microenterprise occupation and poverty reduction in microfinance programs: evidence from sri lanka

World Development Vol. 32, No. 7, pp. 1247–1264, 2004� 2004 Elsevier Ltd. All rights reserved

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

lddev.2004.01.009
www.elsevier.com/locate/worlddev
doi:10.1016/j.wor

Microenterprise Occupation and Poverty Reduction

in Microfinance Programs: Evidence from Sri Lanka

JUDITH SHAW *

RMIT University, Victoria, Australia

Summary. — The microenterprise earnings of microfinance clients in southeastern Sri Lanka arelinked to their initial incomes. Poorer clients face geographic, financial and sociocultural barriers toentry to the most promising microenterprise occupations, leading them to select low-value activitieswith poor growth prospects. In semi-urban areas, poverty impacts could be strengthened bysupplementing loans with nonfinancial interventions encouraging poor clients to select higher-valueoccupations. In arid rural areas, where microenterprises face severe market and infrastructureconstraints, microenterprise development is unlikely to facilitate poverty exit.� 2004 Elsevier Ltd. All rights reserved.

Key words — Asia, Sri Lanka, microfinance, poverty, microenterprise development, rural

livelihoods

* Final revision accepted: 19 January 2004.

1. INTRODUCTION

Most observers agree that in the right cir-cumstances microfinance can increase house-hold incomes, but its impact on poor clientsremains controversial. Several recent impactevaluations have emphasized the nonuniformdistribution of benefits. Most households arebetter off with microfinance, but incomeimpacts vary in magnitude and durability, anda sizeable proportion of clients find that theirpost-credit incomes stagnate or fall (Copestake,Bhalotra, & Johnson, 2001; MkNelly & Dun-ford, 1998, 1999; Mosley, 2001; Sebstad &Chen, 1996; Todd, 2000). Initial income hasbeen identified as a key determinant of impact.An influential cross-country study found thatloans produce the greatest percentage increasesin the incomes of ‘‘upper-poor’’ and nonpoorborrowers who are close to or above nationalpoverty lines. The ‘‘extreme’’––or ‘‘core-poor’’––the poorest 50% or thereabouts ofthose in poverty––were not only less likely toparticipate in microfinance programs; whenthey did participate their post-credit incomeswere less likely to increase. Moreover, suchincreases as occurred were often too small andshort-lived to enable sustainable poverty exit(Hulme & Mosley, 1996). Subsequent stud-ies provide further evidence of a relation-ship between initial income and microcredit

124

impact (Hashemi, 1997; Rahman, 1997;Zaman, 1999).Income promotion through microenterprise

lending remains the primary strategy of mostmicrofinance institutions (MFIs), includingmany which serve poor clients. For poverty-focused MFIs, the questionable efficacy ofmicroenterprise lending at the low end of theincome spectrum makes impact monitoringparticularly important. If their interventionsare not effective, or are producing differentialoutcomes in different client groups, they needto know why. An understanding of the reasonsfor low poverty impacts yields importantinformation for program design and targeting.It may be feasible to improve impacts byredesigning services to better meet the needs ofpoor clients, within the limits imposed by theneed to maintain institutional viability. On theother hand, the poor commonly face nonpro-gram obstacles to microenterprise develop-ment, in the form of unfavorable marketenvironments or inadequate physical infra-structure, over which MFIs have little influ-ence. In such circumstances, a more sociallybeneficial allocation of resources may resultfrom targeting less-poor borrowers who derivegreater benefits from microenterprise loans or,

7

WORLD DEVELOPMENT1248

for MFIs which choose to maintain a povertyfocus, reorienting their activities away fromincome promotion in favor of protectional ornonfinancial interventions.This paper examines the underlying causes of

the income-related impact gap, a topic whichhas received surprisingly little scholarly atten-tion, given its implications for the effectivenessof microenterprise lending in reducing poverty.It investigates the reasons for disparities inmicroenterprise earnings among clients of twoMFIs in southeastern Sri Lanka. As expected,the microenterprises of less poor clients dobetter than those of the poor. Another impor-tant finding is that poverty impacts are differ-entiated by location: poor clients in semi-urbanareas have considerably greater opportunitiesthan their rural counterparts to exit povertythrough microenterprise development services.The semi-urban/rural differential raises impor-tant policy issues for the MFIs, which provide auseful and highly valued service to their near-poor and nonpoor clients, but face challengesin meeting the needs of the poor. There isconsiderable potential for improving impactson the semi-urban poor by encouraging thetake-up of high-earning microenterprise occu-pations. In arid rural areas, however, wheremarket and infrastructure constraints virtuallyrule out poverty-clearing microenterprises, theutility of microenterprise development for poorclients is more problematic. Support for ruralmicroenterprises helps to alleviate some ill-effects of poverty––although it has little impacton poverty incidence––but its benefits are offsetby significant risks and costs for borrowers andlenders, and the generation of negative exter-nalities.The paper is structured as follows: Sections 2

and 3 describe the research context, methodo-logy and findings. Section 4 discusses the rea-sons why poor clients earn less from theirmicroenterprises, identifying geographic,financial and sociocultural barriers to entrywhich deter them from selecting higher-valueoccupations. Section 5 evaluates the MFIs’services to poor clients and makes the case for astronger promotional focus on the semi-urbanpoor.

2. THE RESEARCH CONTEXT ANDMETHODOLOGY

The research was conducted in Hambantotadistrict, a remote region on Sri Lanka’s south-

eastern coast, 250 km from the metropolitanWestern Province and the national capital,Colombo. The district is one of the country’spoorest regions, ranking well below the nationalaverage on household income, employment,literacy, and access to electricity, safe water andsanitation (United Nations Development Pro-gram, 1998). An estimated quarter of its popu-lation of 558,000 is semi-urban, living in andaround five or six regional towns and along thearterial coastal road which connects the regionwith the Western Province. The remaining threequarters live in the mostly arid, sparsely popu-lated rural hinterland. Microenterprises––defined as firms employing between one and 10workers––are by far the largest category ofemployment. Two-thirds of the district’s laborforce are microenterprise owner-operators orunpaid household employees, and another 15–20% are microenterprise wage employees. Themajor microenterprise occupation is small-holder agriculture, which provides the principalsource of livelihood for 60% of the population,and a secondary income source for a further15% (Department of Census & Statistics, 1998).Nonfarm microenterprises––in petty trade,fishing, animal husbandry and a variety ofmanufacturing and service occupations––areclustered in regional towns and along thecoastal road.The case-study agencies are two nongovern-

mental organizations (NGOs): the Women’sDevelopment Federation (WDF) and theHambantota district branch of SarvodayaEconomic Enterprise Development Services(SEEDS). With memberships of 28,000 and10,000 respectively, they are the two largestMFIs in the district, overwhelmingly dominat-ing the local microcredit market in terms ofgeographic spread and portfolio size.The research was conducted over five months

in 1999. The primary research instrument was astructured questionnaire administered to 253respondents, followed by focus group discus-sions and in-depth interviews with 87 respon-dents, and interviews with MFI staff.Respondents were selected randomly frombranch membership lists, the only criterion forselection being a current outstanding balanceon a microenterprise loan. Three location-based categories were constructed: a semi-urban cohort of 115 households, located within3 km of a regional town or within a kilometerof the main coastal road; and two rural groups,28 of whom were drawn from the prosperousUda Walawe irrigation scheme, and 110 from

Table 1. Household income categoriesa

Household income group Definition

Extreme poor Up to 67% of poverty line: household income Rs. 3,350 or less

Poor 67–100% of poverty line: household income Rs. 3,351–5,000

Near-poor 100–150% of poverty line: household income Rs. 5,001–7,500

Nonpoor More than 150% of poverty line: household income more than Rs. 7,500

aAll microenterprise and household income data are expressed in rupees per month.

MICROENTERPRISE OCCUPATION AND POVERTY REDUCTION 1249

the poorer and more remote semi-irrigated andrainfed regions. The household poverty linewas set at Rs. 5,000 per month (about $70 at1999 exchange rates), an inflation-adjustedversion of the national poverty line developedby the World Bank in 1995 (World Bank,1995). Four household income categories wereconstructed, two below the poverty line andtwo above it (see Table 1).The study compares client incomes at the

time of their first MFI loan and at June 1999.Nearly 40% of respondents were recent recruits,having joined the programs in the two yearsprior to the survey, and another 25% hadjoined during 1994–97. Those who joinedbefore 1994 were asked to make June 1994 theircomparison point. Income data were derived byitemizing all household income sources, aver-aged over a 12-month period to take account ofseasonal fluctuations. Information on pre-loanincome levels was obtained principally viarespondent recall. To minimize the risk ofinaccuracies resulting from faulty recollections,respondents were asked to provide corrobora-tive data in the form of lists of housingimprovements and major household assetspurchased and sold since taking the first loan,and subjective assessments of changes in theirliving standards over the relevant period. Inaddition, recollected income estimates werecompared with historical data on agriculturalwage rates, welfare payments and other com-mon income sources, and apparent discrepan-cies were checked with respondents.

3. THE CLIENTS AND THEIRMICROENTERPRISES

This study draws on the classification typol-ogy developed by the Asian Development Bankand others, which groups microenterprises intolow-return ‘‘survival’’ activities and higher-return ‘‘entrepreneurial’’ activities. Entrepre-neurial microenterprises are larger, more highly

capitalized, employ more labor and use moresophisticated technologies. They tend to oper-ate continuously rather than intermittently,reinvest rather than consume surpluses, andhave lower closure rates. They are usuallyprimary household income sources, while sur-vival activities are usually secondary, and aremost commonly operated by men, while sur-vival activities are often operated by women.Their owners are better endowed with technicalskills and business acumen, and with ambition,self-confidence and other personal qualitieswhich are generally held to encourage entre-preneurship (Asian Development Bank, 1997;Cotter, 1996; Ghate, Ballon, & Manalo, 1996;Liedholm & Mead, 1999).The primary distinction between the two

categories arises from disparities in enterprisegrowth potential, which in turn depends onprevailing demand and production conditionsin each occupation. Entrepreneurial occupa-tions operate without excessive pressure fromcompetitors, while the survival occupations arecharacterized by low barriers to entry, undif-ferentiated products, saturated markets, andinefficiencies which limit their competitivenessin relation to larger producers. In entrepre-neurial occupations such as motor mechanicsand carpentry, microlevel producers haveadvantages of scale and access to suitable,affordable technologies which enable them tocompete effectively with nonmicro firms. Bycontrast, microenterprises in many survivaloccupations are not competitive, as the costsand scale of the more efficient productionprocesses employed by larger firms are wellbeyond the capacity of microlevel producers.As a result, microenterprises in many survival-level manual occupations face an uncertainfuture: Hambantota garment-makers, forexample, are unable to compete with factoryproducts; and at least one nonmicro rock-breaking firm, which uses motorized equipmentrather than the mallets and chisels used bymicrolevel producers, has opened for business


in one of the survey locations. Furthermore,entrepreneurial activities tend to be insulatedfrom seasonal fluctuations in input supplies,with regular access to inputs which are usuallyobtained in the cash economy. Survival enter-prises, on the other hand, often rely on inputswhich their owners gather themselves fromlocally available natural resources––such ascoconut husks for coir production, or paddyhusks to feed chickens or fuel brick-makingkilns––supplies of which vary seasonally, and

Table 2. Survival and entrepreneuria

Sector Occupation Fre

Survival

Agriculture Single season paddy

farming, chena

cultivation

Production and

service

Brick-making, coir

production, garment-

making, rock-breaking,

sea-shell crushing

Fishing and

livestock

Lagoon-fishing with

canoe, free-range poultry

and goat-rearing

Trade Small kiosk in owner’s

house or temporary

roadside stall, house-

to-house vending with

hand-cart

Entrepreneurial

Agriculture Dual/triple season paddy

farming, banana and

vegetable cultivation

Production, service,

fishing and

livestock

Carpentry, motor

mechanics, electrical

repairs, food preparation

and sale, jewelery-

making, hair-dressing,

tractor-hire, rice milling,

ocean-fishing with

motorized vessel, cattle

herding, large-scale

poultry-rearing

(500–2,000 birds)

Trade Agricultural wholesaling,

permanent retail

premises on main road

or town center, licensed

weekly market stall in

town center

Total

are often affected by increasing scarcity. Inaddition, production processes in survivaloccupations are prone to climate fluctuations:lagoon fishing, for example, is not viable duringthe annual dry season, and brick-makers usu-ally cease production during the monsoonmonths due to the difficulty of drying bricks inhumid conditions.The Hambantota microenterprises are clas-

sified on the basis of their median occupationincomes (Table 2). Occupations are defined as

l occupations: key characteristics

quency Median earnings Percentage earning

more than Rs. 4,000

163 2,000 8.7

59 3,000 –

40 2,200 15.0

25 1,400 –

39 1,800 15.4

90 6,000 66.0

16 4,300 56.3

43 6,100 65.1

31 6,900 71.0

253 3,000 36.0


entrepreneurial where their median incomesexceed Rs. 4,000, the earnings threshold which,at 80% of the household poverty line, providesa near-certain guarantee of poverty exit. Thesurvival microenterprises fit more neatly intothis categorization than do the entrepreneurialactivities: fewer than 10% of survival enter-prises earn more than Rs. 4,000, but around athird of entrepreneurial activities earn less. TheHambantota data nevertheless support findingselsewhere that occupation is a reliable predictorof microenterprise earnings (Asian Develop-ment Bank, 1997). Although actual earnings donot always reflect earnings potential, particu-larly in new microenterprises, the level at whichearnings will eventually plateau can be pre-dicted with reasonable certainty. This is espe-cially the case in the survival occupations.Lagoon fishing microenterprises, for example,constrained as they are by seasonality, intensecompetition and a dwindling resource base, areunlikely ever to earn more than Rs. 2,000 permonth. In entrepreneurial activities, the upperlimits to earnings potential are often harder todefine. Microenterprises which develop nichemarkets or employ new technologies may risewell above the occupational median income;conversely, those which face significant geo-graphic or financial constraints may struggle toapproach it. The underperforming entrepre-

Table 3. Entrepreneurial and survival activities: medi

Pre-loan

income

Entrepreneurial

occupations

Su

Median

earnings

Frequency Medi

earni

Extreme poor 1,600 3 1,40

Poor 4,000 17 2,00

Near-poor 4,700 26 3,00

Nonpoor 8,700 44 3,70

Total 6,000 90 2,00

Table 4. Initially below-poverty-line clients: median mic

Location Entrepreneurial activities S

Median

earnings

Frequency Medi

earni

Semi-urban 5,000 13 2,00

Arid rural 4,000 3 1,60

Uda Walawe 1,900 4 1,50

Total 4,000 20 1,80

neurial activities in the Hambantota sampleinclude a handful in the latter category, but themajority are new projects which commencedless than 12 months before the survey, andstand a good chance of eventually crossing theRs. 4,000 threshold.As Table 3 shows, the association between

microenterprise performance and householdincome is consistent at all income levels. Theincome-linked disparity in microenterpriseearnings is explained in part by the fact that thepoorest clients in each occupation tend to ownthe weakest microenterprises, but its principalcause is a pattern of occupational segregationin which below-poverty-line clients are clus-tered in survival occupations. The impact ofoccupation selection on earnings is striking. Inall income groups except the extreme-poor,microenterprises in entrepreneurial occupationsearn far more than those which are not. Onlya fifth of poor clients select entrepreneurialactivities, but among those who do, at least halfreach the poverty-clearing threshold. Theirmicroenterprises earn less than those of theirnear-poor and nonpoor competitors, but twiceas much as those of other poor clients in sur-vival occupations. The persistence of very lowmicroenterprise earnings among the extreme-poor, regardless of occupation, suggests theexistence of particular constraints facing this

an microenterprise earnings by initial income status

rvival occupations Total

an

ngs

Frequency Median

earnings

Frequency

0 36 1,400 39

0 63 2,500 80

0 42 3,400 68

0 22 6,500 66

0 163 3,000 253

roenterprise earnings by occupation type and location

urvival activities Total

an

ngs

Frequency Median

earnings

Frequency

0 32 2,000 45

0 63 1,800 66

0 4 1,600 8

0 99 2,000 119


group, an issue which is explored further in thefollowing section.Table 4 compares the microenterprise earn-

ings of initially below-poverty-line clients indifferent locations. Semi-urban clients earnmore than their rural counterparts in both sur-vival and entrepreneurial occupations. Even insemi-urban areas, however, survival activitiesare unlikely to enable poverty exit, with medianearnings well below the poverty-clearingthreshold. The evidence suggests that poor cli-ents are considerably better off selecting entre-preneurial activities regardless of location,although this conclusion is tentative in relationto rural clients because of the very small numberof observations. The most striking differencebetween the semi-urban and rural groups lies inoccupation selection: for reasons which arediscussed below, nearly a third of semi-urbanclients select entrepreneurial activities, in com-parison with fewer than 5% of the rural cohort.Table 5 examines changes in income status

during program participation. One-quarter ofinitially below-poverty-line households exitedpoverty after joining the program. (The netpoverty exit rate is somewhat lower, at 18%, asa handful of above-poverty-line households fellinto poverty during the period of participa-tion.) While the overall poverty impact ofmicroenterprise lending has been positive, it hasnot been uniform. Poverty exit was confined tothose who were initially just below the povertyline: not one extreme-poor household exitedpoverty, although nearly a quarter graduatedto ‘‘poor’’ status. Poverty exit rates were higherfor semi-urban clients (23%) than for the aridrural and Uda Walawe cohorts (9%); and poorclients in entrepreneurial occupations (65%)were more likely to exit poverty than those insurvival occupations (17%).While the MFIs have undoubtedly made a

substantial contribution to the overall improve-ment in living standards, it is important to notethat the reduction in poverty incidence cannot be

Table 5. Income status before a

Pre-loan income

status

1

Extreme-poor Poor Ne

Extreme-poor 30 9

Poor 3 47

Near-poor 1 6

Nonpoor 1 –

Total 35 62

attributed entirely to the impact of loans onmicroenterprise earnings, as the research did notcontrol formacroeconomic conditions and otherexogenous variables which are likely to affectmicroenterprise performance. Moreover, fewclients rely solely on their microenterprises. Inmost households the loan-assisted microenter-prise is one of a number of income sources,which include additional self-employmentactivities, local wage employment, remittancesfrom family members working in the WesternProvince and overseas, and government trans-fers. As microenterprises account on average foronly 56% of household incomes, other incomesources––notably remittances––are significantcontributors to changes in poverty status.

4. WHY THE POOR SELECT SURVIVALACTIVITIES

As the preceding discussion showed, entre-preneurial microenterprises appear to offer aroute out of poverty for the poor (although notthe extreme-poor), but the vast majority con-tinue to opt for survival activities. Rural clientsin particular appear to face significant barriersto entry to the entrepreneurial occupations.This section discusses the geographic, financialand sociocultural factors which influence ofthese factors occupation selection. The relativeimportance varies between locations: the ruralpoor select survival activities because thephysical and market environment presents fewalternatives, while in semi-urban areas, whereconditions favor the development of higher-value activities, human capital and sociocul-tural issues assume greater importance asdeterminants of project selection.

(a) Location

Occupations are unevenly distributed acrosslocations, as Table 6 shows. The distribution is

nd after program participation

999 income status

ar-poor Nonpoor Total

– – 39

25 5 80

33 28 68

2 63 66

60 96 253

Table 6. Distribution of occupations across locations

Semi-urban Arid rural Uda Walawe Total

Nonfarm 113 55 10 178

Survival 57 44 3 104

Entrepreneurial 56 11 7 74

Farm 2 55 18 75

Survival 2 55 2 59

Entrepreneurial – – 16 16

Total 115 110 28 253


particularly uneven with respect to entrepre-neurial activities, fewer than 15% of which arelocated in arid rural areas. Occupationaldiversity and microenterprise incomes arehighest in the densely populated, well-servicedregional towns and settlements along thecoastal road, which support a variety of non-farm activities. In the Uda Walawe settlements,where there is limited scope for nonfarmentrepreneurial occupations, farming providespoverty-clearing incomes for most households.The least propitious conditions are in arid ruralareas, where productivity and technologicalinnovation are impeded by inadequate irriga-tion and power infrastructure, demand is lim-ited by low local population densities and lowincomes, and a narrow range of viable occu-pations combines with low barriers to entry tocreate chronic market saturation.As Table 7 shows, there are striking location-

based disparities in income distribution. With apoverty incidence of 55%, rural clients are morethan twice as likely to be poor as their semi-urban and Uda Walawe counterparts. The factthat the poor are concentrated in remote andresource-deficient regions, where microenter-prise prospects are weakest, makes location akey factor in the income-related microfinanceimpact gap. It also illustrates a central paradoxof rural poverty reduction strategies based on

Table 7. Current household income

Household income group Semi-urban

Extreme poor 9.6

Poor 15.7

Near-poor 23.5

Nonpoor 51.2

Total 100.0

Median household income 7,650

microenterprise development, in that theyappear to be least effective precisely where theyare most needed––in regions which offer fewalternative income-generating opportunities,and in households which are least likely topossess the human and physical resourceswhich enable the best use of such opportunitiesas are available.Poor transport infrastructure is perhaps the

single most important impediment to ruralmicroenterprise development. While most semi-urban and Uda Walawe microenterprises areon or close to reliable bus services, road andtransport quality declines dramatically inremote rural areas. Many rural roads arenominally on bus routes, but in practice busservices are infrequent and unreliable due toinadequate road maintenance and poor regu-lation of the privatized public transport indus-try. During the monsoon season some roads areinaccessible by bus, isolating remote settle-ments for up to two months. In the mostremote areas, dwellings are located up to akilometer from an access route and can bereached only on foot or by bicycle. Poortransport services impose multiple productionand demand constraints on rural microen-terprises. Restricted access to inputs rulesout high-turnover manufacturing activities;and poor linkages limit the access of rural

distribution by location (percent)

Uda Walawe Arid rural All

7.1 20.0 13.8

17.9 35.5 24.6

32.1 21.8 23.7

42.9 22.7 37.9

100.0 100.0 100.0

7,020 4,680 6,250


producers to urban customers, and of ruralhouseholds to urban labor markets, withadverse impacts on local demand.Rural clients are similarly disadvantaged

with respect to power infrastructure. Two-thirds of semi-urban households are connectedto a mains supply, but in arid rural areas, wheremains power is confined to the centers of thelarger villages and the main access routes, onlya fifth of households have electricity. Lack ofelectrical power is a significant impediment tothe development of entrepreneurial micro-enterprises, as it rules out high value-addingoccupations such as carpentry and food pre-paration, which rely on electricity-based tech-nologies.The most varied and vigorous markets are in

town centers, which provide a relatively pros-perous local customer base and function ascommercial and administrative hubs for outly-ing settlements. Town centers support a varietyof microenterprises in various service occupa-tions and the production and trade of food,garments and other consumer goods. Anothercenter of economic activity lies along the arte-rial coastal road, which is the main route fromColombo to the popular pilgrimage center ofKataragama and Yala national park to theeast. Roadside commercial activity is domi-nated by motor repair workshops, tea-shopsand handicrafts producers serving a briskpassing trade of trucks, tourists and weekendbus loads of pilgrims.In arid rural areas, by contrast, low popula-

tion densities and low incomes impose severemarket constraints. Rural nonfarm survivalactivities are the weakest microenterprises inthe sample, with meager earnings and highfailure rates, accounting for 13 of the 18 failedprojects. They are disadvantaged relative totheir semi-urban competitors, as their remote-ness increases the cost and difficulty of reachingthem, restricting turnover and producers’ bar-gaining power. Most rely on local customers inregions where population densities are below100 per square kilometer. As most of theircustomers are farmers and farm laborers, theyare prone to sharp seasonal demand fluctua-tions and many close down during the annuallean season. High local poverty rates restrictoccupational diversity: in poor communities,demand for goods traded in the cash economyis virtually limited to food needs which cannotbe met from home production. As a result,rural nonfarm activities are clustered in pettytrade, fishing, backyard poultry-rearing and

other occupations which combine subsistenceproduction with trade in staple foodstuffs. Thenarrow range of viable occupations contributesto market saturation: in villages which cansupport perhaps a single food retailer, it iscommon to find three or four such enterprises,each with an identical product range of rice,lentils and dried fish, and each with earningswell below the poverty line. The difficulty ofmaintaining entrepreneurial activities in theface of location-related demand constraints isillustrated by the following example:

Example 1: Sunil lives in a sparsely populated aridrural neighborhood 3 km from the coastal road. Heworked as a carpenter’s assistant for several years be-fore opening a retail furniture-making business, oper-ating from a rented electrified workshop on the mainroad. During the first three years his wife borroweda total of Rs. 50,000 from the WDF over the courseof four loans, and the project’s monthly earnings grewfrom an initial Rs. 2,000 to a peak of Rs. 6,000. Afterthree years, Sunil was forced to vacate the workshopfollowing a dispute with the owner. Unable to rentalternative roadside premises, he tried operating thebusiness from his house, but without ready access topassing trade was forced to abandon the project aftera few months.

In the farm sector, entrepreneurial and sur-vival activities are distinguished by their accessto irrigation infrastructure. Hambantota dis-trict is in Sri Lanka’s Dry Zone, which has asingle monsoon season, making the cultivationof a second annual crop dependent on irriga-tion. The Uda Walawe scheme is part of anetwork of major government irrigationschemes designed in the 1970s and 1980s withthe aim of promoting agricultural exports andnational self-sufficiency in rice production.Farms in irrigated settlements are supported byconsiderable public investment, not only inirrigation but also in marketing and agricul-tural extension services, and high-qualitytransport networks links with the westernpopulation centers and ports. For those with ayear-round water supply, Dry Zone farming isan entrepreneurial activity, enabling the culti-vation of two or three annual paddy crops anddiversification into commercial fruit and vege-table production.Most farmers, however, are located in semi-

irrigated or nonirrigated arid regions whereagriculture is an increasingly unprofitable andrisky survival activity. Single-season paddy-farmers receive their water via canals fromcommunally maintained village reservoirs, or


‘‘tanks,’’ which are usually sufficient for a singlepaddy crop during the maha monsoon season,but rarely support a second crop. In the poorestDry Zone regions on the outskirts of irrigatedsystems and in jungle areas, where the tank-fedirrigation necessary for paddy cultivation doesnot exist, farmers practice rainfed chena, a formof slash-and-burn cultivation of drought-resis-tant cereal crops. As chena crops have a lowmarket value and much of the harvest isretained for consumption, cash incomes aretypically very low, averaging less than Rs. 1,000per month.The expansion of farming in Sri Lanka’s Dry

Zone has placed intense pressure on scarce landand water resources, and is threatening fragileeco-systems. In the Hambantota region popu-lation growth, low out-migration, and rising in-migration from the overpopulated Wet Zoneregions to the west have exacerbated an acuteshortage of arable land. The expansion of cul-tivation imposes environmental and economicproblems, with the increasing occupation ofjungle areas on the peripheries of canal net-works by the poorest farmers and newcomers.Deforestation resulting from chena is wide-spread, and the extension of paddy cultivationinto increasingly marginal areas strains thecapacity of village tanks. Faced with decliningyields, farmers either postpone the crisis bydiverting water from nearby streams, or aban-don their most recently cultivated fields, aprocess which further accelerates deforestationand evaporation. In the absence of official sta-tistics on the extent of illegal encroachment,reliable measures of its prevalence are notavailable, but evidence from recent studiessuggests that it is a significant and growingproblem (Institute of Policy Studies, 1998).In addition, policy and economic develop-

ments have adversely affected Dry Zone farm-ers. The world rice price has been decliningsince the 1980s; and in addition, the govern-ment, under pressure to reduce public spendingand liberalize trade policy, has withdrawnproducer price protection measures. Riceimport restrictions were removed in 1995, andthe state-run Paddy Marketing Board, whichtraditionally provided a guaranteed floor priceand acted as a buyer of last resort, was dis-mantled in 1999. As a consequence of thesedevelopments, the real farmgate price of paddyfell by 40% between the early 1980s and late1990s, with most of the decline occurring since1991 (Dunham & Edwards, 1997; Gunawar-dana & Somaratne, 1999). During 1997–98

alone the average producer price fell by 7%. Atthe same time, farmers are facing sharpincreases in input costs, with the scaling back ofsubsidies and privatization of input distributionnetworks. Poverty rates among farmers are wellabove the national average (Aturupane, 1999),and incomes from paddy cultivation are inlong-term decline, with little prospect of areversal in the near future. The consensus indonor and government policy circles is thatsmall-scale paddy-farming is economically un-viable outside the major irrigation schemes(Central Bank, 1998).

(b) Financial factors

The high recurrent costs of some entrepre-neurial enterprises and the ‘‘lumpiness’’ of theirfixed assets create barriers to entry, especiallywhere capital requirements exceed the MFIs’group-based loan size limits (Rs. 30,000 in theWDF and Rs. 50,000 in SEEDS), placing thembeyond the reach of clients who lack additionalfinance. There is a vast gulf between the capitalrequirements of the highest-earning enterprisesand the financial capacity of low-income cli-ents: as Table 8 shows, the monthly expenses ofhigh-value trade occupations represent close toa year’s household income at the poverty line.Even where sufficient credit is available, poor

clients are reluctant to expose themselves to therisks associated with large loans. As example 2illustrates, not only do the most profitableenterprises require large initial capital outlays;they may also undergo prolonged gestationperiods, with low or negative returns during thefirst months of operation. As poor clients can-not afford to operate loss-making concernseven for short periods, they tend to opt for low-risk working capital investments, often com-bining production for home consumption andthe market, which generate meager but imme-diate cash flows.

Example 2: Ranjith owns a ‘‘communications shop’’in a town center, one of two local businesses providinglong-distance telephone and fax services. He openedthe business with an initial investment of Rs. 80,000from his personal savings and a Rs. 200,000 loan froma local bank, with which he installed a telecommunica-tions cable and bought a second-hand photocopierand fax machine. Since joining SEEDS he has takenthree microenterprise loans with a total value of Rs.70,000, and reinvested a substantial proportion ofenterprise profits. He plans to buy a computer andset up an email and internet service. The businessshowed early promise, with gross monthly earnings

Table 8. Mean expenditure on fixed assets and working capital by occupation

Enterprise occupation Mean monthly working capital

and recurrent expensesaMean value of

enterprise assetsb

Survival occupations 6,900 10,300

Agriculture 9,000 14,600

Production and service 4,100 3,500

Fishing and livestock 1,100 2,700

Trade 10,400 15,700

Entrepreneurial occupations 30,900 60,200

Agriculture 18,700 55,000

Production, service, fishing and

livestock

20,400 59,000

Trade 51,900 64,600

Total 15,600 27,400

a Includes raw materials and supplies, wages, rent, transport, fuel and power.b Includes tools, plant, vehicles, livestock. Excludes land and pre-existing buildings but includes building constructionand improvements undertaken since start of project. Details of enterprise asset ownership were taken from a rep-resentative sub-sample of 87 borrowers. Building improvements were assigned a value on the basis of the respon-dent’s recollection of expenditure. Other assets were assigned values on the basis of 1999 retail prices, and do notinclude depreciation.


of about Rs. 5,000; but yielded negative returns duringits first year of operation, as earnings were insufficientto meet repayments on the bank loan. After threeyears net earnings increased to a stable Rs. 15,000.Ranjith was able to meet his high fixed capital costsand support the business through its initial lean periodbecause he is independently well-off: his family ownsfour acres of irrigated paddy-land outside the town,and he combines the business with a second job asan insurance agent.

Financial barriers to entry create a severeimpediment to poverty reduction in rural areas,as the main nonfarm entrepreneurial activi-ties––tractor-hire businesses, rice mills andvegetable wholesaling––have very high capitalrequirements which effectively rule them out asoptions for poor clients. In semi-urban areas,capital constraints are less restrictive. While thelargest entrepreneurial activities, such as thatdescribed in example 2, are not viable optionsfor poor borrowers, many other activities canbe established with moderate initial outlays––on electricity installation, refrigerators, orpower tools––which are accessible via smallloans. Example 3, which illustrates the capacityof credit to galvanize microenterprises whichoperate in favorable demand and productionconditions but face capital constraints, showshow microenterprise lending can supportincremental capital improvements which arewithin the reach of many poor clients. As evenmid-range entrepreneurial activities involve

significantly higher costs than survival micro-enterprises, however, financial considerationsare likely to deter the poorest, most risk-aversesemi-urban households from selecting them.

Example 3: Shiranthi operates a 16-year old tea-shopon the coastal road. During its first years she financedthe business with working capital loans from money-lenders at 10% interest per month. Due to the highcost of credit and limited demand, the shop earned amarginal income during its early years, and Shiranthimade few improvements. Encouraged by the improv-ing fortunes of neighboring businesses with increasesin vehicle traffic in the early 1990s, she decided thatthe shop would benefit from a larger injection of cap-ital, and joined SEEDS in 1995. She has taken threeloans with a total value of Rs. 50,000, with whichshe replaced the original thatched hut with a perma-nent structure, installed electricity, bought chairsand tables, and made down payments on a refrigera-tor and ceiling fan, the balance of which is being fi-nanced by enterprise profits. The initial customerbase of local people has expanded to include truck-drivers and the weekend pilgrim trade, and net earn-ings grew from about Rs. 3,000 in 1995 to about Rs.6,000 in 1999.

(c) Human capital and sociocultural factors

The poor are least likely to possess entre-preneurial aptitudes and aspirations. TheHambantota data lend support to findingselsewhere which indicate that poor clients often


have little understanding of market conditionsand business opportunities in nontraditionaloccupations (Gunatilaka, 1996; Shaw, 1999),and lack the extended social networks whichare key sources of information, skills andbusiness contacts (Barton, 1997; Gunatilaka,1997). They are reluctant to develop the unfa-miliar skills required by the more complexnontraditional microenterprises, and manyreported that the familiarity of the tasks andbusiness relationships associated with tradi-tional survival activities influenced their occu-pation selection decisions. Near-poor andnonpoor respondents were more likely to sur-vey the local environment and base their deci-sion on identified market opportunities, but theresponses of poor clients indicated a tendencytoward a passive approach to enterprise selec-tion: few had investigated alternative occupa-tions, and they were far more likely than othersto cite ‘‘copying others’’ as the primary deter-minant of their choice.There was evidence that sociocultural factors

hinder the development of high-value mic-roenterprises, particularly among poorer cli-ents. In centers of commercial activity, wherethe growth of the cash economy and new formsof economic organization have broken downtraditional patron-client relationships and theirassociated ties of mutual obligation, businessdecisions tend to be based on commercialconsiderations rather than noneconomic crite-ria. In more remote communities, however,social norms may discourage individualism andupward mobility in the poor (Brow, 1996). Theprospect of conflict may deter poorer producersfrom challenging established economic rela-tionships, for example, by introducing mar-keting arrangements which bypass influentialvillage traders (Menike, 1992).In Hambantota, the accumulation of wealth

among the nonpoor appears socially accept-able, but is frowned on when it occurs in poorhouseholds. Respondents were inclined to becritical of poor neighbors who had ‘‘madegood’’; and some appeared to view their ownimproved circumstances as a mixed blessing,emphasizing its less favorable consequences inthe form of unwelcome attention from enviousneighbors and relatives in search of support.Conceptions of the communal good sometimestake precedence over individual gain: whereidentical microenterprises are clustered in asingle community, a common arrangementwhere occupations are caste-based or requireaccess to a localized natural resource, individ-

ual producers face strong social pressure not toengage in price competition. Among brick-makers, for example, competition in terms ofoutput was socially acceptable, and there wereno restrictions on taking on extra labor orworking well into the night, but no producerwas prepared to risk social ostracism bylowering his prices.Low caste and social status may inhibit entry

into entrepreneurial occupations, reducing therange of activities open to extreme-poor andsome poor households. Caste appeared to havea significant impact on opportunity in fishingcommunities, which are physically segregated,and where interactions with members of highercastes are limited to traditional economictransactions such as fish-trading and labor-hire.While low-caste borrowers may be able todevelop enterprises which trade in nonlocalmarkets, social discrimination is likely torestrict their access to retail and service activi-ties which target local customers.Where enterprises depend on the support of

local politicians and officials, there are barriersto entry for the poor and unconnected. Somesemi-urban microenterprises breach zoning andbuilding regulations, occasionally giving rise toattempts by police and other officials to extract‘‘protection’’ payments. Connections with pol-iticians and high-level officials protect ownersfrom such harassment, which may otherwiseconstitute a debilitating unofficial tax. In themobile market-stall trade, which is regulated bya licensing system, permits to set up stalls arescarce and fiercely contested, and were reportedto depend on access to local politicians andofficials. Due to the civil conflict prevailing atthe time of the survey, vehicles using the mainroads were regularly stopped and searched, andvegetable wholesalers transporting goods toColombo reported that letters of introductionfrom the local police chief were essential inorder to avoid long delays at road-blocks.Poorer households are more likely to face

human capital deficiencies. Their mobility andproductivity is often limited by poor healthand undernutrition. Many lack the technicaland business skills required for higher-valuemicroenterprises. They have lower levels offormal education: among below-poverty-lineand above-poverty-line clients, primary schoolcompletion rates were 50% and 80% respec-tively. The household labor supply is morelikely to be limited: poor and extreme-poorhouseholds contained an average of 1.4 able-bodied adults, in comparison with an average


of 1.9 in near-poor and nonpoor households.Extreme-poor households are far more likely tobe female-headed, at 17%, in comparison with a7% female-headedness ratio for the sample as awhole, and are also more likely to be headed bya disabled or elderly adult. In some very poorsemi-urban households, entrepreneurial pros-pects are further restricted by illiteracy andsocial problems such as alcohol abuse,engagement in petty crime and family instabil-ity.

Example 4: Seela lives with her 7-year old son in a re-mote region 10 km from Hambantota town. Her fam-ily’s main source of livelihood was single-seasonpaddy-farming until the year prior to the survey, whenthe family paddy fields were occupied by her brother-in-law following her husband’s death in a road acci-dent, leaving her without a source of income. Shehas taken two loans from the WDF of Rs. 2,500 eachto purchase trailer-loads of granite chunks to be bro-ken into smaller rocks. As she is undernourished andlacks the stamina required for the demanding physicallabor of rock-breaking, and her son is too young toassist her, she takes two months to complete each loadof rocks, a task which most other producers completein a month. The microenterprise provides a netmonthly income of about Rs. 500 after loan repay-ments are deducted.

In rural areas, social restrictions on women’seconomic activity limit their entrepreneurialprospects. Paddy-farming and fishing arerestricted by custom to men, often with devas-tating consequences for female-headed house-holds, as example 4 illustrates. In semi-urbanareas, men predominate in the profitable skilledmanual occupations and mobile trading enter-prises, while women are clustered in low-valuehome-based activities such as garment-makingand coir production.Not all high-value semi-urban occupations

are reserved for men, however. Women arepresent in significant numbers in food pro-cessing; furthermore, the gender boundariesaround nonfarm occupations are less rigid thanthose associated with agriculture and fishing,and they are not completely closed to women.Interestingly, female household heads appearedto have more freedom than married women inproject selection and mobility. Most womenwho enter male-dominated entrepreneurialoccupations are household heads: one widow,for example, assisted by SEEDS-sponsoredtechnical training, was making a moderatesuccess of a semi-urban welding workshopformerly operated by her husband; another wasoperating a mobile vegetable trading business.

These successes indicate that in semi-urbanareas at least, femaleness is not an absolutebarrier to the development of entrepreneurialactivities, and suggest scope for NGO genderawareness programs promoting women’s entryinto higher-value nontraditional occupations.

5. THE CHALLENGE OF SERVING THEPOOR

SEEDS and the WDF have traditionallytargeted a poor, predominantly rural constitu-ency of single-season farmers, landless laborersand fishing communities, but the compositionof the membership base has shifted in recentyears toward less-poor clients in semi-urbancenters and the Uda Walawe settlements.Among the sample, only 37% of new membersjoining during 1996–99 were below the povertyline, compared with 65% of those who joinedbefore 1996, and the proportion of new clientsfrom arid rural areas fell from 56% to 21% overthe same period. Both MFIs have targeted anonpoor clientele in their recent recruitmentcampaigns. In addition to the group-guaranteelending schemes in which poorer clients par-ticipate, they have introduced larger, collateral-based loan packages designed to cover thecapital costs of the largest entrepreneurialactivities. The move up-market is in part aresponse to pressure from donors to reducereliance on subsidized funds and improveinstitutional stability, and gained considerablemomentum from a crisis in cultivation loanrepayments following a drought in 1996.The MFIs employ a two-pronged strategy to

serve their increasingly diverse client base. Theyactively promote entrepreneurial occupations,particularly for the nonpoor, while continuingto support survival activities for rural clientsand the semi-urban poor. They have, however,substantially reduced their support for cultiva-tion in favor of nonfarm activities. Theirapproach has worked well for near-poor andnonpoor clients, but they face challenges inmeeting the needs of the semi-urban poor, andin retaining relevance for their traditional ruralpoor constituencies.

(a) Promoting entrepreneurial activities forpoverty reduction

Both MFIs supplement their lending withnonfinancial microenterprise support services.SEEDS in particular has a highly successful


business development program which hasattracted considerable support from interna-tional donors, with whom it has collaboratedon a variety of projects. Services provided bySEEDS include the identification and promo-tion of low-cost technologies, individual busi-ness advice, and assistance in developinglinkages with markets and suppliers. Theseservices have had considerable success indeveloping established entrepreneurial activi-ties, but little direct impact on poverty, as theowners of such enterprises are mostly nonpoor.There is some evidence that support for high-

end microenterprises may reduce poverty indi-rectly, via employment generation. The generalconsensus is that microenterprises play a sig-nificant role in absorbing unemployment, butas the vast majority are either self-employmentactivities or household-level operations whichemploy unpaid family members, they are notsignificant contributors to job growth outsidethe household (Asian Development Bank,1997; Daniels, 1999; Liedholm & Mead, 1999;Orlando & Pollack, 2000). The impact of crediton paid microenterprise employment, whilepositive, is small (Sebstad & Chen, 1996). Tothe extent that microenterprises do generatejobs, these effects are concentrated in the larg-est, most successful activities (Hulme & Mos-ley, 1996; Mosley, 2001). This was also the casein the Hambantota sample, as Table 9 shows.Wage employment in survival activities is pre-dictably negligible, but nearly a third of entre-preneurial activities employed at least onenonfamily member. Although the povertyimpacts of this quite impressive rate of jobcreation are offset to some extent by the factthat most of the jobs are of very poor quality,being low paid, part-time and intermittent, thedata suggests that support for the enterprises of

Table 9. Nonfarm microenterpris

Enterprise occupation Hous

memb

Survival occupations 98

Production and service 35

Fishing and livestock 24

Trade 39

Entrepreneurial occupations 51

Production, service, fishing and livestock 27

Trade 24

Total 149

the nonpoor may be justified on povertyreduction grounds.Relatively little attention has been given to

the promotion of entrepreneurial activitiesamong the semi-urban poor. The high pro-pensity of this group to opt for survival activ-ities indicates a need for interventions targetingproject selection. SEEDS provides little assis-tance at the point of project selection, however,confining its support for prospective clientsmainly to vocational training in both survivaland entrepreneurial occupations. The mostwell-attended training programs continue to bein traditional survival activities: of the 17nonfarm occupations covered by SEEDS-sponsored training courses in the first half of1999, more than a quarter of participants optedfor just two courses in garment-making andcoir production (SEEDS, 1999). The contentand structure of training courses for entrepre-neurial occupations could be better adapted tothe needs of poor clients. Few can afford toforgo their regular income-generating activitiesin order to attend courses, which sometimesrequire full-time residential attendance outsidethe district. Although training costs are subsi-dized, some find the fees prohibitive, and theremay be a case for the further subsidization ofselected courses for the most promising poorclients.Not only are poor clients unlikely to

participate in training for entrepreneurialoccupations; among those who do participate,post-training microenterprise take-up rates areeven lower. It appears that vocational trainingalone is unlikely to overcome the obstacles ofinformation deficits and anti-entrepreneurialcultural mores. The facilitation of informedselection decisions regarding unfamiliar occu-pations requires the integration of vocational

es: contribution to employment

ehold

ers only

One wage

employee

More than one

wage employee

3 3

3 2

– 1

– –

10 13

9 7

1 6

13 16


training with information on markets, inputsand technologies. In addition, the promotion ofappropriate attitudes through gender aware-ness and other conscientisation programs isneeded to enhance the participation of womenand other disadvantaged groups in nontradi-tional activities. The need for additional sup-port services is illustrated by the example of apoor semi-urban client who, after taking a 10-day screen-printing course, decided againstpersevering with the project, although credit tocover the Rs. 25,000 start-up cost was avail-able. His decision was motivated not by aninformed assessment of potential risks andbenefits, but by a poor knowledge of the mar-ket: he knew of no successful screen-printingprojects in his locality and had little awarenessof demand conditions and opportunities.The MFIs could take a more proactive role

in identifying and promoting occupationswhich are suitable for poor borrowers. Foodpreparation, electrical and mechanical repairs,carpentry and personal services such ashair-dressing are among the most promisingactivities for the semi-urban poor. They arefinancially accessible, as they rely on humancapital and incremental additions of technologyrather than large initial capital investments.Furthermore, they have considerable potentialfor sustained growth, as they are well-posi-tioned to take advantage of proximity to theircustomer bases, and operate in strong marketswith little competition from larger producers.The emerging domestic and international

tourism sector offers considerable potential formicrolevel development. With increaseddomestic activity and the growth of a metro-politan middle class following the liberalizationpolicies of the 1980s and 1990s, together withthe removal of restrictions on vehicle imports,there has been a substantial increase in non-metropolitan vehicle traffic. The wildlife parks,beaches and pilgrimage centers in the Ham-bantota area have potential for further devel-opment as tourist attractions, and the district’spopularity among tourists appears to begrowing: at the time of the survey an interna-tional hotel and a number of guest houses hadrecently opened on the coastal road. Althoughtourism has generated vigorous microlevelindustries in the production of food, handi-crafts, jewelery and garments in Sri Lanka’ssouthwestern beach resorts, microenterprisestargeting tourists are virtually nonexistent inHambantota district. As tourism favors thefemale-dominated food and garment trades, it

presents opportunities for the development ofentrepreneurial activities for women. TheMFIs could encourage such activities throughappropriate awareness-raising and traininginterventions, including language trainingand cultural awareness programs aimed ataddressing a generally poor understanding ofthe market among Sinhalese microentrepre-neurs.As distance and small local markets impose a

ceiling on the region’s capacity to absorb newentrants, poorly conceived interventions riskcreating a problem of oversupply. There maybe a case for restrictions on the supply of creditto new starts in crowded markets. Further-more, it is important to ensure that occupationstargeted for training are appropriate for thelocal environment. District-level SEEDS staffreport that vocational training programs aredesigned in the Colombo head office withoutlocal input, and sometimes have limited rele-vance to local conditions, citing as an examplea recent course in the production of hand-bags, an activity which may be well-suited toColombo markets but for which there is littledemand in Hambantota. An abundance oftraining programs in occupations which havelittle chance of succeeding locally is wastefuland may ultimately reduce the credibility ofnontraditional options.

(b) Support for survival microenterprises

Income promotion strategies, which remainthe primary strategy of most microfinanceprograms, focus on poverty reduction, aimingto move clients ‘‘from a stable ‘below-poverty-line’ situation to a stable ‘above-poverty-line’situation’’ (Hulme & Mosley, 1996, p. 106).Income protection strategies aim to mitigatethe ill-effects of poverty by stabilizing con-sumption, rather than eliminate it by increasingincomes. Very poor clients derive little benefitfrom approaches based solely on income pro-motion, although their well-being may be sub-stantially improved by services which supportcontinuity of access to basic survival needs andempower disadvantaged individuals (Hashemi,1997; Hulme & Mosley, 1996; Montgomery,Bhattacharya, & Hulme, 1996; Rutherford,2000; Sebstad & Cohen, 2000; Zaman, 1999).Carefully planned combinations of protectionaland promotional services may deepen outreachand enable some very poor clients to exitpoverty, as found by a recent study ofBRAC’s poverty-focused IGVGD Program.


The authors found limited impacts on thepoorest and most disadvantaged clients, how-ever, indicating that even when accompaniedby protectional services, income promotionrequires a minimum threshold of economiccapacity (Matin & Hulme, 2003). Incomeeffects are lowest where clients, faced withsevere human capital deficiencies or a hostilephysical and market environment, are least ableto use credit for productive investment.SEEDS and the WDF have judged that the

costs of supporting single-season agricultureoutweigh the benefits, and have adopted apolicy of reducing their exposure to farming infavor of nonfarm microenterprises in arid ruralareas. Their decision was prompted in part byconcerns regarding the environmental andeconomic sustainability of single-season culti-vation, and that access to microcredit may infact work against the best interests of farminghouseholds by encouraging them to remain in adeclining sector. Considerations of institutionalviability also played a part in their decision,which followed a long-term deterioration in therisk profile of single-season farming, and wasprecipitated by a massive harvest failure fol-lowing the 1996 drought, which caused up to90% of cultivation loans to fall into arrears insome rural branches, threatening the viabilityof the MFIs’ district-wide lending programs.From a lender’s point of view, rural nonfarmactivities are less risky than agriculture, as thenonfarm sector is less subject to destabilizing enmasse failures. The lending risk is compoundedby the large working capital requirements ofpaddy cultivation, which are more than twice ashigh as those of most nonfarm activities withcomparable net incomes (see Table 8).Since 1996 the MFIs have imposed a range of

restrictions on single-season cultivation lend-ing. Both have capped loan sizes at Rs. 20,000,and SEEDS has imposed an interest surchargeand a 10% insurance levy on paddy cultivationloans. These restrictions have generated anti-poor trade-offs, as they oblige farmers to sup-plement their microcredit loans with costlyborrowings from moneylenders, and they bearmost heavily on very poor remote householdswhich are least able to develop nonfarm alter-natives. For example, SEEDS requires appli-cants to produce evidence of the legality of theirtenure of the land they intend to cultivate, apolicy which effectively excludes the poorestchena farmers and marginal paddy-growers.Despite the difficulties facing the farm sector,

and the obstacles to accessing credit, paddy

cultivation remains the preferred activity formost rural borrowers. 1 Farming is the mostimportant source of livelihood in Hambantotadistrict, and paddy remains by far the largestsmallholder crop, occupying over 70% ofcropland (Department of Census & Statistics,1998). As a rule, when households can growpaddy, they do: as rice contributes around halfthe calorie intake of rural households, paddycultivation supports food security in an envi-ronment where undernutrition is endemic. Itis a mark of social status in Sinhalese cul-ture, providing farming households with asecure sense of their place in the commu-nity; and it generates substantially highercash returns than nonfarm alternatives (seeTable 2). Efforts to ‘‘steer beneficiaries awayfrom paddy’’ have had limited success, afact which is privately conceded by local MFIstaff. Faced with restrictions on cultivationlending, many farmers have dropped out ofthe MFI programs, preferring to maintaintheir farms with informal sector creditrather than switch to nonfarm microenter-prises.While low poverty exit rates in rural Ham-

bantota demonstrate the limitations of mic-roenterprise development as a promotionalstrategy, survival enterprises serve a variety ofprotectional purposes. Cultivation loans sup-port food production and reduce farmers’dependence on moneylenders; and nonfarmloans support subsistence production of eggsand fish, facilitate risk-spreading throughdiversification, provide short-term stopgapsduring the lean season when farm incomes andlaboring work are scarce, and provide womenwith independent incomes. There may be a casefor continuing to support rural survival enter-prises on the basis of their protectional value,but the decision needs to be based on a carefulcost-benefit evaluation. The considerable costsinclude negative externalities in the form ofenvironmental degradation and high lendingrisks for farm projects; and in nonfarm pro-jects, the creation of excessive competition inweak markets. In addition, there may be asignificant opportunity cost associated with thediversion of resources from alternative pro-motional interventions. There is a strong casefor investigating strategies which may offerbetter prospects for rural poverty exit, parti-cularly those aimed at improving access tononlocal labor markets, which for many ruralhouseholds offer the most viable route out ofpoverty.


6. CONCLUSION

In summary, evidence from the Hambantotasample does not support claims that microen-terprise credit is a broadly effective solution topoverty, although it can work well for clientswho are close to the poverty line and live inenvironments with the conditions necessaryto sustain high-value microenterprises. Asoccupation selection is a key determinant ofpoverty exit, interventions encouraging thesemi-urban poor to select entrepreneurialoccupations are likely to further improve theirprospects. In rural areas, the chief impedi-ments to rural microenterprise development arelargely beyond the control of the programs.Rural microenterprises serve a protectionalfunction, but offer limited prospects for povertyexit, and involve significant environmental andinstitutional costs and risks. For the MFIs,these ambiguities call for a close examination oftheir rural development strategies: the benefitsof support for rural survival enterprises need tobe evaluated and justified, rather than assumed.A final point relates to the impact of public

investment in rural infrastructure on micro-enterprise development. Sri Lanka’s physicalinfrastructure deteriorated considerably in the1990s, as the brunt of fiscal adjustment fell oncapital expenditure, which declined as a pro-portion of GDP from 12% in 1988 to less than5% by the mid-1990s, well below the averagefor low and lower-middle-income Asian coun-tries. 2 By the late 1990s observers were warn-ing that reductions in capital spending werejeopardizing the country’s economic andhuman resource base (Institute of Policy Stud-ies, 1998; World Bank, 1998).Recently, however, there have been encour-

aging signs of renewed activity in infrastructureprovision. The government’s 2002 PovertyReduction Strategy Paper includes a commit-ment to ‘‘pro-poor growth’’, with initiativesaimed at revitalizing rural development andimproving the national infrastructure. Therehas been an increase in government capitalspending, which rose to 5.9% of GDP in 2001,and a commitment, embodied in the PRSP, toachieve 7.3% by 2006; together with the com-mencement of several donor-supported projectsin the power, telecommunications and trans-port sectors, with substantial increases in donorcapital funds projected over the next three years(Government of Sri Lanka, 2002). Of particularsignificance for the Hambantota region is theSouthern Highway Project, due for completion

in 2006, which will link Colombo with thesouthern cities of Galle and Matara. Althoughthe proposed route stops 80 km short ofHambantota town it will, when established,substantially reduce travel times, increase thevolume of nonlocal traffic in the region andimprove the access of local producers to urbanand export markets.While the revival of infrastructure invest-

ment is a welcome development, ongoingbudgetary constraints and policy decisions totarget the high-growth urban and export sec-tors as priorities for capital spending meanthat its impact on poverty-focused ruralinfrastructure is likely to be limited in theforeseeable future. Road development projectscontinue to prioritize strategic national high-ways and the Western Province over the ruraltertiary road network, and power sector pro-jects focus mainly on adding to the capacity ofthe existing grid rather than extending suppliesto new rural users. Moreover, there is a riskthat the privatization of utilities, a strategyactively supported by the World Bank andother donors, will further skew benefits awayfrom the poor, given the high costs ofextending infrastructure to remote areas, thelimited ability of rural users to pay for it and apolicy climate which is generally hostile tosubsidized services.It is no coincidence that the growth of global

enthusiasm for microfinance has taken place ina context of shrinking resources for ruraldevelopment and social services. The dis-appointing performance of microenterpriselending in rural Hambantota highlights theshortcomings of policy prescriptions which seekto privatize the costs of poverty reduction.Development strategies based on self-relianceare unlikely to have much effect in the absenceof an enabling environment which supports theefforts of the poor to develop their productivecapacity. Microenterprise development pro-grams need to be complemented by investmentin social and physical infrastructure; they areno substitute for it. The reorientation ofdevelopment spending towards rural capitalprojects may well go further than any otherinvestment towards reducing poverty. Asdevelopment agencies with a strong nationalprofile, SEEDS and the WDF should make useof their considerable credibility with the SriLankan government and international donorsto lobby for expanded capital spending on ruralservices that are accessible and affordable bythe poor.


NOTES

1. Paddy production is sensitive to short-term fluctua-

tions in economic, climatic and domestic security condi-

tions, but overall it has shown remarkable resilience to

long-term sectoral deterioration. In 1994–95, with good

weather conditions and a general mood of optimism

following the election of a new government and the

prospect of an end to the civil conflict, record acreages

were sown with paddy. A drought reduced 1996–97

output, but in 1998 paddy production bounced back,

rising by 20% over 1997, despite a sharp fall in producer

prices over the period (Central Bank, 1998). There is little

evidence that sectoral decline has dampened the prefer-

ence for paddy: on the contrary, its continuing popular-

ity, together with recent increases in chena production,

suggests that many poorer farmers are withdrawing from

the nonfarm sector and unstable commercial crop

markets in favor of semi-subsistence crops.

2. In the mid-1990s Sri Lanka’s investment in core

infrastructure, at 4.5% of GDP, compared unfavorably

with that of Indonesia (5.5–6%), Thailand (6–6.5%) and

the Philippines (7%) (World Bank, 1998).

REFERENCES

Asian Development Bank (1997). Microenterprise devel-opment: not by credit alone. Asian DevelopmentBank, Manila.

Aturupane, H. (1999). Poverty in Sri Lanka: achieve-ments, issues and challenges. Ministry of Finance andPlanning, Colombo.

Barton, C. (1997). Microenterprise business developmentservices: defining institutional options and indicators ofperformance. Microenterprise Best Practices Project,Washington, DC, USAID.

Brow, J. (1996). Demons and development: the strugglefor community in a Sri Lankan village. University ofArizona Press, Tucson.

Central Bank of Sri Lanka (1998). Annual report.Central Bank of Sri Lanka, Colombo.

Copestake, J., Bhalotra, S., & Johnson, S. (2001).Assessing the impact of microcredit: a Zambian casestudy. Journal of Development Studies, 37(4), 81–100.

Cotter, J. (1996). Distinguishing between poverty allevi-ation and business growth. Small Enterprise Devel-opment, 7(2), 49–52.

Daniels, L. (1999). The role of small enterprises in thehousehold and national economy in Kenya: asignificant contribution or a last resort? WorldDevelopment, 27(1), 55–65.

Department of Census and Statistics, Sri Lanka(1998). Statistical handbook: Hambantota dis-trict, Department of Census and Statistics,Colombo.

Dunham, D., & Edwards, C. (1997). Rural poverty andan agrarian crisis in Sri Lanka, 1985–95: makingsense of the picture. Institute of Policy Studies,Colombo.

Ghate, P., Ballon, E., & Manalo, V. (1996). Povertyalleviation and enterprise development: the need fora differentiated approach. Journal of InternationalDevelopment, 8(2), 163–178.

Government of Sri Lanka (2002). Regaining Sri Lanka:vision and strategy for accelerated development. Gov-ernment Printing Office, Colombo.

Gunatilaka, R. (1996). The National Development TrustFund. Institute of Policy Studies, Colombo.

Gunatilaka, R. (1997). Credit-based, participatory pov-erty alleviation strategies in Sri Lanka: What have welearned? Institute of Policy Studies, Colombo.

Gunawardana, P. J., & Somaratne, W. G. (1999). Non-plantation agricultural economy of Sri Lanka: trends,issues and prospects. Paper Presented at 7th AnnualConference on Sri Lanka Studies, Canberra, Decem-ber.

Hashemi, S. M. (1997). Those left behind: a note ontargeting the hardcore poor. In G. D. Wood & I. A.Sharif (Eds.), Who needs credit? Poverty and financein Bangladesh (pp. 249–257). London: Zed Books.

Hulme, D., & Mosley, P. (1996). Finance against poverty,Vol. 1. London: Routledge.

Institute of Policy Studies (1998). Sri Lanka: state of theeconomy 1998. Institute of Policy Studies, Colombo.

Liedholm, C., & Mead, D. C. (1999). Small enterprisesand economic development: the dynamics of micro andsmall enterprises. London: Routledge.

Matin, I., & Hulme, D. (2003). Programs for thepoorest: learning from the IGVGD program inBangladesh. World Development, 31(3), 647–665.

Menike, K. (1992). People’s empowerment as the peoplesee it. Colombo: INASIA.

MkNelly, B., & Dunford, C. (1998). Impact of credit witheducation on mothers and their young children’snutrition: lower Pra Rural Bank Credit with educationprogram in Ghana. Research Paper No. 4, Freedomfrom Hunger, Davis, CA.

MkNelly, B., & Dunford, C. (1999). Impact of credit witheducation on mothers and their young children’snutrition: CRECER credit with education programin Bolivia, Research Paper No. 5, Freedom fromHunger, Davis, CA.

Montgomery, R., Bhattacharya, D., & Hulme, D.(1996). Credit for the poor in Bangladesh: TheBRAC Rural Development Program and the Gov-ernment Thana Resource Development and Employ-ment Program. In D. Hulme & P. Mosley (Eds.),Finance Against Poverty (Vol. 2, pp. 94–176). Lon-don: Routledge.

Mosley, P. (2001). Microfinance and poverty in Bolivia.Journal of Development Studies, 37(4), 101–132.


Orlando, M. B., & Pollack, M. (2000). Microenter-prises and poverty: evidence from Latin America.Inter-American Development Bank, Washington,DC.

Rahman, R. I. (1997). Poverty, profitability of microen-terprises and the role of NGO credit. In G. D. Wood& I. A. Sharif (Eds.), Who needs credit? poverty andfinance in Bangladesh (pp. 271–287). London: ZedBooks.

Rutherford, S. (2000). The poor and their money. NewDelhi: Oxford University Press.

Sebstad, J., & Chen, G. (1996). Overview of studies onthe impact of microenterprise credit. AIMS Project,USAID, Washington, DC.

Sebstad, J., & Cohen, M. (2000). Microfinance, riskmanagement and poverty. AIMS Project, USAID,Washington, DC.

SEEDS (1999). Quarterly progress report, SEEDS,Colombo, June 1999.

Shaw, J. (1999). A world bank intervention in the SriLankan welfare sector: the national developmenttrust fund. World Development, 27(5), 825–838.

Todd, H. (2000). Poverty reduced through microfinance:the impact of ASHI in the Philippines. Washington,DC: USAID.

United Nations Development Program (1998). Nationalhuman development report. Sri Lanka, UnitedNations, Colombo.

World Bank (1995). Sri Lanka poverty assessment.World Bank, Washington, DC.

World Bank (1998). Sri Lanka: Recent EconomicDevelopments and Prospects. World Bank, Wash-ington, DC.

Zaman, H. (1999). Assessing the poverty and vulnera-bility impact of microcredit in Bangladesh: a casestudy of BRAC. Paper commissioned by CGAP asbackground material for the WDR 2000/01, WorldBank, Washington, DC.

microenterprise occupation and poverty reduction in microfinance programs: evidence from sri lanka

Documents