the Ministry of Com- dence of chronic food insecurity, indicated in terms merce reflected this re-evaluation, with the estimates  23 IMPLICATIONS OF PRIVATE NETS SAFETY FOR PUBLICPOLICY of food needs being drastically scaled back to a short- part of the country, people f r o m Mozambique tend to fall of t w o months for 25,000 families. migrate to South Africa. However, migrationi nnorth- Several NGOs use these data either inplanning e r n areas i s b e l i e v e d t o b e f r o m M a l a w i i n t o the distribution of their direct assistance or in plan- Mozambique. This reflects the relative land-labor ra- ning public works projects. Unlike in the food as- tios i n the two countries and m a y be more of a long- sistance program in r u r a l Zimbabwe, it i s assumed term income-earning strategy by Malawians than a that i t i s normal in food-insecure areas l i k e Gaza t o risk-reducing or risk-coping strategy per se. depend o n food f r o m markets for three months and Similarly, cross-border trade with M a l a w i i s that, therefore, food-for-work programs need only increasing, with f o o d f r o m M o z a m b i q u e b e i n g fill that gap. Similarly, the agency responsible for traded for manufactured goods marketed through disaster relief began i t s recent strategy review with --but n o t necessarily p r o d u c e d i n - - M a l a w i a survey of the responses given by village residents (Whiteside, 1998). Local a n d central government to the question: What w o u l d y o u d o in the event of officials in Mozambique are ambivalent about s u c h flood if there were n o one else beyond this commu- trade, expressing concerns that national food secu- nity t o provide assistance? rity will be harmed, even if household incomes in- I t i s h a r d t o assess the aggregate impact of such crease. Concerns are also voiced that f a r m house- programs. Only 6 percent of respondents in a 1996- holds will be exploited or will sell all of their grain 1997 household survey reported receiving direct and not have enough left over for their o w n con- assistance in terms of money, food, seeds, or cloth- sumption during the lean season (Whiteside, 1998). ing. However, most cluster-based sampling tech- Finally, the government i s concerned that i t will not niques are n o t designed t o gather information o n be able to control or tax this trade. These concerns the type of local programs that the early warning lead t o a variety of barriers, some of which are ad systems often support. Moreover, many of these hoc at the border crossings a n d some of which are in projects are funded by the WFP through food-for- conflict with government policies expressed in w o r k programs, which have been scaled up since Maputo. the 1996-1997 survey was carried out. If these barriers were reduced, the result should Inany case, neither NGOs nor the government be similar t o the type of long-term benefits that ac- are in a position to address idiosyncratic shocks or crue f r o m trade liberalization. Real incomes should the full extent of chronic f o o d insecurity. As inmost rise o n b o t h sides of the border, a n d there should be of the rest of the world, r u r a l households cope with les variation in f o o d prices in southern M a l a w i . r i s k in a variety of different ways including diversi- Whether s u c h trade w o u l d also reduce r i s k in fying f r o m agriculture i n t o livestock (in the south Mozambique i s less clear. I t i s possible that at times of the country) and fishing as w e l l as the use of wild food prices will be lower inMalawi, which will cause foods. It i s unlikely that m a n y households have any current trade flows t o be reversed. However, there savings to draw o n or assets t o sell in the face of a i s little evidence o n the covariance of production or short-term shock after the decapitalization caused prices in the border markets of b o t h countries. by the war. The use of wild foods - a common coping mecha- Many of the food security profiles mentioned nismi nmany countries - has also been mentioned as a above also indicate that households increasetheir wage household coping strategy i n Mozambique. T h i s i s employment andmigration i nresponse to a food short- actually one of a range of ways i n which consumers age. However, the empirical evidence o n the scale of respond to changing income and price signals, and such activities i s scanty. Indeed, given the scarcity of govemments have occasionally designed safety n e t non-agricultural wage employment in r u r a l areas programs with such self-selected consumption pat- where there are n o donor-funded projects, it i s diffi- terns i nmind (Alderman and Lindert, 1998). cult to see h o w they can increase their wage employ- Evidence f r o m the distribution and marketing ment other than through migration. In the southem of food aid in Mozambique during the early 1990s Lkt DEVELOPMENT AFRICAREGIONHUMAN SERIES WORKINGPAPERS illustrates the potential for using self-selection as an m a l feed. This, perception has limited i t s use in effective targeting mechanism for food subsidies. programs designed for less extreme situations than One study of the food aid program found that con- the Mozambique example. sumers w h o preferred to buy white maize rather than The same general point, that governments can yellow maize when their prices were equal w o u l d use consumers' o w n preferences to target subsidies have bought yellow maize i f it h a d been cheaper than or tax exemptions could, in principle, be applied to white despite t h e i r i n i t i a l preference. (Tschirley, other items that consumers buy. Indeed, the Gov- Donovan, and Weber, 1996). Specifically, Tschirley ernment of Mozambique has recently chosen t o ex- et a1 found that a discount of as little as 14 percent o n empt maize and bread f r o m VAT, although, as in yellow maize w o u l d have prompted a quarter of all Zimbabwe, those consumers with higher incomes consumers to shift their purchases f r o m white maize will receive higher implicit savings f r o m the tax ex- to yellow, with low-income consumers being more emption in absolute value than those with lower willing than others to shift. There i s similar evidence incomes. However, the V A T exemption will b e f r o m Zimbabwe and Zambia (Tschirley, Donovan, w o r t h more as a share of income t o low-income and Weber, 1996 and Zimbabwe, 1994) regarding households relative t o wealthier ones. Indeed, there the willingness of consumers t o purchase yellow i s n o one i t e m that i s consumed by all households maize under plausible price regimes. ineither Mozambique or Zimbabwe that w o u l d pro- While some of this evidence was obtained by vide a vehicle for a progressive subsidy (or tax ex- presenting consumers with a hypothetical choice, emption). y e l l o w m a i z e f r o m b o t h t r a d e a n d a i d in Mozambique averaged 348,000 tons a year between Widowhood and Family Break-up 1989-1990 and 1994-1995 (286,000 tons excluding the 1992-1993 drought year). Inthe comparable period, U n l i k e other Southern African countries, in white maize production averaged 384,000 tons (or Mozambique neither customary n o r religious l a w 434,000 tons w h e n the drought year i s excluded). i s recognized by the state (WILSA, 1996). This im- This indicates that marketing channels were capable plies that the majority of women in Mozambique of handling a substantial volume of legal and quasi- have n o legally protected rights of inheritance. In legal trade f r o m the ports t o the interior of the coun- contrast, the l a w i s quite inclusive with regard to try. There i s some evidence also that the pricess for children. By law, there are n o illegitimate children yellow maize and for white maize were linked, al- in the country; regardless of t h e legal status of a though n o t during the drought period w h e n the marriage, any child that i s registered has the same price of white maize, which was scarce due t o the legal rights as any other child. drought, rose substantially above the price of yel- T w o actions can be taken under the existing l o w maize (Tschirley, Donovan, and Weber, 1996). legal regime t o protect the rights of women and chil- Thus, this i s a n example of using self-targeting dren. First, simply registering children can estab- v i a the subsidy of yellow maize in a program of di- l i s h their rights for child support. Second, since a saster management. I t i s less clear, given the data marriage can be registered after the fact, a w o m a n o n consumer preferences, whether there i s a poten- in a customary marriage has the opportunity t o ob- tial niche market in Mozambique or elsewhere in in tain the rights granted to a c i v i l marriage, provided Africa for private trade in yellow maize at non-sub- that her husband has not registered another mar- sidized c.i.f. prices. However, despite these indica- riage. Indeed, until recently the government was tions that consumers m a y be willing t o purchase conducting a campaign t o draw more families un- yellow maize, policymakers in most countries be- der the legal umbrella. lieve that it w o u l d be politically costly for govern- Similarly, the revised 1996 L a n d L a w also en- ments t o distribute the commodity as part of a food shrined women's rights to o w n and inherit title t o a i d program since it i s regarded as mainly for ani- land, in keeping with Articles 66 and 67 of the 1990  25 OF PRIVATE IMPLICATIONS NETS SAFETY FOR PUBLICPOLICY Constitution that recognized the equality of w o m e n c l u d i n g the rights to fields a n d the ownership of a n d men. Thus, in principle, the revised L a n d L a w trees, by means o f oral w i l l s as w e l l as by outright has strengthened security o f t e n u r e for w o m e n transfer during the husband's lifetime. Nonetheless, (Devereux a n d Palmero, 1999). this is not automatic within local custom. The exist- However, the vast majority of individuals are ing research into the r i g h t s of widows, however, unaware of their rights u n d e r f a m i l y l a w (WILSA, consists m a i n l y of case studies a n d is not sufficiently 1996). A l t h o u g h NGOs p r o v i d e some legal assis- broad to a l l o w u s t o m a k e a n assessment o f the ac- tance, this tends to be m a i n l y in the form of media- t u a l protection that such ownership affords. Nor tion in community disputes. Therefore, inheritance are the case studies sufficiently unequivocal t o in- I is still determined largely by customary norms. dicate whether the disruptions o f the w a r a n d the The n o r m s g o v e r n i n g m a r r i a g e (and hence subsequent--albeit still limited--changing nature o f w i d o w h o o d ) differ b r o a d l y between the southern markets are u n d e r m i n i n g the traditional obligations a n d northern parts o f the country. For example, the towards w i d o w s in either the n o r t h or the south. payment of a bride price (locally called lobolo) is m o r e Moreover, according to Platteau a n d Baland (1998), c o m m o n in the south t h a n in the north. Also, the traditional n o r m s evolve w h e n land becomes scarce. northtends t o be matrilineal w h i l e the south i s patri- In such circumstances, there m a y b e a g r o w i n g ten- lineal. This distinction generally affects w h e r e a dency for families to restrict w i d o w s ' rights of ac- w i d o w (and her children) will l i v e after the death cess t o their dead husbands' property. The W o r l d of her husband a n d whether she i s expected t o m a r r y B a n k ' s 1996-1997 P o v e r t y A s s e s s m e n t f o r her deceased husband's brother. I t also influences Mozambique indicates that a quarter of a l l w i d o w s the type o f property that a w o m a n will be a l l o w e d interviewed in a s m a l l a n d specialized survey o f t o o w n outright o r the t y p e that she will have the i n f o r m a l social networks h a d lost access to fields o r right to use. However, in practice, matrilineal soci- houses or b o t h after their husbands d i e d (Dava, Low, eties in Mozambique d o not necessarily afford m o r e a n d Matusse, 1998). However, it i s not clear w h a t social protection t o w o m e n t h a n patrilineal commu- the reasons were for these losses o r whether this i s a nities (WILSA, 1996). greater proportion t h a n in the past. Traditionally, the p a y m e n t of lobolo sealed a In a d d i t i o n to r e l y i n g o n legal and customary contract under w h i c h a woman's children were rec- inheritance and o n kinship, single w o m e n utilize the ognized as her husband's or--in the event of h i s death support of other women, f o r example, inlabor-shar- --her husband's brother's. In the former case, t h i s ing arrangements. Insome contexts, these resemble contract gave the father custody of h i s children if a Roscas both in the nature of the rotation o f the work marriage breaks up, w h i l e in the latter case, it meant schedule a n d in the element of m u t u a l support. that the children's guardian was their uncle rather The analogy can b e carried further as the sharing o f than their mother. The b r i d e price compensated the labor allows a w o m a n w h o cannot c o m m a n d labor wife's birthfamily for her lost labor a n d this resource f r o m within h e r h o u s e h o l d to carry o u t various frequently was used to h e l p the bride's brothers to tasks. Elderly w o m e n are less l i k e l y to participate obtainwives. Also, u n d e r the n o r m s of this commu- insuch exchanges since it i s harder f o r t h e m to make nity, w i d o w s seldom h a d the right to freely choose a contribution. However, inthose places where A I D S a n e w husband. Indeed, remarriage outside the fam- has not devastated a generation, they are more likely ily generally j e o p a r d i z e d h e r r i g h t s t o t h e f i r s t to have children who can assist t h e m o n their farms. husband's property, as m o s t customary inheritance This m u t u a l exchange of labor differs from those n o r m s a i m to keep assets within a family. exchanges (such as ganho-ganho) which, although In the northern areas of Mozambique, w o m e n between neighbors, are exchanges of labor for f o o d have greater rights to p r o p e r t y in their own n a m e or cash and, thus, resemble a casual labor market. t h a n in the rest of the country. Moreover, there i s a W h i l e w i d o w e r s also face economic stress d u e trend t o w a r d transferring p r o p e r t y t o women, in- either t o the loss of their wives' contribution t o the 26 AFRICAREGIONHUMAN WORKING DEVELOPMENT SERIES PAPERS household or due t o the costs involved in remarry- cial has explicitly discouraged institutional care by ing, there i s comparatively little in the literature either the government or NGOs. about any assistance provided t o widowers by in- In summary, the government's strategy f o r formal safety nets. dealing with orphans i s in keeping with i t s overall Participatory assessments have shown, h o w - national approach to social protection. The Govern- ever, that communities recognize that they have a ment of Mozambique explicitly recognizes the lim- responsibility for providing for the children of such i t e d resources at i t s disposal for funding transfer widowers as w e l l as orphans and disabled individu- programs and income support. Implicitly, it also als. As inmany countries in the region, the number recognizes the need t o share the responsibility for of orphans i s predicted t o increase dramatically f r o m risk management with NGOs and c i v i l society. T h i s the already high level of 10-12 percent of children approach i s apparently based not only o n a frank under the age of 15 (UNICEF, 1999). The current assessment of fiscal constraints but also a desire t o base reflects the recent war; while the projections avoid creating a culture of dependency o n govern- stem f r o m the fact that Mozambique has n o t been ment programs. Thus, the Instituto Nacional de able t o avoid the AIDS epidemic. Given that labor Accao Social's currently believes that a n increased i s scarce inmany r u r a l areas of the country, it i s com- share of i t s expanding caseload can be supported m o n for childless families t o take responsibility for through public works programs with direct trans- double orphans, w h o m a y account for 10 percent of fers being limited largely t o the elderly and the dis- the current p o o l of orphans. Indeed, few problems abled. similarly, the government's disaster relief were reported in the placement of the orphans policy i s based o n a n assessment of the capacity of caused by the c i v i l war. However, there i s some local communities t o mitigate the impact of five cat- anecdotal evidence that child-headed households egories of emergencies. Both components of t h i s are appearing in urban areas. Thus, there is some social protection strategy rely o n extensive donor early indications of a partial breakdown of tradi- support a n d the assistance of NGOs to deliver ser- tional coping strategies. vices. The government has n o t earmarked appre- ciable funds for AIDS-related care, relying instead mainly o n donor and NGO funding (UNICEF, 1999). Notes to Chapter I11 As in Zimbabwe, the government of Mozambique sees community approaches as the key t o i t s strat- 1 I t is a surprisingly common error t o assume egy for dealing with the needs of orphans. Indeed, that matrilineal societies are intrinsically less patri- while the government has n o t yet explicitly deter- archal than patrilineal communities. mined h o w i t will foster community care (UNICEF, 2 This i s c a l l e d xivunga in G a z a a n d 1999), the Ministerio d a Coordenacao de Accao So- holominhana in Nampula.  27 OF PRIVATE IMPLICATIONS NETS SAFETY FOR PUBLICPOLICY Chapter I V Conclusion One major advantage of community-based so- service delivery. These, or similar structures, can also cial assistance and co-insurance i s that it minimizes be used t o provide social protection. the problems of asymmetry o f i n f o r m a t i o n a n d To be sure, s u c h structures need local account- moral hazard that makes a formal system of insur- ability and proper incentives--including a h a r d bud- ance prohibitively expensive. Much of the litera- get constraint--to reach their potential. In fact, the ture o n the decentralization of government services advantages of decentralizing social protection are postulates similar advantages in terms of informa- less clear w h e n a fully funded entitlement i s being t i o n f l o w s a n d accountability (Oates, 1972 a n d promoted than they are w h e n a community i s re- I n m a n and Rubenfeld, 1997). quired t o choose i t s priorities given a very limited This implies that one w a y t o capture some of budget. The former gives the community less incen- the benefits of informal systems while tapping the tive t o use i t s information. Moreover, the mixed state's ability to transfer resources over regions and global evidence o n decentralization a n d equity over time i s to use community structures as part of a makes few generalizations valid and provides a n decentralized program. The revitalization of the argument for closely monitoring decentralized ini- Zunde Ra Mambo i n Zimbabwe has some of that fla- tiatives. However, the successful use of local par- vor as--in a quite different context--has the involve- ticipation in targeting drought relief in Zimbabwe ment of councils of elders inUzbekistan in distribut- (Kaseke, Dhemba, a n d Gumbo, 1997) and elsewhere ing public social assistance (Coudouel, Mamie, and (Sharp, 1997) hints that t h i s m a y be possible. Micklewright, 1998). In these t w o examples, struc- However, one additional caveat i s important. tures that predate the state's intervention are being As I n m a n a n d Rubenfeld (1997) noted, decentrali- used to distribute social assistance programs. Even zation m a y achieve efficiency but m a y n o t neces- structures set up by NGOs or the government for the sarily be the best system t o guarantee rights defined express purpose of providing a service c a n benefit ina constitution or national legal code. For example, f r o m having access to local information if they are given that reforms to the legal codes have recently decentralized. Social funds are one w a y inw h i c h the expanded woman's rights t o i n h e r i t or be supported W o r l d Bank has encouraged the decentralization of in a number of African countries compared to those 28 DEVELOPMENT AFRICAREGIONHUMAN PAPERS WORKING SERIES accorded t o t h e m by custom (Gopal a n d Salim, vate initiatives. I t s challenge i s t o devise ways t o 1998), i t i s not clear that decentralization without bolster community programs without undermining countervailing checks w o u l d protect these rights. the norms that have encouraged such fostering in I f national policymakers a i m to preserve in- the absence of state support. This m a y include in- formal social protection mechanisms, they must con- creased community involvement in projects by pro- sider the sensitive issue of possible conflicts between viding co-funding and assisting with inputs t o be rights and custom. This i s made even more com- used in community farming with l a n d allocated by plex by the fact that customs are evolving in re- chiefs for the support of orphans. Alternatively, i t sponse t o changing economic and social circum- m a y involve funding day care and feeding centers stances. Moreover, since colonial rule, local leaders t o reduce some of the burden o n grandparents a n d have c i t e d custom as a means of resisting any en- other caregivers. croachment o n their powers. Thus, custom can be a There i s relevant experience in Sub-Saharan political tool as w e l l as a n expression of community Africa--Malawi has taken the lead in promoting the values. community care of orphans--that points t o b o t h the Practically speaking, one w a y for government potential and the limitations of s u c h a n approach policy to handle rights of inheritance and property for dealing with the consequences of the AIDS epi- settlement w o u l d be t o encourage mediation in con- demic. I t m a y be necessary to support the provi- flicts over rights of maintenance and succession. sion of psychological counseling--to community Moreover, since the central legal system can serve members overwhelmed by the task as w e l l as t o or- as a w a y t o discourage the abuse of customary phans--and technical assistance t o encourage the de- norms (Khadiagala, 1998), measures t o increase ac- velopment a n d expansion of community organi- cess to the legal system are ultimately tools of social zations. The aggregate burden of caring for d u a l protection policy. That is, they may explicitly at- orphans m a y not be large - all but the most pessi- tempt t o replace the autonomy of customary norms mistic projections place this category as only a f e w with mandated rights. If successful, timely legal percent of the age cohort--but the need t o distribute support--either as a direct government program or this burden across a community will require inno- through NGOs--may do more t o avoid destitution vation and local solutions. than w o u l d the provision of cash transfers t o wid- There i s substantial evidence that low-income ows and abandoned women. In addition t o direct households take advantage of market opportunities involvement in guaranteeing access t o courts, gov- t o mitigate fluctuations in their crop yields or in ernments and NGOs can enhance the rights of wid- product prices. T h i s case study has highlighted t w o ows and separated women (as w e l l as their children) examples of national policies that m a y restrict the by assisting them with the registration of marriages ability of households t o manage risk--the limiting and of births where such registration i s a prerequi- of the access that hammer m i l l s have t o maize stocks site for someone to claim h i s or her legal rights. in Zimbabwe and restrictions o n cross-border trade A m o n g other specific measures that take i n t o in Mozambique. M o r e generally, existing policies account the interplay of private and public social a n d institutions that relate t o national or regional protection, policymakers should consider the issue food self-sufficiency must be reviewed t o ensure that of child fostering. On average i t costs 10 times as they take into account household food security a n d much to maintain a child in a n orphanage as in a r i s k management. foster home (World Bank, 1998b). The expense of There i s also evidence that low-income house- institutional care as well as i t s general detrimental holds in Africa use financial institutions as a means consequences for a child's development are good of coping with economic uncertainty. The promo- arguments for the government to f i n d ways t o sup- t i o n of such institutions both accelerates economic port families taking care of orphans. However, the growth and helps households to manage their o w n government should a i m to avoid crowding out pri- risk. However, as i s virtually always the case, there  29 IMPLICATIONS OF PRIVATE NETS SAFETY FOR PUBLICPOLICY i s a caveat, which i s that informal savings and credit that are n o t generally considered t o be social pro- institutions are designed t o serve the needs of their tection measures. Clearly, investments in roads clients. When these institutions are required to take and irrigation are in this category. I t i s rare that development objectives i n t o account, i t i s possible such investments achieve their objective of promot- that they m a y become less flexible t o meet house- ing g r o w t h p r o m o t i o n w i t h o u t also p r o v i d i n g as- holds' savings requirements (Rutherford, 2000). In sistance t o households in mitigating the effects of designing programs that a i m to promote savings and economic shocks. Therefore, i f the designers of credit mechanisms for low-income communities, these investment programs explicitly take t h i s di- policymakers need t o be sensitive t o the fact that mension i n t o consideration - especially w h e n mo- multiple objectives should be carefully balanced. b i l i z i n g c o m m u n i t y participation - t h i s w i l l en- Finally, there are a number of steps by which hance the r i s k r e d u c t i o n a n d g r o w t h impact of governments can h e l p households t o manage r i s k these interventions. 30 AFRICAREGIONHUMAN WORKINGPAPERS DEVELOPMENT SERIES and Maputo City" in Poverty and Well-Being i n References Mozambique: The First National Assessment 1996-7. Maputo: Ministry of Planning and Finance, Processed. Deaton. 1989. "Savings inDeveloping Countries: Theory and Review". World Bank Economic Review. 3(Supple- Ainsworth, Martha. 1996. "Economic Aspects of Child Fos- ment):61-96. terage in Cote d'Ivoire," Research in Population Econom- Dercon, Stefan. 1998. "Wealth, Risk and Activity Choice: ics 8:25-62. Cattle in Western Tanzania" Journal of Development Alderman, Harold. 1996. 'Saving and Economic Shocks in Economics. 55(1):1-42. Rural Pakistan" Journal of Development Economics Devereux, Stephen and Alessandro Palmero. 1999. "Creat- 51(2):343-366. ing a Framework for Reducing Poverty: Institutional Alderman, Harold and Kathy Lindert. 1998. "The Potential and Process Issues in N a t i o n a l Poverty Policy: and Limitations of Self-targeted Food Subsidies" World Mozambique Country Report" Institute of Develop- Bank Research Observer 13(2): 213-230. ment Studies, University of Sussex. Processed. Alderman, Harold and Christina Paxson. 1994. "Do the Poor Fafchamps, Marcel, Christopher Udry, and Katherine Insure? A Synthesis of the Literature on Risk and Con- Czukas. 1998. "Drought and Saving in West Africa: sumption inDeveloping Countries" inBacha (ed.). Eco- Are Livestock a Buffer Stock?" Journal of Development nomics in a Changing World, Vol. 4 "Development, Trade, Economics. 55(2): 273-305. and Environment." London: MacMillan. Foster, G. 1998. "Today's Children - Challenges to Child Care Alwang, Jeffrey and L i r e Ersado. 2000. "Changes inPoverty Promotion in countries w i t h Severe AIDS Epidemics" in Zimbabwe: 1990-1996." Draft. Africa Human Devel- AIDS Care 10 (supplement): 17-23. opment Group 1 , World Bank, and Virginia Institute Gertler, Paul and JonathanGuber. 1997. "Insuring Consump- of Technology. tion against Illness" NBER: Processed. Bendokat, Regina and Mauriza Tovo, 1999. A Social Protec- Gopal, Gita and Maryam Salim. 1998. Gender and Law: East- tion Strategy for Togo. Social Protection Discussion Pa- ern Africa Speaks. (Washington D.C.: The World Bank) per #9920. (Washington D.C.: The World Bank). Grimard, Franque. 1997. "Household Consumption Smooth- Besley, Timothy. 1995. "Non-market Institutions for Credit ing through Ethnic Ties: Evidence from Cote d'Ivoire" and Risk-Sharing in Low-income Countries" Journal of Journal of Development Economics. 53: 391-422. Economic Perspectives 9: 115-127. Hoddinott, John and William Kinsey. 1998. "Child Growth Binswanger, Hans and Mark Rosenzweig. 1993. "Wealth, in the time of Drought" IFPRI: Processed. Weather Risk, and the Composition and Profitability Holleman, J. 1975. Shona Customary Law. Manchester: of Agricultural Investments" Economic Journal 103:(1): Manchester University Press. 56-78. Inman, Robert and Daniel Rubenfeld. 1997. "Rethinking Browning, Martin. 1994. "The Savings Behavior of a Two- Federalism" Journal of Economic Perspectives. ll(4). person Household." Department of Economics, Jayne, T. S. and Stephen Jones. 1997. "Food Marketing and McMaster University: Hamilton, Ontario, Canada. Pricing Policy in Southern Africa: A Survey" World Mimeo. Development. 25(9): 1505-1527. Case, A n n e and Angus Deaton. 1998. "Large Cash Transfers Jensen. 1999. Public Transfers, Private Transfers and the Crowd- to t h e Elderly in South Africa," Economic Journal ing Out" Hypothesis: Evidencefrom South Afnca. Harvard 108(3):1330-1361. University. Processed. Cavendish, William. 1997 "The Economics of Natural Re- Kaseke, Edwin, Jotham Dhemba, and Perpetua Gumbo. 1997. source Utilization by Communal Area farmers of Zim- Transfem'ng Resources to Poor Households: the Case of Safety babwe" Unpublished D. Phil. thesis, University of Ox- Nets in Zimbabwe, MPSLSW and UNICEF: Harare. ford. Khadiagala, Lynn. 1998. "Seeking Justice in Uganda: The . 1999. Incomes and Poverty in Rural Zimbabwe during Adjudication of Women's Property Rights and The Adjustment: the Case of Shindi Ward, Chivi Communal Failure of Popular Justice." Paper presented at the Area, 1993/4 to 1996/7. Oxford University. Centre for American Political Science Association Annual Meet- the Study of African Economies. ing, Boston, Massachusetts. Corbett, Jane. 1988. "Famine and Household Coping Strate- Kinsey, William, Kees Burger, and JanGunning. 1998. "Cop- gies" World Development 16(9): 1099-1112. ing with Drought in Zimbabwe: Survey Evidence on Coudouel, Aline, Sheila Mamie, and John Micklewright. Response of Rural Households at Risk." World Devel- 1998. Targeting Social Assistance i n a Transition Economy: opment. 26(1):89-110. The Role of the Mahalla in Uzbekistan. UNICEF Innocenti Kozel, Valerie and Barbara Parker. 1999. "Poverty in Rural Occasional Papers #63, Florence. India: The Value of Qualitative Research in Poverty Cox, Donald and Emmanuel Jimenez. 1998. "Risk-Sharing Analysis" World Bank. Processed. and Private Transfers: What About Urban House- Low, Jan, James Garrett, and Vitoria Ginja. 1998. "Formal holds?" Economic Development and Cultural Change 47. Safety Nets in an Urban Setting: Lessons from a Cash Dava, Gabriel, Jan Low, and Cristina Matusse. 1998. "Infor- Transfer Program" in Poverty and Well-Being in mal Mechanisms of Mutual Help and Social Protec- Mozambique: The First National Assessment 1996-7. tion: A Case Study from Gaza and Nampula Provinces Maputo: Ministry of Planning and Finance. Processed.  31 IMPLICATIONS OF PRIVATE SAFETYNETS FOR PUBLICPOLICY Lucas, Robert and Oded Stark. 1985. "Motivations to Remit" Udry, Christopher. 1994. "Risk and Insurance in a Rural journal o f Political Economy. 93:901-918. Credit Market. An Empirical Investigation in North- Morduch, Jonathan. 1999. "Between the Market and the State: ern Nigeria" Review of Economic Studies. 61(3): 495-526. Can Informal Insurance Patch the Safety Net?" World UNICEF. 1998. Community Based Orphan Assistance in Zim- Bank Research Obseruer. 14(2):187-207. babwe: Developing and Expanding National Models by Mozambique. 1998. Poverty and Well-Being in Mozambique. Building Partnerships with NGOs, CBOs, and the Private The First National Assessment 1996-7. Maputo: Minis- Sector. Harare: Processed. try of Planning and Finance. Processed .1999. Orphan Programming i n Mozambique: Combin- Mutangadura, Gladys and Ephias Makaudze. 1999. Urban ing Opportunities for Development with Prevention and Vulnerability to Chronic Poverty and Income Shocks and Ef- Care. Maputo: Processed. $ctiveness of Current Social Protection. Harare: M i n i s t r y of Webb, Patrick and Thomas Reardon. 1992. "Drought Im- Public Service, Labor and Social Welfare. Processed. pact and Household Response in East and West Af- Ncube, et al. 1997. Paradigms of Exclusion. Women's Access to rica" Journal of Development Studies. Resources i n Zimbabwe. Harare: Woman and Law in Whiteside, Martin. 1998. When the Whole is More than the Southern Africa Research Trust. Sum of its Parts. Cross-border Interactions on Livelihoods Ncube W. and Stewart. J. 1997. Continuity and Change: The in Southern Malawi and Northern Mozambique. Report Family In Zimbabwe. Harare: Woman and Law inSouth- for Oxfam: UK e r n Africa Research Trust. WILSA. 1996. Right to Succession and Inheritance. Woman and Oates, W. 1972. Fiscal Federalism. N e w York: Harcourt, Brace Law in Southern Africa Research Project, Maputo, and Jovanvich. Mozambique Platteau, Jean-Phillippe and Jean-Marie Baland. 1998. Im- World Bank. 1998a. World Development Report. Oxford Uni- partible Inheritance Versus Equal Division: A Comparative versity Press: N e w York Perspective Centered on Europe and Sub-Saharan Afnca. .1998b. Confronting AIDS. Public Priorities in a Global Namur: Centre de Recherche en Economie du Devel- Epidemic. Oxford University Press: N e w York opment. .1999a. Safety Nets inMalawi: H o w Much?For H o w Rosenzweig, Mark. 1988. "Risk, Implicit Contracts and the Many? World Bank: Processed. Family in Rural Areas of Low-Income Countries" Eco- . 1999b. "Project Concept Document o n a Proposed nomic Journal. 98: 1148-1170. Loan in the Amount of US$250 Million Equivalent to Rutherford, Stuart. 2000. The Poor and Their Money. Oxford the Federative Republic O f Brazil for a Program to University Press, N e w Delhi.. Eradicate Child Labor" World Bank. Processed. Sharp, Kay. 1997. Targeting Food Aid i n Ethiopia. Addis Ababa. Zimbabwe. 1995. Community Based Care for Orphans: Phase Save the Children, UK 11. Processed. Masvingo Province, Department of So- Stewart, Julie and Welshman Ncube. 1992. "Legal and Cul- cial Welfare. tural Framework of Succession in Zimbabwe" inInher- . 1994. The Impact of Maize Market Liberalization in itance Law in Southern Africa. Harare: Woman and Law Zimbabwe's Urban Areas. Ministry of Public Service, in Southem Africa Research Project. Working Paper Labor and Social Welfare: Harare. #6. . 1998a. Poverty i n Zimbabwe. Central Statistics Of- Tschirley, David, Cynthia Donovan, and Michael Weber. fice: Harare. 1996. "Food Aid and Food Markets: Lessons from . 1998b. HIV/AIDS in Zimbabwe. Ministry of Health Mozambique." Food Policy. 21(2): 189-209. and Child Welfare: Harare. 32 AFRICAREGIONHUMAN WORKING DEVELOPMENT PAPERS SERIES Other titles published in the Africa Region H u m a n Development Working Paper Series thus far: Engaging with Adults: The Casefor Increased Support to Adult Basic Education in Sub-Saharan Africa, by Jon Luaglo, February 2001.