Gender Norms in Financial Markets: Evidence from Kenya

World Development Vol. 32, No. 8, pp. 1355–1374, 2004� 2004 Elsevier Ltd. All rights reserved

Printed in Great Britain0305-750X/$ - see front matter

lddev.2004.03.003
www.elsevier.com/locate/worlddev
doi:10.1016/j.wor

Gender Norms in Financial Markets:

Evidence from Kenya

SUSAN JOHNSON *

University of Bath, Bath, UK
Available online
Summary. — The role of institutions––rules and norms––in markets is increasingly recognized indevelopment discourse. This paper considers the role of gender relations for rules and norms infinancial markets. Using evidence from Central Kenya it develops a framework for establishing theinfluence of gender on the demand for and access to financial services, so explaining the genderdifferentiated use of rotating savings and credit associations (ROSCAs). It, first, analyzesintrahousehold norms related to income and expenditure flows and their management, soidentifying gendered patterns of demand. Second, by conceptualizing financial intermediaries asoperating within rules and norms, it allows the influence of gender relations on access to financialservices to be more systematically investigated.� 2004 Elsevier Ltd. All rights reserved.

Key words — gender, microcredit, ROSCAs, savings, Kenya, sub-Saharan Africa

*This paper draws on research carried out under the

Finance and Development Research Programme funded

by the UK’s Department for International Development

(see www.devinit.org/findev) and co-coordinated by

Colin Kirkpatrick at IDPM, University of Manchester. I

also acknowledge support to develop the paper from the

ESRC Research Group on Wellbeing in Developing

Countries at the University of Bath. I am particularly

grateful for comments on earlier drafts from Haroon

Akram-Lodhi, Daphne Chang, James Copestake, Bar-

bara Harriss-White, Allister McGregor, Nici Nelson,

Sarah White, and Dorothy McCormick and her col-

leagues at the Institute of Development Studies at the

University of Nairobi, along with the comments of

anonymous reviewers. Particular thanks are also due to

Samuel Kareithi, Edward Kinyungu, Margaret Maina

and John Muriuki for research assistance. Comments

and suggestions from many other colleagues have

also been invaluable. The author is however entirely

responsible for errors and omissions. Final revision

accepted: 23 March 2004.

1. INTRODUCTION

There is increasing recognition of the impor-tance of institutions––rules and norms––inmarkets, not least from leading actors such asthe World Bank. However, this interest has yetto result in a systematic analysis of the influenceof gender as a social institution that patternsfinancial markets. Rather, it is suggested thatmicrocredit schemes targeted to women havebeen designed in a way that means that gendernorms have been successfully overcome.The purposes of this paper are twofold. The

first is to investigate the influence of gender onthe use of a range of financial services in a localfinancial market. The paper therefore presentsevidence on patterns of financial service use inthe financial markets surrounding the town ofKaratina in Central Kenya which show that theuse of rotating savings and credit associations(ROSCAs), savings and credit co-operatives(SACCOs) and borrowing from friends andrelatives are gendered. While the explanationfor gendered SACCO use is readily apparent asit is linked to male control over cash crops,gendered ROSCA use and borrowing fromfriends and relatives require further explora-tion. Since ROSCAs were the most widely usedservice in this research I focus primarily on this

135

finding and reflect further on borrowing fromfriends and relatives in the conclusion.The widespread use of ROSCAs in the

financial markets of many developing countriesis well documented (Adams & Fitchett, 1992;Bouman, 1995; Geertz, 1962) including their

5

WORLD DEVELOPMENT1356

extensive use by women in Kenya (Anderson &Baland, 2002; Kimuyu, 1999; Nelson, 1995)and by women in many other parts of the world(Ardener & Burman, 1995). Ardener notes anumber of features of women’s involvement inROSCAs: (a) the fact that social sanctions suchas disgrace and shame act as mechanisms whichforce people to repay; (b) the value membersplace on the social networking and solidaritywhich ROSCAs offer; (c) the fact that women’sROSCAs tend to have smaller and lower con-tributions than men’s; (d) that women oftenprefer single-sex ROSCAs and will activelyexclude men; and (e) the opportunities forsecrecy that ROSCAs offer in savings accumu-lation. But, Ardener herself makes a plea for amore detailed gendered analysis of ROSCAuse, thereby suggesting that while this much isknown, there is a need to answer a deeperquestion: how can this gender differentiation inROSCA use be better explained?The second objective is therefore to explain

this gender-differentiated pattern of ROSCAuse by developing an analytical approach thathas the potential to offer a more systematicexplanation of the role of gender in patterningthe access and use of financial services, andhence the operation of financial markets ingeneral. The analysis has two components.First, an analysis of the gendered demand forfinancial services. Approaches to theorizingsavings and credit in economics treat savings asa residual after consumption, and credit pri-marily as an input into a production function.Here I treat financial services––whether savingsor credit––as a means of making the time-basedtransition from income to expenditure forindividuals rather than households. Further,drawing on the extensive research analyzing theintrahousehold dynamics of production andconsumption in developing countries I analyzeintrahousehold resource allocation throughexamination of the control of income fromhousehold activities; the gender division ofexpenditure responsibilities and the way intra-household financial management actually takesplace, in order to enable gendered patterns ofdemand for financial services to be identifiedand understood. This analytical approach isused with data from Kenya to show why theflows of income and expenditure women man-age are better suited to the type of financialservice that ROSCAs offer, compared to thosethat men manage.Second, the analysis turns to the supply side.

Financial services are provided by a range of

types of financial intermediary such as banks,parastatals, savings and credit co-operatives,moneylenders, rotating savings and creditassociations and so on. These operate withinoverlapping sets of rules and norms––bothinternal to the intermediary itself, and withinwider sets of legislated rules and societal normssuch as those of shame and reputation.Whether men or women are able to engage withthese rules and norms, and how they do so,determine which services they are able to use.Hence the supply side of the market is alsogendered due to the design of financial inter-mediaries. Data from Kenya show the differingability of men and women to engage with therules required for ROSCAs to function effec-tively and hence further explains gender differ-entiated use.The paper concludes by considering the

implications of the approach and its findingsfor analysis of gender in financial markets and,finally, for current debates about the empow-erment impact of microfinance and consequentapproaches to practice.

2. GENDER, CREDIT ANDINTRAHOUSEHOLD RELATIONS

Recent development discourse has started torecognize the role of institutions––rules andnorms––in markets, and this is demonstratedby the World Bank’s World DevelopmentReport of 2002 (WDR, 2002) entitled ‘‘BuildingInstitutions for Markets’’ (World Bank, 2002).In addition the World Bank has also broughtthis emphasis on institutions into its analysis ofgender. ‘‘Engendering Development’’ (WorldBank, 2001) states that

Societal institutions––social norms, customs, rights,laws––as well as economic institutions, such as mar-kets, shape roles and relationships between men andwomen and influence what resources women andmen have access to, what activities they can and can-not undertake. . . Even when formal and informalinstitutions do not distinguish explicitly betweenmales and females, they are generally informed (eitherexplicitly or implicitly) by social norms relating toappropriate gender roles (p. 13).

But, while the WDR (2002) dedicates achapter to the role of norms and networks inmarkets, and makes repeated reference to therole of social norms in access to resources andrights, gender (or gender differentiated norms)merits only three mentions, and the chapter on

GENDER NORMS IN FINANCIAL MARKETS 1357

financial systems makes no reference to genderat all. Indeed, analysis of gender bias in relationto the financial sector as a whole has beenobserved to be relatively undeveloped (vanStaveren, 2001).Examples of microcredit programs tend to be

used as an antidote to this lack of systematicanalysis of the interaction of gender relationsand institutional arrangements in financialmarkets. ‘‘Engendering Development’’ makesuse of examples from the microfinance field toillustrate the ways in which microcredit pro-grams and informal financial systems may bedesigned to better enable access by women. Theoverall message is that in financial marketswomen lack access to collateral largely becauseof gender norms surrounding women’s access toproperty, and that microcredit programsthrough the use of alternative forms of collateral,along with simplified application procedures andthe reduced need to travel, have overcome theseconstraints (WDR, 2001, p. 173).This apparently simple story is questioned

however by recent debates over the impact ofmicrocredit programs on women’s empower-ment that start to reveal the complexity andrange of impact on gender relations thatintroducing loans into the household can have.Starting from the 1980s, the targeting of credittoward women was justified by a ‘‘publictranscript’’ (Rahman, 1999, p. 69), especially inthe context of Bangladesh, in which loans towomen enabled them to increase their earningswhich would be used more effectively to sup-port family welfare than those of men (pri-marily an efficiency driven argument) while atthe same time enabling women to be empow-ered in the wider societal context. A number ofstudies offered evidence for this impact path-way (Amin, Becker, & Bayes, 1998; Hashemi,Schuler, & Riley, 1996; Khandker, 1998). But astrong challenge to this narrative came in theform of an analysis of women’s control overloan use in Bangladesh, which indicated that inthe majority of cases women did not themselvescontrol the use of the loan, rather that a womanwas likely to hand it over to her husband(Goetz & Gupta, 1996, p. 53). Subsequentstudies using anthropological methods havefurther revealed the diverse nature of intra-household impacts and reported that increasedmale violence against women, both verbal andphysical is possible (Rahman, 1999; Schuler,Hashemi, & Badal, 1998).Kabeer seeks to resolve the problem of these

starkly contradictory findings by pointing out

that the research methodology matters, arguingthat negative evaluations tend to focus onantagonistic intrahousehold relations ratherthan co-operative ones while quantitativestudies do not impose such a preference (Kab-eer, 2001). Her study therefore asked womenthemselves to evaluate the impact and sheconcludes from their testimonies that loans hadgiven women the scope to improve their bar-gaining position within the household. Thisapproach does not however negate the factthat particular experiences may have beenproblematic or disadvantageous or answerthe ‘‘central question for policy orientedresearch. . . What are the conditions that enableor undermine women’s capacity to negotiate abetter deal within and beyond the household?’’(Hart, 1995, p. 64).Compared to the extensive debate triggered

in the Bangladesh context, there is relativelylittle detailed evidence on the intrahouseholdimpact of microcredit from elsewhere. Mayouxexamined seven microfinance programs inCameroon and also questions the assumptionof harmonious intrahousehold relations thatprograms most often adopt. She points out theimplications of polygyny in creating complexityboth between husbands and wives as well asamong women and of hierarchical householdand kin relations which circumscribe women’sroles and constrain their bargaining power inmarital disputes, both in patrilineal andmatrilineal systems (Mayoux, 2001, p. 453).The key point to highlight from these debates isthat many aspects of gender relations, bothwithin the household and in the wider society,come into play and mediate the impact of loansgiven to women (Mayoux, 1999), but the mi-crocredit and women’s empowerment debatehas been more concerned with evaluatingoverall outcomes than undertaking detailedanalysis of intrahousehold gender relations andfinancial arrangements in specific contexts andanalyzing the dynamic processes set in motionby the introduction of credit.This seems rather surprising given the femi-

nist critique of unitary household economicmodels which questions assumptions of incomepooling and sharing (Folbre, 1986b; Kabeer,1994), and the wealth of empirical material onintrahousehold dynamics and their implicationsfor interventions and policy––particularly inthe agricultural sector of developing countries(Guyer, 1980; Hart, 1995; Longhurst, 1982).This literature has in turn contributed toreformulated models of the household as a


collection of individuals who bargain over thedistribution of the gain in welfare derivedthrough co-operation (Manser & Brown, 1980;McElroy & Horney, 1981), although these toohave been criticized for containing highlyeconomistic notions of the bases of bargainingand co-operation in the household (Folbre,1986a; Kabeer, 1994; Sen, 1990).Despite this extensive criticism of household-

level analysis, the demand for credit tends to beseen as either an input into the productionfunction of farm households or firms, or as ameans of household consumption smoothing(Case, 1995). Nor have theories of saving beenpurposefully disaggregated by gender althoughthere is the potential to do so, for example,Friedman’s permanent income hypothesis inwhich consumption is based on the agents viewof her long-term income and short term chan-ges––or transitory income––are saved; orModigliani’s life-cycle theory in which age isthe main factor influencing saving in anticipa-tion of retirement needs (see for example Sti-glitz, 1993).It therefore appears possible and necessary to

build from earlier detailed analysis of intra-household relations to consider the implica-tions of the organization of production,consumption, and related income and expen-diture within the household for the demand forfinancial services. At the same time an analyt-ical framework is needed that can accommo-date our understanding of the ways in whichgendered norms such as those concerningproperty rights influence access to financialservices.

3. ANALYTICAL FRAMEWORK: THEINFLUENCE OF RULES AND NORMS IN

FINANCIAL MARKETS

This paper looks at the role of genderednorms from two perspectives: first, in terms oftheir influence on the demand side and, second,in terms of influencing access to financial ser-vices on the supply side.

(a) Intrahousehold gender norms and thedemand for financial services

In order to examine the influence of gendernorms on the demand for financial services, westart by examining intrahousehold relationsrelating to access and control of income andexpenditure. In order to develop this analysis, I

draw on Rutherford’s view of the way poorpeople manage their money which emphasizesthe way savings and loans facilities enable flowsof income to be turned into the ‘‘usefully largelump sums’’ of cash required to finance neces-sary expenditure (Rutherford, 1999, p. 1).Hence, financial services are seen as offeringindividuals and households a means of makingthe time-based transition from income toexpenditure (or vice versa) through either sav-ings or loans facilities.The approach used here has three steps.

First, we consider how income streams fromdifferent income generating activities in whichhouseholds are engaged enter the householdand are allocated and controlled by husbandsand wives. Second, it considers intrahouseholdpatterns of expenditure responsibilities. It doesthis by analyzing data from both husbands andwives as to their view of the balance ofresponsibilities in a number of expenditureareas. The categories offered to respondentswere five: husband’s sole responsibility; hus-band makes a major contribution and wifemakes a minor contribution; equal contribu-tions; wife makes major contribution and hus-band makes minor contribution; wife’s soleresponsibility (see also Stichter, 1987). Ofcourse, the results were not entirely consistentand husbands and wives disagreed about theextent of their relative contributions. While thiscan also be seen as evidence of bargainingbehavior, this is not the main focus of theanalysis here since the purpose is to revealbroad patterns of expenditure responsibility.‘‘Agreement’’ here simply meant that theaccounts were broadly consistent and wasanalyzed by considering how far apart theresponses were. If they were in categories thatwere only one removed from each other in theabove list, this was regarded as ‘‘minor’’ dis-agreement, if more than one category apart,this was regarded as ‘‘major’’ disagreement.The approach then considers how intra-

household financial management is organized.A study of household financial management inthe United Kingdom has classified the mecha-nisms through which husbands and wivesmanage income streams to meet necessaryhousehold expenditure (Pahl, 1989). 1 Pahl usesthe word management to indicate, ‘‘theadministration of household finances; anactivity which includes the payment of bills fordaily living expenses’’ (p. 29). The managementsystem classification does not however sayanything about the knowledge of one partner


of the income of the other although Pahldefines control as indicating overall direction ofthe finances of the household, incorporating thenotion of ownership of the money or propertyunder discussion. Burgoyne points out thatmanagement systems shade into one anotherand are difficult to classify, explaining thatdifferences may relate to attitudes about claimson income even though management remainsunder the control of one person (Burgoyne,1990). 2 This raises the critical question of therelationship between claims and control, and itis differences in attitude on these that make theline between independent management andshared management a very thin one. Never-theless, while the data did not enable theseissues to be examined in depth, the analysisdoes allow us to draw conclusions about thepatterns of access and control of income andexpenditure within the household and theimplications that these have for the demand forfinancial services that enable individuals toconvert flows of income to meet their expen-diture responsibilities.

(b) Norms, rules and access to financial services

We next consider the implication of genderrelations and gendered norms for access tofinancial services by drawing on an approach toanalyzing financial intermediation that recog-nizes the role of rules and norms.Financial intermediation requires the con-

version of savings into loans, and financialintermediaries operate within overlapping setsof rules and norms to perform this function(Johnson, 2003b). They have internal rules––orpolicies––which determine, for example, howmuch and to whom they will lend. Muchattention has been paid to the design of suchrules especially in the credit literature, high-lighting the importance of screening borrowersand enforcing contracts due to the associatedinformation asymmetries and transactionscosts (Hoff & Stiglitz, 1990; Stiglitz & Weiss,1981). Further, many of the rules that enableintermediaries to function are part of the widercontext in which they operate, for example,government regulation of the financial sector,and laws that are part of the wider formal legalsystem such as property and contract law. Inaddition, informal (i.e., not legislated) normsand sanctions also play a role.First, the formal regulation of the financial

sector through bank regulatory requirements,co-operative or parastatal laws have different

elements depending on the type of intermedi-ary. The need for bank regulation is particu-larly driven by the need to protect the interestsof depositors whose interests may diverge fromthose of the shareholders. This is usually donethrough mechanisms such as liquidity andreserve requirements, criteria about the qualityof lending, nature of collateral and so on. Thesetypes of requirement do not usually apply toparastatal lenders or to co-operatives where, inthe latter case, for example, owners and usersare the same people and have roles to play insetting the internal rules.A second set of rules within which financial

intermediaries operate are those of formalproperty and contract laws, which determinehow contracts can be designed and enforced.Whether lenders are regulated by governmentor not, they can appeal to these laws––hencemoneylenders who are not in any way regis-tered may also use property and contract lawsto enforce contracts with borrowers.Formal laws are also supported by informal

norms in providing the basis of social order(DiMaggio, 1994; Platteau, 1994). Hodgson,for example, following Durkheim, argues that

whenever a contract exists there are factors not reduc-ible to the intentions or agreements of individuals thathave regulatory and binding functions for the contractitself. . . there exists a binding set of rules to whichthere is not explicit or detailed reference by the partiesinvolved. . . [A]ll market based and contractual sys-tems thus rely on essentially noncontractual ele-ments––such as trust and moral norms––to function(Hodgson, 1999, p. 252).

These norms are often enforced throughsanctions such as guilt, shame and informa-tional sanctions, which may for example dam-age someone’s reputation (Posner & Rasmusen,1999). Such norms and sanctions are largelybased in social and cultural meaning andpractice, including for example moral and reli-gious values, and have independent motivatingpower to cause action to take place (Elster,1989). Hence they do not need external sanc-tions to be effective, as they can be internalizedand followed even if violations are not observedor exposed. Shame, or the anticipation of it,can be a sufficient internal sanction. As a result,these play a key role in enforcing rules, butactors do not necessarily respond to them aspart of a rational optimization process, thoughthey may use them to ex-post rationalize self-interest.


The key issue here is that in order to gainaccess to financial services, individuals must beable to engage with the norms and rules of anyparticular financial intermediary. It is wellknown that women who lack access to formalproperty rights or asset ownership due to gen-der norms are unable to engage effectively withthe rules of banks or lenders that require col-lateral. This approach therefore allows a meansthrough which the influence of gender normson access to different types of financial servicescan be investigated across a range of financialservice providers. This paper uses this per-spective on financial intermediation to explorewhy women use ROSCAs more than men.

4. RESEARCH METHODOLOGY

This paper draws on findings from researchthat set out to investigate the embeddedness oflocal financial markets in social relations,through investigation of access to and use arange of financial services. The methodologydeveloped to do this sought to do so in anholistic way, exploring formal and informalfinancial services, investigating both the supplyof and demand for financial services and bystarting with an open-ended view of what thekey dimensions of embeddedness might be,such as, age, gender, ethnicity, wealth and soon.In order to explore both the supply and

demand sides of a definable financial ‘‘market,’’it was necessary to identify an appropriatelocation. Karatina is a small vibrant town inMathira Division of Nyeri District in Kenya’sCentral Province. Its location on the slopes ofMt Kenya means that it has fertile volcanicsoils and good rainfall. Its choice was driven bythree factors: first, the presence of a range offormal financial services; second, the presenceof microfinance institutions 3 for a period oftime which would ensure their operations wereadequately understood by users; third, thepotential to identify Karatina as a town centreserving a reasonably well-defined rural hinter-land. Field research was undertaken in Kara-tina and two nearby rural sublocations for atotal of eight months in three visits over theperiod September 1999–February 2001.First, a supply-side survey was undertaken

which involved: (a) in-depth interviews with 37formal financial service providers and (b) theidentification and interview of informal sectorproviders ranging from groups to moneylend-

ers. Cross-sectional data on prices and volumeswere collected along with details of productsand services. Open-ended questioning exploredtrends and informant’s explanations of chang-ing provision.Second, a two-stage demand-side survey was

undertaken. A random household survey of 150households spread equally across the three keylivelihood types in the area––tea farming, coffeefarming and town-based enterprise andemployment––collected basic household-levelinformation on livelihood profiles. This datawere then used to identify a purposivesubsample of 68 in-depth individual interviews(based on location, age, sex, and marital status)in 48 households in order to investigate bothquantities and informant’s explanations of theirfinancial service use. Hence, both husband andwife were interviewed in 24 households.Data from the supply side were compiled to

establish overall market dimensions and adetailed understanding of the products on offer(Johnson, 2004). Quantitative data analysis offirst round household and second round indi-vidual interview data was carried out. While the68 individual interviews were purposefullychosen, the data were pooled to examine pat-terns of access to and use of financial services inrelation to a number of socioeconomic char-acteristics including gender, age, marital status,education, location, main income source,expenditure poverty, landholdings and otherasset holdings. The analysis used both crossta-bulations and logistical regressions to establishwhich socioeconomic characteristics wereimportant in determining use. 4

Once the importance of gender had beenestablished after analysis of the detaileddemand-side data, the final field visit included13 focus group discussions and participatoryrural appraisal (PRA) exercises held with bet-ter-off and less well-off men and women in allthree livelihood areas to explore the genderdivision of labor and control of income inrelation to the following main income sourcesin the rural areas: coffee, tea, dairy, horticul-ture, food and casual labor. Participants wereasked to make relative assessments of men andwomen’s contributions and the allocation andcontrol of income. The differential participationof men and women in ROSCAs was also dis-cussed.The findings presented below first draw on

the findings of the demand-side survey topresent an overview of financial service use forthe sample of 68 individuals, highlighting where


these results were gendered. It then goes on touse the data on married households to exploreintrahousehold financial flows supported byextensive qualitative data collected from FGDsas well as the in-depth individual interviewees.The subsequent section on ROSCA participa-tion uses qualitative data drawn from both ofthese sources and the initial open-ended inter-viewing about ROSCAs with groups and indi-viduals on the supply side.

5. GENDERED PATTERNS OFFINANCIAL SERVICE USE IN

KARATINA’S FINANCIAL MARKETS

Table 1 summarizes the findings on financialservice use. The three most used types of sav-ings service were rotating savings and creditassociations (ROSCAs––48.5%), bank savingsaccounts (45.6%) and savings and credit co-operatives (SACCO) based on a common bondof cash crop tea or coffee production (33.8%).ROSCAs are systems in which a number ofpeople form a group and contribute an agreedamount on a regular basis. At each meeting thefund is usually given to one person who takesall of the money, until everyone in the grouphas received the money in turn. The order ofrotation may be determined by ballot, by age orseniority or other social systems of prefer-ment. 5

Analysis of the use of these services by gen-der showed that ROSCAs were used by 65.7%of women but only by 30.3% of men. Thisresult was significant at the one per cent level

Table 1. Frequency of use of financial service types (%)

N ¼ 68 Savings

accounts

Loans

ROSCA 48.5 48.5

Banks/building societies 45.6 7.4

Cash crop SACCOs 33.8 19.1

Employee SACCOs 10.3 8.8

Post office 8.8 N/A

Independent ASCA 8.8 5.9

MFI 7.4 7.4

NGO managed ASCA 5.9 7.4

Parastatal lenders 0 1.5

Insurance group 39.7 N/A

Moneylender N/A 0

Loan from friend or

relative

N/A 17.6

Source: Johnson (2003b).

and logistic regressions supported the signifi-cance of gender as a determining factor. Inter-estingly there was no difference in the use ofbank savings accounts by gender, but men andwomen were significantly more likely (at 5%level) to use cash crop SACCOs than women(45.5% compared to 22.9%). Since the maincash crops are tea and coffee and men usuallycontrol these crops, it is they who are morelikely to join the SACCO. Interestingly allwidows in the rural areas were members of cashcrop SACCOs, indicating that they take overthe accounts when the husband dies.When it came to borrowing, the most used

service was cash crop SACCOs––used by 19.1%of respondents––and it was being older, lesseducated and living in the rural sublocationsthat were significant factors affecting use ratherthan gender. This reflects the fact that peoplewho have worked in employment or business(usually men) return to the rural areas to farmwhen they retire. A further factor contributingto age differentials in access is the extent of landdivision. Young people are now less able toderive an adequate income from the smallfarms they may be allocated and are forced toseek employment and business opportunitieselsewhere––sometimes leaving the managementof the tea and coffee farms to their fathers.The second most common source of loans

was friends and relatives––used by 17.6% ofrespondents. Interestingly 27.3% of men usedthis source but only 8.6% of women––signifi-cant at the five per cent level––and the logisticregressions indicated that being young, edu-cated and male significantly raised the likeli-hood of borrowing from this source. Afterthese two loan sources, 6 the use of other loansources was low: employee SACCOs (8.8%),banks and building societies (7.4%), microfi-nance institutions (7.4%) and accumulatingsavings and credit associations managed bylocal NGOs (managed ASCAs) (7.4%).It is interesting to note that there were a total

of only five loans (7.4%) from banks andbuilding societies but none of these borrowershad formally used land as collateral although inone case a title deed had been deposited in thebuilding society but not legally charged. Threewere loans to women––two through a bank runmicrocredit scheme. But the more interestingfinding of the research was the difficulty formen of using land as collateral not simplybecause of transactions cost but because wivesand other family members had to be consultedin the process of mortgaging land and were


unlikely to agree. The importance of land in thelocal social and cultural context means that thesocial relations surrounding it produce a situ-ation where it’s use as collateral is heavilyconstrained so helping to explain the overalllow use of bank loans (Johnson, 2004).In summary, gendered patterns of use were

therefore evident for ROSCAs, and cash cropSACCOs and borrowing from friends and rel-atives. As indicated above, explanations for theuse of cash crop SACCOs can easily be foundsince the norm is that men control cash cropproduction. The next sections explore the rea-sons for gendered ROSCA use, and we returnto the finding that young men make more useof borrowing from friends and relatives in theconclusion.

6. ENGENDERING THE DEMANDFOR FINANCIAL SERVICES

This section uses the analytical approachlaid out above to explore the intrahouseholddivision of labor, control over income andexpenditure and financial arrangements tounderstand the gendered demand for financialservices, and then explains how ROSCAs areparticularly appropriate to meeting thisdemand for women rather than men.

(a) Gender division of labor and control ofincome

Data from focus group discussions are usedto explore the relationship between the genderdivision of labor and control over income fromthese income-generating activities.Historically among the Kikuyu men tradi-

tionally allocated land to women for foodcrops. Women managed the land and con-trolled the output from it including incomefrom the sale of any surplus (Fisher, 1954;Leakey, 1956). This division of labor is littlechanged for subsistence food crops, but cashcrops and off-farm income are now veryimportant to the household economy. Menmainly controlled income from tea and coffeeeven though women often undertook signifi-cant amounts of work on cultivation. In better-off households, it was more likely that womenwould be given a portion of the income to useas they wished, whereas in less well-off house-holds it was more likely that a wife would haveto ask her husband for money for specificpurposes.

Income from dairy and horticulture exhibitedmore complicated patterns of allocation andcontrol that appeared to differ between the teaand coffee zones and with household wealthlevels. Men were more likely to control dairyincome in better-off households in the coffeezone compared to less well-off households, andcompared to households in the tea zone. Apossible explanation is that with the decline ofcoffee incomes in the late 1990s, dairy incomehas become a relatively more important sourceof income for these men, than those in the teazone where tea income has been more stable.Moreover, in these households, dairy incomemay constitute amounts that––if receivedthrough a Dairy Co-operative––enable men tocollect the money as a lump sum at the monthend and retain their membership of the co-operative as a potential source of loans. But,women and men in these households reportedthat some of this money is immediately given tothe wife in order to ‘‘encourage them to carryon doing the work’’ and women in this areathought that they undertook the major part ofthe work since they usually undertook the dailyjob of cutting and carrying the napier grass tostall feed them. Whereas, in the less well-offhouseholds in the coffee zone, milk was morelikely to be sold locally to neighbors or to milkbrokers who came to the door, as thesehouseholds preferred a daily cash flow ratherthan a lump sum at the month end. Since menin these households often undertook dailycasual labor, it was their wives who receivedand managed this income. These householdswere also less able to feed cows well due to lackof land for fodder, milk yields were thereforelikely to be lower. The husband therefore leftcontrol of this income to his wife to cater fordaily household expenses.In the coffee zone, vegetables were also cul-

tivated as a cash crop because there are watersources that can be used for irrigation, andbecause of its access to markets due to prox-imity to the main road to Nairobi. This pro-duction is for local markets rather than exportand mainly involves tomatoes, sukuma wiki (atype of kale), green peppers and aubergines. Inthe poorer households horticulture can be animportant source of income and men are morelikely to control it even though both men andwomen reported that the work they did on itwas about equal. It is easier for poorer house-holds to rent or borrow a piece of land to growvegetables than, for example, to afford anothercow as virtually all dairy cows are zero-grazed


and sources of fodder can be difficult to obtain.So the control of dairy income may be in thehands of the woman, whereas the man controlsthe horticultural income. In the better-offhouseholds women reported that they do moreof the work on horticulture while men thoughtit was fairly evenly split. But, this horticulturalincome is in the main controlled by women.This, they argued, was because it was smallcompared to the coffee income and can be usedfor daily expenses. This means that women donot necessarily have to ask for money fromtheir husbands for this.Income from food crops is generally small as

these are not cultivated as cash crops andwomen have always controlled the granariesand been allowed to sell small quantities as longas this did not affect household food supply.Where casual labor was the main source ofincome there were more likely to be discussionsat the end of the day about how it should bespent. Where it was a supplementary incomewomen were able to keep this income forthemselves.Data from the detailed individual survey for

the 24 married households, in which husbandsand wives were interviewed separately, indi-cated that wives in 21 of these had access to anincome source of their own (see Table 2). Ki-kuyu women have a long history in businessand trade (Leakey, 1956; Robertson, 1997) andnine had access to income via business. Ofthese, six involved women clearly controllingthe income for themselves: five of these weretrading: two in food and three in clothes, whilethe sixth was a tailor. In the other three busi-nesses husbands were heavily involved: all ofthese involved shops, two rural grocery shopsand an electrical goods shop in town. Onlythree women did not have direct access to anyincome source. Two of these were very youngwomen who had not been married long; theother was a very old woman whose husbandhad always been the provider. Young and

Table 2. Sources of wives’ main income in marriedhouseholds

Number of cases Total

Employment 5

Business 9

Casual labor 5

Income from crops or dairy 2

None 3

Total 24

newly married women in the tea zone particu-larly noted that ‘‘there is a problem when awoman has an income higher than the manbecause men see power in the income that theyhave and think women will become proud andfeel that they can control men with theirincome,’’ they went on to explain that starting abusiness for both of them can prevent theproblem so that ‘‘it seems that the businessmakes them equal.’’ Hence in the majority ofhouseholds (21 out of 24) women had access toan income source although in three of thesecases the income source was not independent ofthe husband’s influence. Women’s reportedincomes were however approximately 38% oftheir husband’s.This evidence suggests three main findings

regarding patterns in the gender division oflabor and control over income streams. First,men control the sources of income that gener-ate the largest income streams. Second, thismeans that control of particular income sour-ces, for example, tea, coffee and dairy, differsdepending on the relative importance of thisincome to the household. Third, that incomestreams from resources that generate small butregular amounts and on which women work,may often be assigned to women. These fea-tures of income sources mean that the incomethat men control tends to involve largeramounts than women, and some of these have alumpy and periodic, or sometimes irregularpattern, as in the case of tea and coffee income.By contrast women control smaller but moreregular income flows.

(b) The intrahousehold division of expenditureresponsibilities

The analysis of married couples views aboutthe balance of their expenditure responsibilitiesfound high levels of disagreement. As indicatedabove, respondents were asked to indicate theirrelative responsibility using five categories. Innone of the categories does the number ofhouseholds in which there is total agreementover the split in responsibilities reach a half ofthe 24 households.There was most agreement that asset pur-

chase was a male responsibility. There was stillconsiderable disagreement, however with menunderreporting women’s contribution accord-ing to women. The explanation for this mightbe that women often buy household assets,such as furniture, from ROSCA funds andthink that they are contributing to household

Table 3. Mean difference in monthly expenditure asreported by husbands and wives by expenditure category

Expenditure

category

ðn ¼ 21Þ

Husband’s

expenditure

estimate as %

of wife’s mean

Mean differ-

ence (Kshs)(husband’s less

wife’s estimate)

Overall

expenditure

170 5,365

Own food 154 857

Bought food 136 447

Non food

household items

182 435

Transport 391 1,872a

Clothes 414 5,651a

Health 1,949 13,663

Education 207 1,522

Other

expenditure

791 1,151

aDifference significant at 5% level.


assets––types of assets that men tend to over-look. Men on the other hand tend to think ofpurchase of larger assets such as land, buildinga house, bicycles and so on.The purchase of farm inputs and payment of

schools fees was also mainly a husband’sresponsibility. In the case of school fees therewas a reasonably high level of agreement withmen reporting it as their sole responsibility intwo-thirds of cases and the lowest level ofmajor disagreement across all categories. Thisis likely to reflect the fact that as an importantand major expense it is an area that has to beexplicitly budgeted and agreed on. Most menalso reported health expenditure as their soleresponsibility. But the cases of major dis-agreement showed a slightly different pattern toother categories: in this case men reported thatwomen were making higher relative contribu-tions than they themselves reported. Thedynamics here are often that the smaller day-to-day illnesses of children for which they requirea trip to the clinic or some tablets from the localshop, tend to be paid for out of women’spockets but men see themselves as responsiblein major cases of illness. It may be that wherecases of major illness are not present men dotherefore recognize this contribution.The purchase of food, nonfood items and

clothes for children were the areas where therewas least agreement and also categories wherefewer men reported that these were their soleresponsibility. While these are not cases wheremen necessarily concede that women have themajor responsibility it is clearer that responsi-bility is shared. The findings also indicate quitehigh levels of major disagreement. This mayreflect that where women have their ownincomes, they tend to spend on these items andthis is not necessarily discussed with the hus-band. Men may therefore tend to be unawareof some of the expenditure made in these areas.Rent, transport, and business investment

were categories of expenditure that applied toaround one-half of the households. Rent wasclearly a male responsibility in town-basedhouseholds. On the other hand transport tendsto be an expense that is paid for by whoever isincurring it. So that the greater mobility of menmeans that they have significantly higher levelsof expenditure on transport than were reportedby women (see Table 3). Business investmenttends to be another instance where it dependson whose business it is and it is primarily thatperson’s responsibility. Supplementary schoolexpenses such as transport and books were not

major expenses and were more likely to be anexpense to which women contributed. Finally,expenditure on clothes showed that clothesbought for the husband was the category inwhich there was overall the highest agreementthat this was the husband’s responsibility: 21men and 18 women. There was more disagree-ment over whose responsibility it was to buyclothes for the woman: the same number ofwomen reported that it was their sole respon-sibility as men reported that it was theirresponsibility.Given the level of disagreement over these

household expenditure responsibilities, it wouldseem likely that these might be reflected inhousehold expenditure reporting. An analysisof the previous month’s household expendituredata where husband and wife had indepen-dently estimated this (for the 21 cases in whichdata existed) indicated that the mean differencein total reported monthly expenditure was Kshs5,365 (see Table 3). This is approximately one-third of mean monthly expenditure reported bymen and over a half of that reported by women.Further, husbands reported income levels thatsuggested that 91% of total household expen-diture was funded from their own income,while women reported income levels whichamounted to just over 50% of their estimate ofhousehold expenditure.Husbands on average reported higher

expenditure than their wives in all expenditurecategories and these differences were particu-larly significant for transport and clothes. 7


These differences can be more easily under-stood in the light of the above analysis ofexpenditure responsibilities. It is clear that fortransport expenditure men have responsibilityfor their own expenditure and are more likelyto be mobile than women, hence it is a categorywhere women are less likely to be able to makea clear estimate of their husband’s expenditure.Similarly where men were purchasing clothesfor their family and themselves their wives wereless likely to have a clear view of his expendi-ture on clothes, and men were more likely to beresponsible for buying clothes for their wivesthan their wives were for buying clothes forthem. Hence the difference may be explained bythe husband’s purchase of his own clothes.These perspectives underline the point thatboth handle income and have separate expen-diture responsibilities with neither necessarilyhaving a clear view of their spouse’s contribu-tion.

Table 4. Classification of househo

Classification Definition

Independent

management

Both have separate sources of i

and neither has access to all hou

funds

Pooling system All household income is shared

both have access

Housekeeping

system

Man usually gives wife a fixed s

for housekeeping and retains th

Male whole

wage system

Husband manages all household

finances

(c) Intrahousehold financial management

Analysis of the qualitative data collectedduring the in-depth interviews for the 24 mar-ried households enabled us to characterize theirfinancial management system using Pahl’sscheme (see Table 4). The independent man-agement system best reflects the majority ofcases in our data set. The ‘‘housekeeping’’system can, however, run alongside the inde-pendent management system. Even in house-holds where the woman did have anindependent income source the husband maygive money to the wife for household expenses,as this income might be too small to make asignificant contribution.The evidence from Mathira suggests that this

in turn largely depends on the nature of co-operation and conflict within the household. Ittherefore depends on a co-operative relation-ship between husband and wife in how they mix

ld financial management systems

Households conforming to this

pattern ðn ¼ 24Þ

ncome

sehold

This system best characterises the majority

of households (15)

and Three households were found to clearly

operate a system of this nature. Two in

Chehe were dependent on casual labor

undertaken by both husband and wife as

their main source of income. They described

bringing the money they had earned to the

table and deciding what needed to be bought

with it

The third case was where the wife manages

an electrical shop and the husband operates

the related workshop––on a daily businesses

they pooled what they had earned and

planned how to use it

um

e rest

Five cases can be described as operating on

this basis, however, the amount that is pro-

vided is not fixed. In two cases the wife had

no independent income. Husbands may also

give wives money for household expenses

even where she also has an independent

source of income

One household where the wife had no inde-

pendent source of income appeared to fit this

type of system it did not appear that the wife

handled any cash, rather that the husband

was entirely responsible for purchasing

everything for the household


and match their income streams to meetexpenditure priorities. One point that is clearlydistinct between the European context and thatof Mathira, is that the presence of bank orsavings accounts was ubiquitous in Pahl’ssample and their joint use was common. InMathira, by contrast, the system is on thewhole a cash management system that is sup-ported by the use of accounts but rarely oper-ates in a way in which all money flows into andout of accounts. This suggests that the natureof claims and control may be very differentsince pooling income in bank accounts requiresconsiderable a priori agreement on use, whereasseparate cash income streams does not. Underindependent management in which women andmen control their own income, men and womenadopt financial services independently in orderto manage these income and expenditure flowswhich may be means of ensuring separationand entrenching independent control. Indeedthe qualitative evidence suggests that this wasan important function of financial service use.The very few couples who reported greater jointuse of bank accounts tended to be better edu-cated, and either employed or operating abusiness in town.In this sample, the evidence on co-operative

discussion was mixed but there was a cleartendency for the accounts of husbands andwives to be more consistent where it appearedthat there was greater discussion between themabout the levels of income and division ofresponsibilities. This was often also related tolevels of tension and conflict in the householdrelating to financial and other matters, such asmen’s behavior. Hence, one woman reportedthat, when she saw that her husband drank sherealized that they needed to plant tea so as to beable to pay for school fees when the childrenwere older, and they agreed that the tea moneywas mainly for this purpose. She said that shedid not know what he earned although he couldask to see her payslip [both were teachers], butif she asked to see his payslip then he would notshow it to her. She complained: ‘‘You know,African men are not there to tell you what theyhave in their pockets––that is their secret. . . hismoney is to do what he wants with’’ (HH90W).Social norms about household provisioning

are contested and men take the view that it istheir responsibility to provide, althoughwomen’s role in household provision has his-torically been strong among the Kikuyu (Clark,1980; Fisher, 1954; Kershaw, 1975). A marriedwoman stated it thus: ‘‘Men believe they are

going to do everything in the house but theycan’t manage’’ (HH106) and so she explainedthat she assisted him. Her husband expressed asimilar view saying it was his responsibility toprovide and that his wife’s role was to supple-ment. This belief appeared to be particularlyimportant for young men. There appeared to besome sensitivity among young married men inthe tea zone about the extent to which theirwives might have their own income and handlemoney. This seemed to be a result of the needfor young men after marriage to establishthemselves as household heads.

(d) Implications for the demand for financialservices

To summarize, this analysis of the genderdivision of labor and control of income sug-gests that where both women and men con-tribute labor and men have primary control ofthe resources involved in a productive activity,men tend to control the sources of income thatproduce the highest returns. These incomestreams in turn enable them to meet theirresponsibilities for areas of expenditure such asasset purchase, farm inputs, school fees andhealth care. These expenditures tend to be rel-atively large and lumpy and their incomestreams are often also uneven. Women aremore likely to be allocated streams of incomethat will enable them to purchase food andother basic household items that are requiredon a regular basis. Hence their income streamstend to be smaller and more frequent.Financial management in these households

can best be characterized as independent whereboth men and women have access to their ownincome sources and neither has access to allhousehold funds. But men have the ability toreallocate income streams where these derivefrom resources over which they have primarycontrol, but this is constrained by a basic normthat women have control over subsistence foodgrains. Men may also make claims on women’sincome but women may also seek places to savethat enable them to resist these claims, espe-cially by ensuring that the money is not physi-cally kept in the house.Due to the fact that men and women have

some independence of access and control overtheir income streams, the different character ofthese income streams, and the expenditureresponsibilities they must meet with them willresult in different demands for savings andloan services. First, since women’s incomes are


generally lower than men’s and they tend to usethis income primarily for household necessitiesand their personal expenditure, they havesmaller amounts to save and the sums theyrequire as lump sums to finance expenditurewill also be smaller than men’s. In order toconvert this small regular flow into larger sumsto purchase clothes or household items, womenused ROSCAs. For example, a widowexplained that when her husband was alive sheused her income from casual labor which was‘‘little by little’’ to join many groups and sheused the payouts to purchase household uten-sils such as cups. She had also used the groupsto buy other things for the house since he diedsuch as the sofa set and the glass for the win-dows (HH22FHHH).Even women who do not have a direct source

of their own income may use savings frommoney given them by another householdmember to buy food to join a ROSCA. Thiswas well illustrated by the case of a woman wholived with her husband and mother-in law andwho took Kshs 20 daily from the money hermother-in-law gives her to buy food for RO-SCA contributions although her husband didnot know about it. Women are also likely toneed funds quickly and conveniently to meetimmediate food or nonfood requirements (suchas fuel) or small health expenses in relation tochildren’s ailments. Hence the accessibility ofsavings is also a concern, but for those inROSCAs this is frequently overcome by bor-rowing from another member who is likely tolive very close by.Second, men have access to lumps of income

from cash crop payments or salaried employ-ment. In association with this they have greaterresponsibility for providing for lumpy pay-ments such as school fees and asset purchases.In order to provide for lumpy expenditure, theyrequire places to save their income or placesfrom which they can borrow to meet thesecredit needs. The regularity of ROSCA opera-tions does not match these requirements. Apartfrom uneven income flows, men’s expenditure isstrongly affected by seasonality for farm inputsand school fees. ROSCAs offer members thepayout in rotation and hence men’s demand isalso seasonally concentrated where the pur-chase of clothes, household utensils and smallhousehold assets by women can take place atany time of year.This analysis has demonstrated that when

financial services are seen as a means ofbridging the transition from income to expen-

diture, then demand by women and men isheavily differentiated due to the gender normsof control over resources and the allocation ofincome from them. The allocation of thisincome along with gender norms regarding theroles women can play in meeting expenditureresponsibilities determines the nature ofdemand. In this case ROSCAs have character-istics that make them a more effective means ofconverting women’s income flows into the lumpsums they require than for men.

7. THE SUPPLY SIDE: THE ROLE OFGENDERED NORMS AND SANCTIONS

IN ACCESS TO ROSCAs

The second component of the analyticalframework proposed above argued that accessto financial services depends on the ability ofactors to engage with the rules and norms thatenable different types of intermediary to oper-ate. This section explains women’s greater useof ROSCAs as arising from the greater effec-tiveness of the informal sanction of shame forwomen compared to men and traces the his-torical social origins of this.

(a) Differing experiences of men and women inROSCAs

The most consistent aspect of men andwomen’s own explanations of differential RO-SCA use concerned the social consequences ofnonpayment and how these differed betweenthem. Informants said that women felt shame ifthey were unable to make a contribution totheir ROSCA––a woman would rather borrowfrom a friend to take money to the group thannot pay as she did not want to be embarrassedand ‘‘spoil’’ the group. Nor did she want theshame of people coming to take things from herhouse. As one woman put it:

It is embarrassing to go to a gitati [group] withoutmoney. . . you are going to be told to give them theirmoney––you have helped yourself and used the moneyand today you are giving another person, so you haveto pay it back. You are going to spoil your name andpeople will be fearing you––seeing you as someonewho doesn’t pay, I would prefer to ask someone togive me the money than to go to the gitati [group]without the money. . . if you can’t pay they may cometo take your security, your TV. . . It is not good. . . itwill be auctioned. Would you be happy? The childrenwill be asking questions––how do I answer them?(EJN, 142).


It appeared that men also sought to avoidputting their wives in a situation where theymight experience shame through not being ableto pay. Both wives and husbands explained thathusbands had refused when they asked to join aROSCA. These were cases where the wife hadasked to join but where the husband wouldhave to give her the contributions as she had noindependent income of her own. If the husbandfelt that he could not reliably provide the con-tributions he felt it was better for the wife notto join, as he did not want her to be embar-rassed in the group by his failure to bringmoney for her. Where women had their ownincomes or were given housekeeping money,then it was usually up to them how they man-aged it. If they wished to make contributions togroups from this, they were usually free to doso. 8

By contrast, male informants reported manycases of men’s groups that had failed. In anumber of cases fights had occurred over groupbreakdown and some young male informantscould show the resulting scars. Some of thesecases involved con men claiming to start groupsand canvassing the young men in town to makecontributions. These groups never met and bythe time it came to their turn they would notget the money. They knew that even if theytook the case to the police they were unlikely toget their money back and taking the case tocourt was too expensive.Men themselves explained this as the result

of a more individualistic culture among themand described themselves as having a ‘‘don’tcare’’ attitude. Men reported that they did notlike the strictness of the rules involved ingroups and the fines they would incur if theydid not attend. Other men also explained thatmen could not mix with each other since theydo not want to be seen as needing to dependon others and so being seen as inferior. Theydid not like to consult each other when theyhad problems but preferred to keep theirproblems to themselves: ‘‘men are proud anddo not trust each other. . . every man isclever. . . you know someone is going to messyou up’’ (i.e., if you get into a group withthem) (FGD6).Often the rules of the group were not written

down and men realized that nothing couldreally be done if they did not pay, so they didnot fear default. If a woman was not able tocomplete her payments in the group then thegroup could go and repossess the items but aman would resist this, as a woman reported:

‘‘You go to his place to start selling things andhe will just say no’’ (REM, ARG, 121).These women further reflected that men did

not feel shame in the way that women did, witha local saying: ‘‘thoni cia arume ciri igoti’’ whichin translation means ‘‘the shame of men is atthe back of the head.’’ She translated this to say‘‘a man does not care about anything if hebelieves he is unable to pay at that time’’(REM, 121). The word for shame in Kikuyutranslates as something between shame andshyness and the saying refers to the fact thatwhen a woman feels this it is evident in herexpression, and she is unable to look someonein the eye, but a man will not feel it and willhold his head up high, in this case, when tellingothers that he has no money to bring to thegroup.This is not to say that men did not join

groups––whether men’s or women’s––or thatsome did not value them. In fact a number ofyoung men demonstrated their eagerness tojoin a ‘‘good’’ group––which usually meant areliable women’s group––and lamented theirinability to gain access to such groups. Someyoung men reported being ‘‘carried’’ by womenin groups. This was where a young man, whohad a good relationship with a woman whotrusted him, joined a group through her. Hewould then give her the contributions but onlyher name appeared on the register. This is oneof the few ways a young man could join reliablegroups when he was not yet married and couldnot therefore do so through his wife.Many men reported however that they had

now joined self-help groups due to the hardtimes they were facing as the economic down-turn since 1997 had ‘‘brought the men to beingable to be humble in their homes’’(FGD5).Men said that they preferred to start their owngroups because they saw women’s groups asdealing with ‘‘small things,’’ i.e., householditems such as utensils, where as they needed tosave for purposes such as school fees andinvestments. One men’s group said that theyhad started as men only (in the rural area) butwives came on behalf of their husbands––usu-ally because the husband was not reliable––andso it became hard to exclude them (FGD4).As was seen in the last section, men have

expenditure responsibilities that involve bulkypayments for school fees, asset accumulationand so on. Thus if groups are to be a means offinancing these responsibilities, men need to bein groups where the payments are higher inorder to achieve this within an acceptable


period of time. This evidence suggests that forwomen, embarrassment or shame acts as asufficient informal sanction to enable ROSCAsto operate. For men however this informalsanction is not effective and more formal meansbecome too costly. The next section exploresthe historical social reasons why women weremore able to operate in these groups than men.

(b) The role of groups in Kikuyu socialization

This research clearly indicates the ongoingsocial importance of women’s groups. As oneyoung man put it: ‘‘To a woman a gitati is herlifestyle, it’s her way of life. . . a woman is not awoman until she is in a gitati’’ (YIK, S34). Thissocial context enables them to learn how tospeak properly in public and the older womeneducate the younger women on how to handlethings in their homes and discuss problems withtheir husbands. Their ability to buy a sweateror a smart dress with the ROSCA payout andbe dressed smartly and cleanly enables them togain in self-esteem and mix freely with richerpeople––‘‘You can become a woman’’ (FGD3).This was judged important by town womenwho felt that their husbands would run away ifthey were not smart because he would compareher to a woman dressed smartly that he hadmet in a bar. But although he wants her to dresswell he will not provide the money to do this.‘‘Men just need to see the woman in thehouse. . . so it’s upon her to get the way to dressand manage the house’’ (FGD3).One of the main roles that groups perform

for women is that of providing themselves withhousehold utensils. They noted that that if youenter the house of a woman who is in a groupthen it will be noticeable because she will havecups, plates and sufurias (cooking pots).Women explained that while a man will beinterested in building a house, he would not beconcerned with buying what goes inside it.While some men expect those things to be thereeven though they will never buy them, otherswill not care: ‘‘if there is no cup to drink fromhe will just tell you to bring the tea in the su-furia!’’ (FGD3). Wainaina underlines the socialimportance of having such utensils by sayingthat women without them are ‘‘shy to come outand meet with other women, or to allow visitorsin to their homes’’ (quoted in Ardener, 1995, p.13).Anthropological sources offer evidence of the

social and economic role that groups haveplayed in Kikuyu society in the past. There is

extensive evidence of the role of work groups ofeither sex involved in agricultural and othertasks (Fisher, 1954; Stamp, 1975). Stamp alsonotes the existence of women’s lodges, whichconsisted of women married into a particularlineage and had economic, social and judicialfunctions with evidence that there was animportant age-grade system operating amongwomen based on circumcision and maritalstatus with the highest status being accorded towomen whose first child had been circumcised(nyakinyua) and these were regarded as leaderswho were capable of disciplining others. Theseage-sets gave women power ‘‘apart from’’ menrather than over men, and Robertson arguesthat their functions both legitimized and per-petuated patriarchal values while also defininga women’s sphere and forbidding male entry(Robertson, 1997, p. 243). But in reviewingchanging patterns of women’s collective actionover 70 years, Robertson argues that thesegroups have moved away from the role of‘‘producing properly socialized adults. . . to amore class-based women’s solidarity involvedwith promoting women’s economic activities’’(p. 240). It was after independence that ROS-CAs clearly became one of the activities ofthese groups and in the 1970s and 1980s thatwomen’s groups expanded hugely. Thisexpansion was in part spurred by the broaderideology and practice of Harambee (Kiswahilifor pulling together) promoted by the govern-ment and often supported by donors (Mwaniki,1986; Udvardy, 1998). These groups undertooka wide range of activities: joint farming, or jointselling of their labor in agriculture; water pro-jects; accumulation of funds to buy land forcultivation or building; setting up their own co-operatives and so on. Savings funds and RO-SCA activities were often a core activity ornecessary feature to enable the projects to takeplace (Maas, 1986; Nelson, 1995; Robertson,1997).By contrast, male age-sets traditionally pro-

vided a means for collaboration across lineages(Muriuki, 1974). But even when they did,Lonsdale suggests that male solidarity was notstrong, especially after men married, quotingthe Kikuyu saying: ‘‘helpful age-mates becomerival householders’’ (Lonsdale, 1992). More-over, since age-sets no longer exist, the sociali-zation process for men has not involved groupmembership and it has certainly not beenadapted to current circumstances in the waythat Robertson describes for women. Hence itcan often be the case that a young man only


learns about groups from his wife once he hasgot married (AIG, 17).The importance of groups to women––often

now with ROSCAs as a core activity––has alsothen to be viewed in historical context. Theyhave been and remain a means through whichwomen learn about and fulfill both their socialand economic roles. Hence, in contrast to menit can be argued that women have retained theskills to pursue collective action of this kind.

(c) Summary

When financial intermediation is seen asrequiring rules or norms in order to enforcerepayment, whether formal (i.e., legislatedrules) such as contract law or informal normsand sanctions operating through reputationand shame, then the ability of different socio-economic groups to interact with different typesof financial organization will in part depend ontheir ability to interact with the rules and normsinvolved.Respondents contrasted the individualistic

and opportunistic norms of men’s behaviorwith the shame that women would experience ifthey failed to repay. It is important to empha-size, however, that the argument here is notthat men do not experience shame. Rather thatin the context of ROSCA activity this socialsanction does not ensure men’s compliance tothe rules of the group as effectively as it does forwomen.Anthropological accounts of the ways in

which women and men were traditionallysocialized support the view that women’sgroups have evolved and adapted to take onnew roles in contemporary society while age-sets have declined as a basis for male solidarityoffering no enduring means for crosslineagecollaborative activity. The greater dependenceof women on female kin and neighbors forlivelihood support––for example, the ways inwhich women reciprocate and support eachother with food and other transfers––may alsohelp explain this (see Francis, 2000).

8. CONCLUSIONS

This paper has laid out an analyticalapproach to examining the role of gender infinancial markets by examining its influence onthe demand for financial services and on access.This approach was applied to the finding thatthe most widely used service in Mathira is

ROSCAs but that their use was gendered. Theevidence demonstrated that the nature of theincome streams which men control could belarge and periodic, even irregular, compared towomen’s, which were smaller but more fre-quent. This meant that women rather than menwere more likely to have access to the regularcontributions that ROSCAs require. Moreover,men’s expenditure responsibilities were alsolumpy, and in many cases are tied to particulartimes of year. The ability of ROSCAs to caterto these needs is limited for two reasons. First,ROSCAs offer members the payout in rotationand hence men’s demand is likely to be sea-sonally concentrated where as the purchase ofclothes, household utensils or small householdassets by women can take place at any time ofyear. Second, the greater scale of men’sexpenditure requirements is another factor. Inorder to gain adequate lump sums from ROS-CAs, men would need to contribute higheramounts. Given the apparent weakness ofinformal social sanctions in men’s ROSCAsand the larger amounts they would need to savein them to make them useful, ROSCAs are amuch more risky place for them to save com-pared to formal financial institutions. Alterna-tively, more cost effective enforcementmechanisms would need to be found for men’sROSCAs.This analysis also helps explain the finding

reported in section two that being young edu-cated and male raised the likelihood of bor-rowing from friends and relatives. Since men donot have social relationships that enable themto join groups to meet their financial needs theyare more likely to turn to friends and theyexplained that this was how they were mostlikely to borrow. The young educated men thatthis finding referred to were unmarried, livingin Karatina and running their own businesses.This meant that they needed funds, had nocollateral––because they were too young tohave inherited land––and had not yet had theopportunity to learn about groups or thebehavior required to manage to borrow fromthem.This approach can also accommodate the

role of gender norms in access to property,since collateral requirements are also rules thatsome financial intermediaries require in orderto operate. A second gendered pattern of use incentral Kenya was that women were far lesslikely to use cash crop SACCOs than men. Thisis also then easily incorporated by this analysissince a rule of access to cash crop SACCOs is


being a tea or coffee farmer and it is usuallymen that hold the license and hence join theseSACCOs. The approach of looking at the rulesand norms that govern different types offinancial intermediary and the ability of menand women operating within gender norms toengage with them, presents a more completeanalysis than the analysis of property rightsalone.

(a) Implications for analysis and policy

This approach therefore suggests that finan-cial markets are certainly segmented by genderbut may also be fragmented by it. Nissanke andAryeetey distinguish between a segmentedmarket and a fragmented market on the basisthat fragmentation is evident when returnsadjusted for risks and transactions costs are notcomparable between market segments (Nis-sanke & Aryeetey, 1998). The term segmenta-tion on the other hand describes a process ofnormal specialization between products (e.g.,with respect to prices) that are evened out oncerisk-adjusted returns are calculated. Manyresearchers have recognized the segmentationof informal financial markets and suggestedthat this is due to variables such as locality,type of lender, the trade, service or sector beingfinanced (Ghate, 1988). Remaining variationsin returns are therefore explained in terms ofimperfections in the market due, for example,to barriers to entry. Policy then concentrates onthe removal of barriers to entry in order thatcompetition can erode excess profits.This research demonstrates that gender

norms present barriers to entry, first, on thesupply side, not only as has been convention-ally understood in terms for example of prop-erty rights, but also through the interaction ofmen and women with informal norms whichsupport the operation of financial intermedia-tion. Second, on the demand side, gendernorms present obstacles by gendering theirdemand for services. Approaches to removingthese gendered barriers to entry are thereforerather complex since they require changes to arange of gender norms and practices that aredeeply embedded in social behavior. ‘‘Buildinginstitutions for markets’’ therefore requires awillingness to address these, for example,through tackling norms regarding intrahouse-hold income distribution and financial man-agement.The evidence presented regarding the gen-

dered role of shame as a social sanction

enforcing repayment in group-based financialarrangements might also lead us to argue thatmoral hazard is itself gendered with men beingmore willing and likely to shift the costs ofborrowing onto others than women (see alsovan Staveren, 2001). This has two potentialpolicy implications, first, since providers may infact face lower costs of provision to womenthan men because of this, we would expect tosee gender differentiated prices as a result andpolicy could actively support such a develop-ment. 9 Alternatively, policy might attempt toinfluence male behavior toward repayment, forexample, by making sanctions more effectiveand severe, or by moral suasion and the pro-motion of positive role models.

(b) Implications for microfinance policy andpractice

These findings give further scope for reflec-tion on the microcredit and women’s empow-erment debate summarized at the outset. First,it suggests that microfinance programs targetedtowards women and intent on their empower-ment may have inherently contradictory policyobjectives since this at the same time makesinstrumental use of their gendered subordina-tion. Rutherford (1995) has criticized the vir-tually exclusive targeting of women inBangladesh noting their lack of mobility, andhence availability in villages of Bangladeshwhen meetings are held and comments on theirgreater deference to authority which makesthem easier for MFI workers to manage. WhileMFI programs in Kenya are not exclusivelytargeted to women, credit officers alsoexplained that groups whose members are atleast half female were easier to manage. Thefact that women are likely to take with them thesocial historical context of their group activitiesand compliance to informal sanctions fromROSCAs into microcredit groups then pro-duces an inherently contradictory position forthem.Second, the analysis above presents a

framework through which intrahouseholdfinancial arrangements can be investigated andunderstood. The microfinance industry needsto adopt such an approach if it is to be suc-cessful in developing effective strategies forgrowth. The controversy that was precipitatedby Goetz and Sen Gupta’s analysis of loancontrol and use concerned different construc-tions of how these arrangements took place,since it appeared that supposed independent


control by women of loans did not concur withthe reality. Those familiar with microfinance inBangladesh were aware of this disparity butnever challenged the ‘‘public transcript’’ whichhad been used to promote the industry else-where. Key differences between the Bangladeshcontext and this part of Kenya are that womenhave significantly higher levels of autonomy inmanaging their own income streams; greateraccess to the public realm and hence ability toconduct their own businesses; and retaingreater control of the resulting income. Giventhese differences it should be expected thattargeting loans to women is likely to have sig-nificantly different impacts on intrahouseholdrelations in different contexts. In particular,since intrahousehold relations can fundamen-tally affect the room for maneuver that womenhave, they may not be able to adequatelyexpand their businesses to absorb higher loansizes (see also Johnson, 2003a).Third, from the practical perspective of the

microfinance industry, the above analysis sug-gests the need for microfinance institutions(MFIs) to start to tackle demand in a gender-differentiated manner. The increased emphasisin the industry on making services more client-led (Cohen, 2002) and developing the marketresearch and product development techniquesto achieve this (Wright, 2003) must also rec-ognize the need for market research methodsthat are gender aware and the potential forwomen targeted products. The related chal-

lenge is however to ensure that productsdesigned for women do not entrench them insubordinate positions by responding only towomen’s practical needs identified throughmarket research. For example, improving theavailability of cheap, safe and cost efficientservices for the small amounts of savings whichwomen wish to make (Vonderlack & Schreiner,2002) may address their practical needs, but itis necessary to ensure that their strategic needs(Moser, 1993)––which may be less apparentfrom market research––are also tackled. Thedesign of loan products to date has concen-trated on financing women’s productive ratherthan reproductive roles, but products couldalso be designed to facilitate improvements inhousing, healthcare, cooking and plumbingsince these are labor and resource saving forwomen (Johnson & Kidder, 1999). By reducingthe burden of reproductive work, this is alsolikely to alleviate the constraints women face inmaking successful use of enterprise loans.Finally, a gender analysis (rather than one

solely focused on women) must also point outthat these findings also show that some menalso lack access to collateral. Young, educatedmen find themselves in a position where theylack the social relationships that would enablethem to borrow through these mechanisms.This in turn suggests that policy in microfi-nance needs to re-think how access for men,and young men in particular, can beimproved.

NOTES

1. I have not found any analysis of this nature for

developing country contexts.

2. Burgoyne analyzes financial management by Euro-

pean couples who have remarried in which she uses

categorizations of shared and pooled management. For

her, the difference between shared management and

pooling is an attitudinal one about the claims that each

can make in relation to the income of the other. That is,

even if income is not pooled so that access is equal, it is

putatively pooled in the sense that all is available for

household use even if the management arrangements

remain separate.

3. The microfinance sector refers to ‘‘microfinance

institutions’’ although they may more correctly be

referred to as organizations. Unfortunately, this can

confuse the discussion of institutions in markets but I

use the term MFI because this is now an institution

within that sector!

4. Given the small sample size, significance testing in

crosstabulations and the use of logistical regressions

was only possible where subsamples were sufficiently

large.

5. Alternatively the payout may be auctioned with the

person willing to take the largest discount receiving the

payout. The remainder of the funds are then divided

among those who have not yet won the payout.

6. ROSCAs are regarded here as a savings service

although all but the last member to receive the payout

effectively obtains a loan. If these are included as a loan

service also, they are clearly also the most used loan

service.


7. The figures in Table 3 for health show that men’s

expenditure on this was on average almost 2000% of

women’s which underlines the point made above about

the nature of their expenditure responsibilities, but the

actual differences were not significant due to the small

sample and large variation.

8. An exception to this arose in Chehe (tea zone) where

less well-off young men did not like their wives belonging

to groups. This seemed to be due to a sense of insecurity

in their marriage. Some wives explained that their

husbands did not like them to bring things for the

house because they felt the women would boast about it

and undermine their status as head of the household.

9. Lower prices for women drivers have been seen in

the UK motor insurance sector to reflect women as more

careful drivers than men.

REFERENCES

Adams, D. W.& Fitchett, D. A. (Eds.). (1992). Informalfinance in low-income countries. Boulder, CO: West-view Press.

Amin, R., Becker, S., & Bayes, A. (1998). NGO-promoted microcredit programs and women’sempowerment in rural Bangladesh: quantitativeand qualitative evidence. Journal of DevelopingAreas, 32(2), 221–236.

Anderson, S., & Baland, J.-M. (2002). The economics ofROSCAs and intrahousehold resource allocation.Quarterly Journal of Economics, CXVII(3), 963–995.

Ardener, S. (1995). Women making money go round:Roscas revisited. In S. Ardener & S. Burman (Eds.),Money-go-rounds: the importance of rotating savingsand credit associations for women (pp. 1–20). Oxford/Washington, DC: Berg.

Ardener, S.& Burman, S. (Eds.). (1995). Money-go-rounds: the importance of rotating savings and creditassociations for women. Oxford/Washington, DC:Berg.

Bouman, F. J. A. (1995). Rotating and accumulatingsavings and credit associations: a development per-spective. World Development, 23(3), 371–384.

Burgoyne, C. B. (1990). Money in marriage: howpatterns of allocation both reflect and conceal power.Sociological Review, 38, 634–665.

Case, A. (1995). Symposium on consumption smoothingin developing countries. Journal of Economic Per-spectives, 9(3), 81–82.

Clark, C. M. (1980). Land and food, women and powerin nineteenth century Kikuyu. Africa, 50(4), 357–370.

Cohen, M. (2002). Making microfinance more client-led.Journal of International Development, 14(3), 335–350.

DiMaggio, P. (1994). Culture and economy. In N. J.Smelser & R. Swedberg (Eds.), The handbook ofeconomic sociology (pp. 27–57). New York: RussellSage Foundation.

Elster, J. (1989). Nuts and bolts for the social sciences.Cambridge, UK: Cambridge University Press.

Fisher, J. M. (1954). The anatomy of Kikuyu domesticityand husbandry. Nairobi: Overseas DevelopmentAdministration.

Folbre, N. (1986a). Cleaning house: new perspectives onhouseholds and economic development. Journal ofDevelopment Economics, 22(1), 5–40.

Folbre, N. (1986b). Hearts and spades: paradigms ofhousehold economics. World Development, 14(2),245–255.

Francis, E. (2000). Making a living: changing livelihoodsin rural Africa. London and New York: Routledge.

Geertz, C. (1962). Rotating savings and credit associa-tions: the middle rung in economic development.Economic Development and Cultural Change, 10(3),241–263.

Ghate, P. B. (1988). Informal credit markets in asiandeveloping countries. Asian Development Review,6(1), 64–85.

Goetz, A., & Gupta, R. S. (1996). Who takes the credit?Gender, power and control over loan use in ruralcredit programs in Bangladesh. World Development,24(1), 45–63.

Guyer, J. I. (1980). Food, cocoa and the division oflabour by sex in two West African societies. Com-parative Studies in Society and History, 22(3), 355–373.

Hart, G. (1995). Gender and household dynamics: recenttheories and their implications. In M. Quibria (Ed.),Critical issues in Asian development (pp. 39–74).Oxford: Oxford University Press.

Hashemi, S., Schuler, S., & Riley, A. (1996). Rural creditprograms and women’s empowerment in Bangla-desh. World Development, 24(4), 635–654.

Hodgson, G. (1999). Evolution and institutions. Chelten-ham, UK: Edward Elgar.

Hoff, K., & Stiglitz, J. E. (1990). Imperfect informationand rural credit markets––puzzles and policy per-spectives. World Bank Economic Review, 4(3), 235–251.

Johnson, S. (2003a). Gender relations, empowerment andmicrocredit: moving on from a lost decade. Mimeo.

Johnson, S. (2003b). ‘‘Moving mountains’’: an institu-tional analysis of financial markets using evidencefrom Kenya. Unpublished Ph.D., University ofBath.

Johnson, S. (2004). ‘‘Milking the elephant’’: financialmarkets as real markets in Kenya. Development andChange, 35(2).

Johnson, S., & Kidder, T. (1999). Globalization andgender––dilemmas for microfinance organizations.Small Enterprise Development, 10(3), 4–15.

Kabeer, N. (1994). Reversed realities: gender hierarchiesin development thought. London: Verso.

Kabeer, N. (2001). Conflicts over credit: re-evaluatingthe empowerment potential of loans to women inRural Bangladesh. World Development, 29(1), 63–84.


Kershaw, G. (1975). Changing roles of men and womenin the Kikuyu family by socioeconomic strata. RuralAfricana, 29, 173–193.

Khandker, S. R. (1998). Fighting poverty with micro-credit: experience in Bangladesh. Dhaka: UniversityPress.

Kimuyu, P. (1999). Rotating savings and credit associ-ations in rural East Africa. World Development,27(7), 1299–1308.

Leakey, L. S. B. (1956). The economics of Kikuyu triballife. East African Economic Review, 3(1), 165–180.

Longhurst, R. (1982). Resource allocation and thesexual division of labour: a case study of a MoslemHausa village in Northern Nigeria. In L. Beneria(Ed.), Women and Development: The Sexual Divisionof Labour in Rural Societies (pp. 95–118). New York:Praeger.

Lonsdale, J. (1992). The moral economy of Mau Mau:wealth, poverty and civic virtue in Kikuyu politicalthought. In B. Berman & J. Lonsdale (Eds.),Unhappy valley––conflicts in Kenya and Africa. Lon-don: James Currey.

Maas, M. (1986). Women’s groups in Kiambu, Kenya: itis always a good thing to have land. Leiden: AfricanStudies Centre.

Manser, M., & Brown, M. (1980). Marriage andhousehold decision-making: a bargaining analysis.International Economic Review, 21(1), 31–44.

Mayoux, L. (1999). Questioning virtuous spirals: micro-finance and women’s empowerment in Africa. Jour-nal of International Development, 11(7), 957–984.

Mayoux, L. (2001). Tackling the down side: socialcapital, women’s empowerment and micro-finance inCameroon. Development and Change, 32, 435–464.

McElroy, M. B., & Horney, M. J. (1981). Bash-bargained household decisions: toward a generaliza-tion of the theory of demand. International EconomicReview, 22(2), 333–347.

Moser, C. (1993). Gender planning and development:theory, practice and training. London: Routledge.

Muriuki, G. (1974). A history of the Kikuyu 1500–1900.Nairobi: Oxford University Press.

Mwaniki, N. (1986). Against many odds: the dilemmasof women’s self-help groups in Mbeere, Kenya.Africa, 56(2), 210–227.

Nelson, N. (1995). The Kiambu group: a successfulwomen’s ROSCA in Mathare Valley, Nairobi (1971–1990). In S. Ardener & S. Burman (Eds.),Money-Go-Rounds: The importance of Rotating Savings andCredit Associations for Women (pp. 49–70). Oxford/Washington, DC: Berg.

Nissanke, M., & Aryeetey, E. (1998). Financial integra-tion and development: liberalization and reform in sub-Saharan Africa. London and New York: Routledge.

Pahl, J. (1989). Money and marriage. Basingstoke:Macmillan.

Platteau, J.-P. (1994). Behind the market stage wherereal societies exist, Parts I & II. Journal of Develop-ment Studies, 30(3), 533–577.

Posner, R. A., & Rasmusen, E. B. (1999). Creating andenforcing norms, with special reference to sanctions.International Review of Law and Economics, 19, 369–382.

Rahman, A. (1999). Micro-credit initiatives for equitableand sustainable development: who pays? WorldDevelopment, 27(1), 67–82.

Robertson, C. (1997). Trouble showed the way: women,men and trade in the Nairobi area 1890–1990.Bloomington and Indianapolis: Indiana UniversityPress.

Rutherford, S. (1995). ASA: the biography of an NGO,empowerment and credit in rural Bangladesh. Dhaka:Association for Social Advancement.

Rutherford, S. (1999). The poor and their money. Oxford:Oxford University Press.

Schuler, S. R., Hashemi, S. M., & Badal, S. H. (1998).Men’s violence against women in rural Bangladesh:undermined or exacerbated by microcredit pro-grammes. Development in Practice, 8(2), 148–157.

Sen, A. K. (1990). Gender and co-operative conflicts. InI. Tinker (Ed.), Persistent inequalities: women andworld development (pp. 123–149). Oxford: OxfordUniversity Press.

Stamp, P. (1975). Perceptions of change and economicstrategy among Kikuyu women of Mitero. RuralAfricana, 29, 19–44.

Stichter, S. (1987). Women and the family: the impact ofcapitalist development in Kenya. In M. G. Schatz-berg (Ed.), The political economy of Kenya (pp. 137–160). New York: Praeger.

Stiglitz, J. E. (1993). Economics. New York: W.W.Norton.

Stiglitz, J. E., & Weiss, J. (1981). Credit rationing inmarkets with imperfect information. American Eco-nomic Review, 71(3), 393–410.

Udvardy, M. L. (1998). Theorizing past and presentwomen’s organizations in Kenya. World Develop-ment, 26(9), 1747–1761.

van Staveren, I. (2001). Gender biases in finance. Genderand Development, 9(1), 9–17.

Vonderlack, R. M., & Schreiner, M. (2002). Women,microfinance and savings: lessons and proposals.Development in Practice, 12(5), 602–612.

World Bank (2001). World Development Report: Engen-dering Development: Through Gender Equality inRights, Resources and Voice. New York: WorldBank and Oxford University Press.

World Bank (2002). World Development Report: Build-ing Institutions for Markets. New York: World Bankand Oxford University Press.

Wright, G. (2003). Client responsive product develop-ment. Small Enterprise Development, 14(3), 12–21.

Documents

Gender Norms in Financial Markets: Evidence from Kenya