Maize Is Life Maize Research andSmallholder Production ...pdf.usaid.gov/pdf_docs/PNABZ963.pdf · the Office ofSustainable Development / Pro ... MVTS NSCM NSSA OPV RDP SAL I-III USDA

Productive Sector Growth and Environment DivisionOffice of Sustainable DevelopmentBureau for AfricaU. S. Agency for International Development

Maize Research in Africa (MARIA)Project Case Study

Maize Is LifeMaize Research and Smallholder Productionin Malawi

Melinda Smale

July 1993

(Publication services provided byAMEX International, Inc.

··;t41'·1

This report was prepared with support from theU.S. Agency for International Development /Bureau for Africa / Office of Analysis, Re­search, and Technical Support / Division ofFood, Agriculture, and Resources Analysis (nowthe Office of Sustainable Development / Pro­ductive Sector Growth and Environment Divi­sion); in accordance with a contract with theU.S. Department of Agriculture / Office of In­ternational Cooperation and Development.

t•

Contents

Glossary of Acronyms and AbbrcviationsGlossary of Definitions

V

VI

l. Introduction i

Scope and Purpose of Case Study 2Structure of tl)c Report 2

2. Methodology 4

J. The Cultural and Economic Significance of Smallholder Maize Production 5

Tables: 3.1. Maize as Percent of Total Daily Caloric COl/sumption, per Caput, 61961-1977

3.2. Maize as Percent oj Total Area Cultivated hy Smallholders. 1958/59, 61968/69, 1980/81-1990/91

3.3. Maize Value as Percent 0/ Total Value a/Smallholder and All 7Agricultural Output, 1970-1986

4. An Inventory of Varietal Innovation 8

Table: 4. I. Source, Flint Content, Type, and Approximate Release Dates or Maize 9Varieties in Malawi

5. The Impact of Maize Research 10

Farm Family Impact IIHousehold Characteristics ojAdopters and Nonadopters I IManagement Practices ofAdopters and Nonadopters 14Resource Availability and Allocation, Adopters and Nonadopters 15Household Income and Consumption. Adopters and Nonadopters 17Yield and Economic Risks of Hybrid Maize Adoption 18Returns to Labor and Total Factor Productivity, Hvbrids and Local Maize 19

Agricultural Development Division Impact 20Changes in Maize Technology 20An Example ofSmallholder Welfare ~lleets 24

National Impact 26

Figures: 5.1. Malawi Maize Production-Three Yield Scenarios 275.2. Malawi Net Maize Imports-Constant Per Capita Consumption 275.3. Malawi Agricultural GDP-Three Yield Scenarios 27

Map: 1. Agricultural Development Divisions of Malawi 12Tables: 5.1. Relationship of Farm Household Characteristics and Hybrid Maize 13

Adoption

III

5.2. Local Maizc1Yields. Hybrid Maize Adopters and Nonadoplers5.3. Fertilizer U e on Local Maize. Hyhrid Maize Adopters and NOl/adopters5.4. Selected Agi JI10mic Practices. Local and Hl'hrid Maize, .

5.5. Resource Avc/ilahility and Allocation Indicators. Hyhrid Maize Adoptersal/d Nonadopters

5.6 Maize Importence in Household Income al/d Consumption. f{}Jhrid MaizeAdopters and' Nonadopters

5.7. Lahor Returns and Total Factor Productivitl'. Hybrids and Local Maize:. .5.8. Varietal Adoption Characteristics hy Agricultural Development Division, I

1980~/990 I

5.9. Variety as Percent (?lAggregate Maize Area and Aggregate Output. by ,Agricultural bevelopment Division. / 980~8 /

5. / O. Mean Maiz~' Yields hy Variety. by Agricultural Development Division,/980--8/

5. / I. !/Iustrative Wellare Changeslor Hyhrid Maize Adopters in Blantyre.Kasllngu. and Mzuzu Agricultural Development Divisions. 1fJ89--91

6. Factors Affecting the Impact of Maize Research on Varietal Adoption

Varietal ReferencesThe Role of Research

Maize Research Themes and PrioritiesFlint vs. DentHybrids vs. Open-pollinated VarietiesLength of Growing SeasonSoil Fertility

Human Resources, the Organization of Research. and Research CapacityDiscontinuitiesInsti tution-Building

Input Distribution, Marketing, and Price PoliciesSeed SupplvSeed DiflusionMarketing System

Promotional Efforts and Fanner LearningFarming Systems

Figure: 6.I. ADMARC Sales of Improved Maize Seed, /981/82-1989/90 andPreliminary 1990/9/

7. Conclusions

Lessons LearnedWindows of Creativity

References

Appendixes

A. Principal Data SourcesB. Economic and Policy Context of Smallholder Maize Production

IV

[4151616

[7

[921

22

23

24

29

29313132333435353535373738394041

38

42

4244

45

4951

D

t

I

Glossary of Acronyms and Abbreviati1ns

ADDADMARCAESARTSASA

CAROCGIARCIMMYT

DARDAP

EA

FAOFSNMFSR

GDPGOM

.HYV

IITA

LLDP

MADIAMKMOAMVTS

NSCMNSSA

OPV

RDP

SAL I-III

USDA

Agricultural Developmcnt DivisionAgricultural Development and Marketing CorporationAgroeconomic SurveyAdaptive Research TeamsAnnual Survey of Agricultural

Chief Agricultural Research OfficerConsultative Group on International Agricultural ResearchInternational Maize and Wheat Improvement Center

Department of Agricultural ResearchDiammonium phosphate

Enumeration Area

Food and Agriculture OrganizationFood Security and Nutritional MonitoringFarming Systems Research

Gross Domestic ProductGovernment of Malawi

High-Yielding Variety

International Institute for Tropical Agriculture

Lilongwe Land Development Programme

Managing Agricultural Development in Africa (World Bank)Malawi kwacha-US$1 = 4.23 kwacha (as of mid 1993)Ministry of AgricultureMaize Variety and Technology Adoption Survey

National Seed Company of MalawiNational Sample Survey of Agriculture

Open-Pollinated Variety

Rural Development Projects

Structural Adjustment Loans I - III

United States Department of Agriculture

v

Glossary of Definitions

Composites: Improved open-pollinated varieties developed from the genetic combination of awider range of parental material than is usually the case for synthetics.

Hybrids: Improved maize for which pollination is carefully controlled at all stages of theseed production process. For most conventional hybrids, parental matenal consistsof inbred lines for which plants are self-pollinated over several generations. Thecontrolled crosses of two, three, or four of these inbred lines lead to conventionalhybrid seed. Where plants are allowed to randomly pollinate, genetically-causeddeterioration in the yield of hybrid maize is high implying that, for fanners whohave begun using hybrid seed, it is usually economically rational to purchase newseed every year.

OPVs (Open-pollinated varieties): Maize varieties in which seed production is the resultof random pollination of silking plants by pollen from the tassels of other plants.These can include farmers' varieties as well as improved varieties developedthrough synthetic or composite breeding strategies. Genetically-caused deteriora­tion in the yield of OPVs occurs relatively slowly over time implying that, forfarmers who have begun using seed of improved OPVs, it is usually economicallyrational for them to use their own grain as seed for several generations beforereplacing it with higher quality purchased seed.

Synthetics: Improved open-pollinated varieties developed from a breeding and seed-produc­tion procedure similar to the one used to produce hybrid maize, but for which actualseed production takes place under open-pollinated conditions. Resulting varietieshave a relatively narrow genetic base.

VI

}

f

1. Introduction

Maize replaced millets a~d sorghum as thedominant foodgrain crop i~ Malawi only 60 to70 years ago. but over t~ree-quarters of thenation's cultivated area is !now sown to maizeeach cropping season. Per caput, the quantityof maize Malawians consume as a starchy stapleis perhaps the greatest in the world. "Maize isour life (chimanga ndi moyo)." and the ideal ofproducing sufficient maize for the maize por­ridge (nsima) needs of the household "informseveryone's actions and rationales for their ac­tions before, during, and after the maize har­vest.'" Each "hungry season" when their maizestocks have been depleted, many farm house­holds face undernutrition as maize prices riseprohibitively and supplies at local market out­lets fluctuate. Food preferences and the risksassociated with relying on product marketsimply that, in Malawi, farm household deci­sion-making is motivated by the objective ofproducing enough maize to satisfy annual sub­sistence needs.

As staple food requirements to sustain agrowing population increase, diffusion of suit­able higher-yielding varieties has become a foodsecurity imperative.2 [n the short-term, land­saving technological change can only beachieved in Malawi through adoption of seed­fertilizer technology. Soil fertility maintenanceby traditional methods such as fallowing and

IFrom villagers' statements, cited in Peters. 1988.

21n post-World War II Nyasaland, Kettlewell al­ready expressed concern over declining productivity asIncreasing population pressure on the land tended to­wards maize monocropping. Even before independencehe described parts of the Southern Region of Malawi ashaving high population densities and large numbers offarm households with less than a hectare of land(Kettlewell, 1965).

rotation has become increasingly difficult asfarmers expand their maize area and monocropin an attempt to secure family grain require­ments in the face of chronically low maizeyields. Releasing land for the cultivation ofother food crops that are essential to improvingnutritional standards and for production of ex­port crops that earn valuable foreign exchangecannot be accomplished without improvingmaize yields. J

Malawi has a labor-land ratio that is highby African standards (Binswanger and Pingali,1988) and agroclimatic conditions that are fa­vorable for a seed-fertilizer transfonnation.Malawi's maize research program has releasedhybrids, synthetics, and composites for over 30years but, until the late 1980s, no more thanabout 10 percent of aggregate maize area wassown to hybrids or first-year open-pollinatedvarieties. Recent data (C[MMYT/MOA; FSNM)demonstrate that, especially in higher-poten­tial, maize-producing zones, the percentage offanners sowing hybrid maize has grown rap­idly in the past four seasons. Aggregate area inhybrids has remained fairly low because, evenwhen farmers have adopted hybrid maize, theycontinue to devote a large proportion of theirhousehold's maize area to local varieties.

Certain consumption preferences ofMalawian fanners, among other features of inputsupply and distribution, have been frequently

'Similarly, Kettlewell wrote that conditions inMalawi demonstrated the vulnerability of a food sup­ply so dependent on a single crop, and the importanceof reducing the proportion of land under maize tofacilitate crop rotation and production of other crops.At that time as now, "the first part of the problem ofagricultural progress was to raise the yield of maize perunit area throughout the country" (p. 258).

cited as factors limiting the popularity of hy­brid varieties. Malawians reveal a distinct con­sumption preference for the flinty varietiesloosely categorized as "local," or "maize of theancestors (chimanga cha /lwko!o)." These vari­cties are more efficiently processed into thefine white flour (li/a !Vovera) used to preparethe preferred type of porridge, and their hardgrain is more resistant to weevil attaqk in stor­

age than most of the denty, white hyhrids thathave been introduced in the past. For this rea­son, hybrid maize was until recently,promotedas a cash crop, although some substitution ofhybrid maize for local varieties in consumptionis increasingly perceptible and is unavoidablefor the food-deficit households who representthe majority in Malawi, In recognition of theimportance of consumer preferences insmallholder adoption decisions, the Departmentof Agricultural Research (DAR) has periodi­cally released semiflint OPVs. For the 1991-92season, DAR also released two new semiflinthybrids and promotional efforts are emphasiz­ing improved processing and storability traits.Evidence suggests that the new semi flint hy­brids perform well relative to both denty hy­brids and local maize in terms of yield, pro­cessing, and storage characteristics (Smale eta!., 1993; Jones and Heisey).

As a case study, the history of maize re­search in Malawi is of policy interest for twoprincipal reasons. First, although various fac­tors suggest that the agroeconomic setting isfavorable for HYV adoption, farmer adoptionrates have risen very slowly. Understandingadoption patterns in Malawi has implicationsfor other maize-producing and -consumingzones. Second, although the significance offlintmaize preferences in household decision-mak­ing has long been recognized by the breedingprogram, a perceptible tension appears to haveexisted historically between the recognition ofgrain quality as a trait and the importance ofyield criteria. For farmers who grow improvedvarieties as a cash crop, processing and storageefficiency is of no significance and yield at

2

harvest is critical. Maize-deficit farmers whowant to consume their maize are concernedabout yield from the mortar. Flintiness andyield criteria have also been related to the issue,spurred by donor involvement, of whether hy­brids or OPVs should be emphasized.

Consumer preference for flinty maize maybe a relevant issue for breeding programs in thebroader region encompassing the maize-pro­ducing zones of Tanzania, Zambia, and per­haps Mozambique. In all of these zones, maizeproduction by smallholders IS a critical foodsecurity issue, and similar preferences are likelyto affect adoption decisions.

Scope and Purpose of Case Study

The objectives of the Malawi case study are to(I) provide a profile of the more significanthistorical changes in farmers' varietal choice(research impact as expressed by adoption rates);(2) summarize associated changes in aggregatemaize production figures and related welfarestatistics and suggest how these figures mighthave evolved under different technical changescenarios; and (3) document major factors inresearch, extension, and promotion of theseinnovations that may have influenced the speedand breadth of technological change. The par­ticular emphases of the Malawi case study arethe rate of varietal change and the role of con­sumption preferences in both farmer deeision­making and the breeding program.

Structure of the Report

The report begins by presenting evidence toillustrate the central importance of maize in thesocioeconomy of Malawi. 4 The following sec­tion presents a time chart and outline of maizevarietal releases and maize research activity. [n

4Appendlx B presents a review of secondary litera­ture about aspects of the nation's approach to develop­ment policy that havc affcctcd smallholder maize pro­duction.

I

Section V, indicators of the effe

t·ts of varietal

adoption on farm families, maj r maize-pro­ducing zones, and selected m. croeconomicvariables are presented and interpreted. SectionVI summarizes some of the major factors that

3

have influenced the impact of maize researchin Malawi, and the concluding section suggestssome of the speci fie lessons that can be drawnfrom that experience.

I

2. Methpdology

Both primary and secondary sources providethe information and data ror the study. A re­view of secondary literature provides the back­ground and context for maize research impactsection (Appendix B). A combination of time­series data from the Annual Survey of Agricul­ture (ASA), recent findings from the MaizeVariety and Technology Adoption Survey(MVTS), and selected figures from the FoodSecurity and Nutrition Monitoring Survey Re­ports are used to generate the adoption rate

lThe major data sources arc described in AppendixA. The data from the surveys are broadly representativeof the major maize-producing zones of Malawi.

4

statistics, characterize the impact of technicalchange on farm families and Agricultural De­velopment Divisions, and identify some of thefactors that have affected adoption rates (objec­tive I).5 Secondary data are used to developillustrative scenarios that express the effects ofchanges in maize seed technology on maizeproduction, national food security, and agricul­tural GDP (objective 2). Interviews with keyactors in the research system form the basis ofobservations relevant to objective 3.

3. The Cultural and Economic Significanceof Smallholder Maize Production

Among Malawians, nsima (stiff maize porridge)and food are synonymous.c, Based on her inter­views in Zomba, Peters reports that, for mosthouseholds, to cultivate without cultivatingmaize was an "impossible thought." To haveland is to grow maize, and growing maize is away of life (1989, p. 48). The CIMMYT/MOAdata from Mzuzu, Kasungu, and Blantyre Ag­ricultural Development Divisions confirm thatall survey households grew maize and, in1989-90 and 1990-91 cropping seasons, de­voted from 75 percent (Kasungu) to 98 percent(Blantyre) of their individual farm area tomaize. 7

The FAO Food Balance Sheets show al­most a negligible change in maize as a percentof per caput daily caloric consumption from thelate colonial period through the 1970s (Table3.1). A net increase in total and maize caloriesis distinguishable over time. Although compa­rable figures have not been compiled for the1980s, per caput utilization of maize as food inMalawi still ranked among the highest in the

60nly in somc zones around the lakeshore wouldcassava or rice be considered more important as astarchy staple.

7The Report Oil an Economic Survey of" Nyasaland

1958-59 reportcd that, on average, nearly 100 percentof an individual cultivator's land was sown to maize inthe Southern Region, as compared with about 55 per­cent in the Central Region. As a point of comparisonwith today's cstimates, Blantyre is located in the South­ern Region, and Kasungu in the Central Region.

tMaize is bclieved to have gained ItS oomlflantposition in the Malawian diet between 1850 and 1910(Agroeconomic Survey, 1982). Williamson (1956) notesthat although Lacerda (1790s) and Livingstone (1850s)rcfer to maize in their writings, both recount that thestaple crops were millets, such as fingcr millet, andsorghum.

:;

world (CIMMYT, 1990).KIn a nation in which agriculture constitutes

an estimated 40 percent of GDP (Pryor, 1988),provides an estimated 45-50 percent of wageemployment in the "modern sector" (Kydd andChristiansen, 1982; Pryor, 1988), and employsover 80 percent of the total labor force (Gulhati,1989), maize is by far the dominant crop interms of hectarage for all 8 Agricultural Devel­opment Divisions (ADDs) except Ngabu. Esti­mates of maize area as a percent of total culti­vated area are shown in Table 3.2 for the 5major maize-producing ADDs of Malawi, overtime. Q Differences among zones reflect alterna­tive cash crop opportunities (tobacco andgroundnuts, particularly in the Central Region),larger land areas to diversify food crops (mil­let, cassava, sweet potatoes, and beans, espe­cially in the Northern Region), and subsistencerequirements/land ratios (highest in the South).Considering sampling errors, differences overtime are not evident over the brief period re­ported in the 1980s, but are likely to haveemerged gradually between the late colonialperiod and present. At the national level, mostestimates now predict that from 75 percent to85 percent of total area cultivated bysmallholders is sown to maize.

Finally, in an economy in which the value

"Here, area is defined in terms of primary crop. Inthe NSSA and ASA, maizc IS only rarely recorded as asecondary intercrop. Counting the intercroppcd maizcarea in tcrms of thc secondary erop would increascother crop arca (especially for pulses and beans) as apercent of total cultivated arca. The signi ficance ofintercropping may have declined over time, and ap­pears greatest today in specific areas of high populationdensity in the South, Dedza, and Ntchcu in LilongweADD, and in Rumphi, Mzuzu ADD.

Table 3.1 Maize as Percent of Total Daily Caloric Consumption,Per Caput, 1961-1977

---Per Caput Daily Calories--- Maize

Year Maize Total Percent of Total

1961-65 (aver.) 1395 2092 671967 1379 2038 681968 1428 2149 661969 1500 2208 681970 1552 2313 671971 1530 2358 651972 1533 2336 661973 1520 2331 651974 1525 2331 661975 1417 2201 641976 1467 2265 651977 1448 2215 65

Source: Food and Agricultural Organization of the United Nations, Food Balance Sheets 1967 -1977,Rome, 1980; International Maize and Wheat Improvement Center (CIMMYT) 1989/90 World Maize

Facts and Trends, Mexico, D.F. 1990.

Table 3.2 Maize as Percent of Total Area Cultivated by Smallholders,1958/59, 1968/69, 1980/81 -1990/91

-------------Agricultural Development Division" ------------- All

Year Blantyre Liwonde Lilongwe Kasungu Mzuzu Malawi

1958/59 66-751968/69 88 89 81 70 771980/81 89 88 76 67 751985/86 91 83 76 731986/87 89 80 66 691987/88 93 82 66 761988/89 82 67 771989/90 96 82 831990/91 98 70 81 75-85

'Figures for Mzuzu exclude Nkhata Bay, a cassava-producing zone, and for the last two years in Blantyre

and Kasungu they exclude portions of the ADDS that are considered to be less representative with

respect to maize production.

Source: Report on an Economic Survey of Nyasaland 1958-59, Federation of Rhodesia and Nyasaland.Ministry of Economic Affairs; National Sample Survey of Agriculture, 1968-69 and 1980-81, NationalStatistical Office, Government of Malawi; Annual Survey of Agriculture 1985-1989, Ministry ofAgriculture, Government of Malawi; CIMMYT/MOA Maize Technology and Varietal Adoption Survey,1989-91.

6

1

Table 3.3 Maize Value as Percent of Total Value of Smallholder andAll Agricultural Output. 1970-1986

(million 1978 MK)

·· __ ··---···Real Value of Output·····_-_···· Maize % of Output ValueYear Smallholder Estate Total Maize Smallholder Estate

,1970 150 41, 191 59 40 311971 175 44 219 76 43 401972 184 50 234 80 43 341973 163 59 222 72 44 331974 169 59 228 81 48 361975 156 76 232 74 47 321976 186 84 269 74 40 271977 21 1 99 310 78 37 251978 217 102 319 84 39 261979 223 119 342 112 50 331980 202 130 332 82 40 251981 188 128 316 75 40 271982 189 151 340 121 64 361983 195 160 356 105 54 301984 210 151 352 104 50 301985 213 89 421986 213 75 35

Source: Smallholder and estate output estimates and GDP deflator from Pryor, F.L., Income Distribution

and Economic Development in Malawi: Some Historical Statistics. World Bank Discussion Papers 36.Washington. D.C., 1988; Nominal maize producer prices from Gulhati, R., Malawi: Promising Reforms,Bad Luck, World Bank, Washington, D.C., 1989; Maize production figures from Food and Agricultural

Organization of the U. N.

of estate production is considered to be theprimary generator of foreign exchange earn­ings and national revenues, maize value as apercent of total value of smallholder and allagricultural output is high--especially for sucha low-priced commodity. The estimates shownin Table 3.3 indicate that the percent of maizevalue in smallholder output ranged from 35percent to over 60 percent from 1970 to 1986and is only slightly lower as a percent of thevalue of all agricultural output, including thatof estates.

'OFerguson et a!. (199()) argue that "in the absenceof agricultural inputs that stabilize production, declines

in crop diversity may result In greater nutritional risk atleast for certain strata of farmers" (p. 276).

7

The nutritional implications of the domi­nance of maize calories in the dieeo and soilfertility implications of monocropped maizehave been cause for policy concern. Releasingland for the production of other foodcrops andpotential cash crops is now a priority for nutri­tional, soil fertility, and income reasons-at anational and a farm household level. However,the fact that the majority of farm householdsare maize deficit may mean that the most thatcan be accomplished through improving maizeyields in the short to medium term is to closethe household food deficits, reduce the threat offood imports, and slow the expansion of maizearea.

4. An Inventory of Varietal Innovations

Until the early 20th century, the maizes grownin the region were smaIl-statured, flint typeswith shorter season materials grown on alluvialsoils and swampy areas during the dry seasonand longer season varieties cultivated as part ofthe rainfed crop complex (Blackie, (989). Ac­cording to Miracle's (1966) historical descrip­tion of maize in tropical Africa, flinty varietieswere, until recently, more popular than denlyvarieties because of lower susceptibility to in­sect infestation.

Although dent hybrids cover almost allmaize area in Zimbabwe and over two-thirds ofmaize area in Kenya, flinty local varieties stillpredominate in southern Tanzania, eastern Zam­bia, and Malawi. The trade-off betweendentiness and flintiness is related in part toyield and in part to the supply of germplasm.To this date, suitable breeding material for denthybrids is much easier to locate and the result­ing hybrids that have been produced have supe­rior yields. II

An approximate time chart and descriptionof varieties released by Malawi's maize breed­ing program is shown in Table 4.1. 12 Perusal ofthe maize types released and imported over the

II An example is the difference in yiclds reportedby Bolton (1974) for LH II and SR52. Part of the yielddi fference between LH 11 and SR52 is due to the factthat LH II was a double-cross and SR52 is a single­cross hybrid. Breeders may have also been undulyinf1ucnced by North American studies that show aslightly higher yield potential (approx. 5 percent) fordent maize.

I~Sources include Kydd (1989), Bolton (1974),Kettlewell (1965), Darrah and Penny (1974), and oralinterviews with B.T. Zambezi, Senior Maize Breederand Maize Commodity Team Leader. The informationis approximate.

8

years illustrates how the emphasis of the breed­Ing program has oscillated between, but has~lways included, both hybrids and OPVs. Re­lated to maize type is the issue of dent and flint'characteristics.

Brown (1963) writes that prior to indepen­dence, very little mass-selection had been doneby farmers. In the 1940s, S. Hoyle began col­lecting, identifying landraces and inbreedinglocal material to produce pure lines. In 1954,R.T. Ellis initiated a breeding program whichled to the development of a number of hybridsand several synthetics. Among these, the bestcultivars (subsequently released) were consid­ered to be LH II, a semiflint hybrid, and threesemiflint synthetics (SV28, SVI 7, and SV37).Each synthetic was bred for a distinct regionand agroclimatic zone.

Breeding activity appears to have ceasedalmost completely during the early 1960s withthe transfer of power. Breeding lines deterio­rated and, in 1967, the hybrid program wasofficially discontinued. When the new breeder,Bolton, arrived from Tanzania in 1970, theemphasis of the program shifted from synthet­ics to composites. Bolton and other officialsbrought materials from Kenya, Tanzania, andZimbabwe. The program released the compos­ites UCA ("Malawianized" from Tanzanianmaterial) and CCA (developed primarily fromlocal materials), but imported the hybrids SR52(Zimbabwean), RlOO (Zimbabwean), and H632(Kenyan).

In 1977 the hybrid program was restored.In response to the popularity of SR52 in othercountries of the region and among estates inMalawi, and to reduce exorbitant import costs,program researchers began screening Zimba­bwean and South African hybrids with the ob-

I

Table 4.1 Source, Flint conteni Type. and Approximate Release Datesof Maize arieties in Malawi

Year Name Flint Type SourceContent

1959 SV17 semiflint synthetic Bred locally1961 LH 11 semiflint hybrid Malawi-Mexican-Zimbabwean inbred

lines, bred locally1966 SV28 semiflint synthetic East African Agriculture and Forestry

Research Organization inbred lines, bredlocally

1967 SV37 semiflint synthetic Bred locallySR52 dent hybrid Imported, Zimbabwean-bredR200 dent hybrid Imported, Zimbabwean-bred

1971 UCA semiflint composite Tanzania-bred (adapted)CCA semiflint composite Local materials/some exotic, bred locally

HG32 semident hybrid Imported, Kenya-bred (LLOP only)R201 dent hybrid Imported, Zimbabwean-bred

1978 MH12 dent hybrid (Adapted) Zambian SR521983 NSCM41 dent hybrid Ciba-Geigy 4141, import license for Fl1984 MH14-16 dent hybrid RSA-Zimbabwean germplasm, bred

locallyCCC semiflint composite Local. C1MMYT, RSA and Zimbabwean

material, bred locallyCCO semiflint composite liTA/CIMMYT/RSA material, bred locallyR215 dent hybrid Imported, Zimbabwean-bred

Tuxpeno composite Imported, CIMMYTKalahari composite Imported, RSA-bred

EarlyPearl

1990 MH17 semiflint hybrid MH 12 and CIMMYT material, bredlocally

MH18 semiflint hybrid MH 16 and CIMMYT material, bredlocally

jective of selecting high-yielding hybrids toreplace SR52. Their work culminated in therelease ofMH 12 and the development ofbreed­ing lines for MH 14-MH 16. That decision marksa change in emphasis to breeding for yield withdent hybrids, although the importance of flinti­ness was still recognized in continued workwith composites. In the mid- and late 1980s,

length ofgrowing season emerged as a researchtheme. The semiflint, shorter season composite

CCD and shorter season denty hybrid MH16were released. The dent hybrids NSCM41,

9

RlO 1, and Rl15 were also imported. 13 Only inthe past few years were the themes combined inthe rapid development and release of two newsemi flint hybrids, MH 17 and MH 18, one ofwhich (MH 18) has a relatively shorter growingseason.

IlThere is perpetual debate about the flintIness ofthese varieties. Although some farmers report that theyprocess better than other dent hybrids, the breedersalways rank them at the same point as other hybrids onthe flint-dent spectrum.

5. The Impact of Mai1ze Research

)

JLIst before independence, Brown (1963) de­scribed maize as the staple food, called it asubsistence crop, and reported that it was grownon ridges laid out in contours and cultivated byhoc. Based on the estimates reported inKettlewell (1965), about 0.6 million hectareswere cultivated in maize. Kettlewell's yieldestimates were 0.7 t/ha for the North, 1.1 t/hain the Central Region, and 0.9 t/ha in the South.Weighted by area, Kettlewell's figures suggestan average unfertilized maize yield of 0.9 t/ha.Assuming no exports and using a figure of I. Itons/annum for the maize subsistence require­ments ofa family of four, Brown estimated thata maintained national yield of 3 to 4 t/ha was areasonable research objective to assure maizeself-sufficiency.

Today, hand-hoe cultivation with ridging isstill the dominant form of land preparationexcept in some areas of the North where farmsizes are relatively large and labor require­ments correspondingly burdensome (ASA andCIMMYT/MOA). About twice as many hect­ares are sown each year to maize, and the popu­lation has more than tripled. To meet the samemaize subsistence requirements on the samemaize area reported by Brown, average maizeyields would need to be nearly 7 t/ha. Givencurrent estimates of minimum per caput maizerequirements (230 kgs), average maize yieldsof 3 tlha would meet the same objectives whilepermitting land to be diverted for the produc­tion of other food and cash crops. Data onnational yields show a positive trend, but onlywhen taken over a long time period (FAO andUSDA).

A comparison of similar figures and pointestimates has suggested to some that little haschanged since the breeding program began

10

(Kydd, 19X9). On the contrary-given a rap­idly growing population and the dominance ofmaize in the diet, the impact of maize researchand graqiual HYV adoption has clearly been tocounteract the yield-reducing effects of declin­ing soill fertility and expansion into marginallands, enabling the nation to continue to be, inmost years, self-sufficient in maize. Nationalmaize yields, at approximately 1.3 tlha, areprobably far short of research goals, but na­tional hybrid maize yields (of smallholders) areroughly 3 t/ha. In the high-potential adoptionzones such as parts of Mzimba District, theKasungu, and Lilongwe plains, average maizeyields have, in fact, increased. Today's averagemaize yields for the major maize-producingzones would be closer to 1.3 tlha for the North,1.5 tlha in Central Region, and 1.0 tlha in theSouth. 14 [n contrast to the point estimates com­puted for the colonial period, average yields forunfertilized local maize appear to have prob­ably declined to approximately 0.7 tlha(C[MMYT/MOA).

As in other HYV adoption settings (Blackie,[990), initially the main beneficiaries of maizeresearch have probably been larger producersand smallholders on better quality land. Thesefarmer subgroups enjoy management or physi­cal resource advantages and are better posi­tioned to bear the economic risk of using creditand experimenting with new varieties. Recordedin case studies during the 1970s and 1980s, thelarger farmer bias may have been especiallytrue in the early years of the Lilongwe researchand extension program and in various pilotefforts (Roberts, 1972; Chipande, 1987; Hansen,

14Bascd on National Crop Estimates for the late1980s.

\

1986; Anderson, 1975). Based on the seed salesestimates reported by Quinten and Sterkenburgand FAO area figures, 0.3 percent of aggregatemaize area in Malawi was sown to hybrids in1970/71, of which, about two-thirds of the areawas located in Liiongwe District and most ofthe seed was imported. The same sources pro­vide estimates of 2.3 percent ofaggregate maizearea under first-, second-, or third-year synthet­ics with two-thirds of the area again in LilongweDistrict. A Lilongwe Land DevelopmentProgramme (LLDP) survey in 1971 showed 53percent of farmers growing synthetics and 17percent sowing hybrids." One source cited byKydd (Schulten and Westwood, 1972) estimatedas much as 8 percent of total cropped area wasplanted to improved varieties (mostly synthet­ics) in the early 1970s. At perhaps the peak ofcomposite maize diffusion, the 1980/81 NSSAshows a fairly significant percentage ofsmallholders growing composite varieties inMzuzu, Kasungu, and Salima ADDs.

Over time, as in Zimbabwe and Kenya (al­though much more slowly), a larger subset offarmers have been able to adopt recommendedvarieties on part of their maize area, with andwithout credit or project schemes. The upwardslope in the aggregate diffusion curve for hy­brids and broadening in the cross-section ofadopters is especially evident in the last threeyears (Smale et aI., 1992; FSNM). Economet­ric results, descriptive statistics, and secondarysources suggest that multiple factors are associ­ated with the propensity to adopt. Farmer learn­ing, greater willingness to substitute denty hy­brids for local maize in consumption, shortergrowing season and lesser yield risk with hy­brids, as well as institutional factors such asimproved seed distribution and greater flexibil­ity in package diffusion, have undoubtedly con-

';An interesting item reported in the survey reportwas that when hybrid maize growers were asked whythey continued to grow local or synthetic varieties, oneof their responses was a "stated loyalty" to a certainacreage of local and synthetic varieties.

II

tributed to growing adoption rates in the lastfew years (Smale, 1992). The effects of thelatest research breakthrough-the release oftwonew semiflint hybrids-are now being assessed(Smale et aI., 1993; .Iones and Heisey).

The followi ng subsections use theCIMMYT/MOA and ASA survey data to por­tray some of the likely effects of hybrid maizeadoption on farm households, between ADDs,and on various national economic indicators.Survey zones and ADDs are shown in Map I.The narrative focusses on hybrids for severalreasons. First, there are obvious difficulties inmeasuring the amount of improved OPV mate­rial in farmers' fields and associated techno­logical impact. Second, seed sales, secondarysources, and the two survey data sources sug­gest that the area under first-year OPVs hasbeen declining over the past decade, while thearea under hybrids has increased. Third, forbetter or for worse, today's emphasis of themaize research program, largely enforced bydonors, is hybrid development. However, anyhypothesized research impact is recognizablyunderstated if it excludes synthetics and com­posites. For example, first-year composites, ascompared to hybrids, continue to be grown onabout 10 percent of maize area in Salima ADD,a primarily lakeshore environment. The impactof all composite material in Salima is ofcoursegreater than that figure indicates.

Farm Family Impact

Household Characteristics ofAdopters andNonadopters

[n the [989/90 season, all but a few surveyfarmers grew local maize. [n Kasungu andMzuzu survey zones, over a third of farmersalso grew hybrid varieties, but in Blantyre only14 percent sowed hybrid maize. Even whenfarmers planted hybrids they continued to de­vote the major portion of their maize area tolocal maize. Both adopters and nonadopters

Map 1. Agricultural DevelopmentDivisions of Malawi

preferred to consume local maize, althoughsome substituted their own or purchased hybridmaize during maize-deficit seasons.

Those who adopt are more likely to bemale, members of credit clubs, and to operatelarger areas (Table 5.1). Wealth influences op­portunities for adoption, and credit relaxes ex­penditure constraints, facilitating adoption-ifonly for a season. The larger the land area, themore likely is the household to qualify forcredit or to have alternative crops that generatecash income. Female heads of households whoare divorced or widowed tend to be less wealthyand are less likely to be club members-andtherefore have fewer opportunities to adopt.The primary diffusion mechanism for the seed­fertilizer technology package has been the for­mal credit system, which has favored jointhouseholds and larger farms. This interrelatedcluster offactofs, which often translates looselyinto "control over resources," is associated withthe probability of adoption but disguises diver­sity in the adopter population.

12

The figures also demonstrate that female­headed households, noncredit club members,and smaller farmers do adopt Certain culturaltraditions imply de jure female-headed house­holds are more prevalent in the South wherefarm sizes are also smaller. 1(, Women in thatregion generally have matrilineal rights to land,but small farm size has constrained their choices.Given consumption preferences and, until re­cently, recommendations for growing hybridsin pure stands where intercropping is moreprevalent, many women probably didn't feelthey had "enough land" to grow hybrid maize(Hirschmann and Vaughan, (983). Maize isclearly a woman's crop to the extent that it is afood crop, but in any region, all members of thehousehold, when present, work in the maizefields. Hybrid maize purchased on credit maybe more of a "men's crop" in the North, forexample, where cultural traditions are alsopatrilineal. In no sense, however, is the conceptof "women's crops" and "men's crops" par­ticularly useful in the analysis of hybrid maizeadoption in Malawi. 17 Producing sufficientmaize is the common objective of every indi­vidual in any Malawian household.

Anecdotal evidence from the 1989/90CIMMYT/MOA survey illustrates what differ­ences in farm size among adopters implies.Farm size is related to farming systems andfanner objectives, and not just to credit eligi­bility. One of the subsets of hybrid maize grow­ers was found in Thyolo. These farmers grewshort-season hybrid maize on tiny plots to con­sume or sell green in Blantyre city for supple­mentary food or cash during the hungry season.They also worked off the farm to meet their

16Peters' distinction between de {acta (male absentand engaged in employment activities) and de )lIre

(divorcee or widow) female-headed households is ap­propriate in explaining different probabilities of adop­tion.

17Even in Zomba, Peters concludes that there areno gender-linked crops in the way the concept is usedto describe West African systems (1989).

\

Table 5.1 Relationship of Farm Household Characteristics andHybrid Maize Adoption

ADOPTION CHARACTERISTIC

HouseholdCharacteristiclSubgroup

Sex of Household Head'

femalemale

Credit Club Membership'

yesno

Farm Size Class'

less than 0.7 haO.7to1.5hamore than 1.5 ha

Local Maize Subsistence Ratio'

less than 11 or above

Percent of SubgroupSowing Hybrid Maize

1738

7617

133656

3340

Mean Percent of MaizeArea Sown in HybridMaize by Adopters

3943

4440

444437

3048 ..

* statistically significant differences between subgroups (5 percent), chi-square test.* * statistically significant differences between subgroups (5 percent) t-test.+ actual local maize output/minimum stated maize subsistence requirements.

Source: Maize Variety and Technology Adoption Survey, CIMMYT/MOA. 1989-90.N == 420 farmers in Blantyre. Mzuzu. and Kasungu Agricultural Development Divisions.

local maize consumption needs and to buy theirinputs. By contrast, some of the hybrid maizegrowers in the Kasungu and Mzuzu areas soldover 2 tons of hybrid maize in the previousyear, producing 3 to 4 tlha yields by applyinghigh analysis fertilizer and using animal draftpower for land preparation. These fanners alsohad enough land to produce large outputs oflocal maize, satisfying their consumption re­quirements at the same time that they earnedprofits from their hybrid maize. Both sets offarms may have grown hybrid maize for differ­ent economic reasons.

13

Noncredit club members also adopt. In 1989/90, hybrid maize adopters in the Blantyre sur­vey zone were more likely to be self-financedand to have first learned about improved seedfrom other farmers rather than extension agents.The fact that, in the past, credit packages haveconsisted of seed and fertilizer in fixed quanti­ties also means that land allocated to hybridmaize by credit users has exhibited a lumpinessaround O.4-ha (I-acre) intervals. For hybridmaize growers who are not credit club mem­bers, there is greater variation in hybrid maizehectarage.

Table 5.2 Local Maize Yields, Hybrid MaizeAdopters and Nonadopters

(kilograms per hectare) .

r

Characteristic

Mean observed maize yields,objective yield estimates

Unfertilized localFertilized localAll maize

Mean expected maize yields,farmers' estimates

Unfertilized localFertilized local

Adopters

832'1,351'1,806'

853'1,536'

Subgroup

Nonadopters

720'1,184'881 '

704'1,31 T

'Statistically significant differences between subgroups (5 percent). t-test.

N = 420 farmers in Blantyre, Mzuzu, and Kasungu Agricultural Development Divisions.

Source: Maize Variety and Technology Adoption Survey. CIMMYT/MOA 1989-90_

The CIMMYT/MOA data also confirm thatin Malawi, adoption patterns, sex of householdhead, farm size, and credit club membershipvary by zone. In aggregated figures, differ­ences in these variables as they relate to adop­tion are to a large extent differences associatedwith agroeconomic zone. Within zones, differ­ences are less evident. For example, within theBlantyre survey zone female-headed householdswere no less likely to adopt than male-headedhouseholds while in the Mzuzu zone, they were.Similarly, although pronounced among theKasungu and Mzuzu survey farmers, differ­ences in the likelihood of adoption betweenfarm size classes were not signi ficant amongthe Blantyre survey farmers.

Finally, sex of household head, credit clubmembership, and farm size may affect prob­abilities ofadoption, but are less likely to influ­ence the proportion ofmaize area adopters plantin hybrids. The household characteristic that ismore likely to affect land allocation to varietiesby adopters is the ratio of local maize subs is-

14

tence requirements to the local maize outputtheir land can produce.

Management Practices ofAdopters andNonadopters

Adopters in the survey zones both obtained andbelieved they could obtain higher yields fromtheir local maize (Table 5.2). Partial explana­tion for this finding is provided by evidencethat adopters were more likely to apply fertil­izer to their local maize and, when they used it,they applied a higher rate of Nlha (Table 5.3).Often fanners reallocate some of the fertilizerreceived on credit as part of a hybrid maize ortobacco package to their local maize, but inrecent years fertilizer has been available Oncredit specifically for local maize and someclub members purchase additional fertilizer withcash.

Fertilizer application does not explain all ofthe difference between actual and observed localmaize yields for adopters and nonadopters,

\

Table 5.3

Characteristic

Fertilizer Use onAdopters and

I

Local ~aize,Nonadppters

Hybrid Maize

Subgroup

Percent of farmers applyingfertilizer to focal maize"

Mean kgs N per hectare, farmersapplying fertilizer to local maize

Adopt!'!rs79'

49'

Nonadopters39

32'

. Statistically significant differences between subgroups (5 bercent), t-test.

., Statistically significant differences between subgroups 15 percent!. chi-square test.

N =420 farmers in Blantyre, Mzuzu and Kasungu Agricultural Development Divisions.

Source: Maize Variety and Technology Adoption Survey. CIfv1MYTIMOA, 1989-90.

however. The fact that unfertilized local maizeyields differ between the groups suggests thatother management or human capital variablesmay playa role.

Between varieties, as expected, farmersdevote more labor to land preparation for hy­brid maize because they more frequently plantit on fallowed land (Table 5.4). Although hy­brid maize tends to be planted later, more timeis required in planting because of greater plant­ing densities and, according to many surveyfarmers, because "greater care is needed tofollow recommendations." More hybrid areathan local maize area is also weeded twice.

Resource Availability and Allocation,Adopters and Nonadopters

Adopters tend to have both larger total areasand larger areas in other crops (Table 5.5).Although, on the average, maize as a percent ofhousehold cultivated area differs statisticallybetween adopters and nonadopters (because ofsmall standard errors), the difference is hardlymeaningful. Even after farmers have adoptedhybrid maize, they continue to sow a largeportion of total cultivated area in maize bothbecause of the dominance in the diet and the

15

economics of the cropping system. [n general,hybrid maize area substitutes for local maizearea rather than releasing land for cultivation ofother crops. Per hectare net returns are prob­ably higher in most years for hybrid maize thanfor many of the alternative crops smallholderscan grow (groundnuts, beans, cassava, sweetpotato). In Mzimba District of the Mzuzu zone,hybrid maize is a cash crop. Among the surveyzones, perhaps the greatest reallocation of farm­ers' area is found among Kasungu farmers whohave the opportunity to grow highly remunera­tive tobacco. Kasungu farmers were also morcwilling to consume their own hybrid maize.

The fanns of adopters also have greatercarrying capacity (hectares per adult over 12years ofage) to support the starchy staple needsof the family. The very slow decrease in thepercent of fann area sown to maize as thelabor/land ratio rises underscores the impor­tance in farm household objectives of attempt­ing to satisfy maize subsistence requirements. IX

Controlling for farm size and labor capacitydocs not diminish the most salient feature of

ISThe significance of the subsistence constraint III

farm household decision-making is supported by sev­

eral types of econometric results (Smale, 1992).

Table 5.4, Selected Agronomic Practices, Local an Hybrid Maize

r

Characteristic

Land Preparation

Percent of plots sown after fallow"Percent of plots ridged by ridger"

Planting

Percent of plots planted after Dec. 15"Mean plant density (1 OOO/ha)

Weeding

Percent of aggregate area weeded twice

Intercropping

Percent of aggregate area intercropped

Maize Variety

Local Hybrid

4 109 15

17 2731' 35'

63 76

16 13

'Statistically significant differences between subgroups (5 percent), t-test ... Statistically significant differences between subgroups (5 percent), chi-square test.N =420 farmers in Blantyre, Mzuzu, and Kasungu Agricultural Development Divisions.

Source: Maize Variety and Technology Adoption Survey, CIMMYT/MOA, 1989-90.

Table 5.5 Resource Availability and Allocation Indicators,Hybrid Maize Adopters and Non-Adopters

Characteristic

Mean farm size (ha)Maize areaArea in other crops +

Mean hectares/adult « 12 yrs)

Mean percent of cultivated area in maize

Adopters

1.68'1.42'.26'

.60'

86'

Subgroup

Non-Adopters

1.07'0.92'0.14·

041'

90·

Hectares/adult class

> 0.25.25 to .39040 to .59< .60

Mean annual earnings from off-farm labor (MK)

95868585

136

Percent area in maize

95928683

143

• Defined by primary crop. Statistically significant differences between subgroups (5 percent), t-test.N = 420 farmers in Blantyre, Mzuzu, and Kasungu Agricultural Development Divisions.

Source: Maize Variety and Technology Adoption Survey, CIMMYT/MOA, 1989-90.

16

Table 5.6 Maize Importance in Household Income and Consumption,Hybrid Maize Adopters and Non-Adopters

Characteristic

Mean value of crop output (MKIMaize outputOther primary cropsOther crops interplanted with maize

Maize as percent of value of primary crops(MK)

Maize as percent of value of annual income(MKj'

Mean maize output (kgsl per adult « 1 2years)

Mean minimum annual maize subsistence

requirements (kgsl

Mean minimum annual maize subsistencerequirements (kgsl per hectare

Adopters

899'743'127'29"

86

65'

942'

1,067'

947'

Subgroup

Non-Adopters

363'257'62'44"

87

49'

314'

982'

1,394'

+ Annual income defined as sum of off-farm wage and in-kind earnings, livestock sales, value of primaryand interplanted crops produced, remittances and value of maize stocks at planting.

. Statistically significant differences between subgroups (5 percent), t-test .

.. Statistically significant differences between subgroups (5 percent), t-test.N = 420 farmers in Blantyre, Mzuzu, and Kasungu Agricultural Development Divisions.

Source: Maize Variety and Technology Adoption Survey. CIMMYT/MOA. 1989-90.

fanning systems in Malawi (Table 5.5).When cultivated following recommenda­

tions and even when adapted to most farmers'conditions, farmers use more labor per hectarefor hybrid maize than for local maize varietiesas they are typically grown. 19 On the average,however, adopters do not appear to reallocatelabor from off-farm to farm activities but withinfarm activities (Table 5.5).

I"There is no particular rcason why high levels ofmanagement could not be applied to fertilized localmaize as well!

17

Household Income and Consumption,Adopters and Nonadopters

The mean value of total crop output for adopt­ing households is 2.5 times the value fornonadopters, primarily because of increasedmaize output, but also as a result of their othercrop production. The importance of maize as apercent of the total crop value is the same forboth groups, while maize as a proportion oftotal annual income flows increases in signi fi­cance for adopters (Table 5.6).

Average maize output per adult triples withhybrid maize adoption. Mean minimum annualmaize subsistence requirements are higher forthe adopting households because they tend to

be larger; however, because their farm sizes arealso greater, the amount of maize per hectarethey need to produce to meet their require­ments is lower. Consequently, adopting house­holds are better off both with respect to abso­lute maize output and maize output relative torequirements.

The boost in maize output could, other fac­tors held constant, imply improved caloric in­take and, through maize sales, a eli versi fied diet

(more oils and protein) for adopting house­holds. Other factors are likely to dilute, but notoff<;et, the apparent consumption and nutritionalgains. First, because many of the adopters areclub members, some of their hybrid maize out­put is used to repay loans. Even when hybridmaize is not sold to repay loans, denty hybridswere usually sold to meet cash needs becauseof their poor storability and processing charac­teristics and may have had less of a direct effecton nutrition than the new semiflint hybrids. Tothe extent that local maize is more frequentlyintercropped than hybrid maize, growing hy­brid maize could have a slight negative effecton nutrition. Since most adopters also growlocal maize and, in zones where intercroppingis frequent, hybrid maize is increasingly inter­cropped, the last effect is likely to be negli­gible.

As a positive effect of hybrid maize adop­tion on nutrition, farm households that growearlier maturing hybrids are able to consumemore green maize in the hungry season andharvest earlier. If it is true that mgaiwa (whole­meal flour) is more nutritious than ufa woyera(refined white or "pure" flour), adopting house­holds who consume their own hybrid maize aswhole grain flour may also recei ve some nutri­tional benefit.

Potentially, the food security position ofhybrid maize growers could be less precarious,but the food security impact of hybrids is prob­ably more evident on an aggregate than on ahousehold level. Without the hybrid maizeoutput marketed by adopters, maize-deficithouseholds would probably have to pay higher

18

maize prices in the hungry season-i fthey couldprocure maize at all. In part, the marketingsystem for hybrid maize has operated to redis­tribute the less preferred varieties, at a cheaperconsumer price, from production surplus todeficit areas. When it is valued in terms ofnational food security, the shadow price ofhybrid maize output is greater than its nominalvalue.

Yield and Economics Risks o/Hybrid MaizeAdoption

A comparison of either observed or expectedcumulative yield distributions for fertilizedhybrid maize, fertilized local maize, and unfer­tilized local maize demonstrate that the fertil­ized hybrids grown in Malawi are less riskywith respect to yield than either fertilized orunfertilized local varieties. On the other hand,relative riskiness of net returns (one aspect ofeconomic risk) depends on the pricing relation­ships assumed. If local maize is given a valuepremium expressing superior processing andstorage efficiency, and households are assumedto produce local maize only for home con­sumption, fertilized local maize appears lessrisky than fertil ized hybrid maize. When theconventional,assumptions used to compare prof­itability are employed, the results are inconclu­sive and depend on the nature of individualfanners' attitudes toward risk. In other words,for all fanners, yield prospects are less riskywith hybrids. For some fanners hybrid maizecultivation poses more of an economic riskthan local maize production. The fact that nosingle technology dominates with respect toriskiness of returns suggests that farmers maybe able to reduce total economic risk by sowinga portfolio of varieties.

The cumulative distributions also show thatthe total probability of negative returns, or"downside risk" is always greater with fertil­ized hybrid maize relative to fertilized or un­fertilized local maize. When fanners operatewith limited resources, producing a small sur-

Table 5.7. Labor Returns and Total Factor Productivity,Hybrids and Local Maize

Maize TechnologyFertilized Fertilized Unfertilized

Characteristic Hybrid Local Local

Yield (kgs/hal 2;774 1,264 745Price (MK/kgl q.29 0.29 0.29

Transport and harvesting costs d.04 0.04 0.04,

Gross Returns (MK/ha) 694.50 316.00 186.25

Seed Costs 1 (MK/hal 37 6.5 6.5Fertilizer2 196.35 72.1Credit Charges 28.00 8.65

Variable Costs (MK/ha) 261.35 87.25 6.50

Gross Margins (MK/ha) 432.15 228.75 179.75

Gross Margins/Person-hourJ (MK/hour) 1.16 0.66 0.59

Total Factor ProductivitY' 1.49 1.10 0.95

'25 kgs/ha2 For hybrid maize, 170 kg/ha urea and 85 kg/ha DAP; for local maize, 75 kg/ha urea and 20 kg/ha DAP.3 Six-hour days; 62 person-days for hybrid maize, 58 person-days for fertilized local maize. and 51

person-days for unfertilized local maize.4 Rental rate for land = MK 123.50 (Jere, 1990).N = 420 farmers in Blantyre, Mzuzu, and Kasungu Agricultural Development Divisions.

Source: Maize Variety and Technology Adoption Survey, CIMMYT/MOA, 1989-90.

plus in one year and a deficit in the next, therisk of low or negative economic returns maybe of primary importance in their decision­

making. 2o

Returns to Labor and Total FactorProductivity, Hybrids and Local Maize

Returns to labor in maize production for local

maize (fertilized and unfertilized) and hybridmaize (fertilized) are shown in Table 5.7. Theseries have been constructed using experimen-

20 Details of method and construction for the aboveanalyses are reported in Smale et al. (1992). Economet­ric evidence that sowing hybrid maize does not involvegreater yield risk than sowing local maize in DowaWest, Kasungu ADD, is also found in Bulla (1990).

19

tal data for labor hours and CIMMYT/MOAsurvey data on returns, expenditures, and wages.The figures are comparable to, but lower than,those calculated in representative budgets byPlanning Division, Ministry of Agriculture.Under farmer conditions in 1989/90, on the

average, adoption of hybrid maize roughlydoubled returns to labor in maize production.

Preliminary estimates of total factor pro­

ductivity (the value of output divided by thetotal value of inputs) were also calculated forthe three maize technologiesY Output, variable

~1 Farm-level prices were used. No attempt wasmade to distinguish tradeables from non-tradeables,establish world market reference prices, or account forthe effects of policy on relative prices. In addition tothese caveats, prices do not include premiums for flint

in!uts, and the rental rate for land were valuedas shown in Table 5.7. Minimal capital costsw rc not valued. Estimated total factor produc­tivity for unfertilized local maize is 0.95; forfertilized local maize 1.10; and for fertilizedhybrid maize, 1.49. The figures suggest thatunfertilized local maize, still the dominant tech­nology, is relatively unproductive in Malawi'sland-scarce conditions. Fertilization (at aver­age rates for the sample) improves estimatedtotal factor produeti vity by approximately [5pekent. Adoption of hybrid varieties plus fer­tilization increases it by over 50 percent. The

I

predominance of maize in the cropping systemeven when total factor productivities are gener­ally so [ow may be explained by the lack ofalternative crops, the conventional pricing as­sumptions employed, or both.

For example, less conventional assumptionsmight reflect such considerations as (I) themajority of farm households produce less thantheir maize subsistence requirements; (2) yieldlosses in processing can be as high as 25 per­cent for denty hybrids; (3) storage losses foruntreated denty hybrids are also very high; and(4) costs ofprocuring fertilizers are much higherfor farmers who are not club members. Withthese assumptions, the comparisons of tech­nologies favors fertilized local maize. The samecalculations can be produced with various setsof assumptions (that are meant to characterizevarious farmer subsets) and generate contradic­tory sets of figures.

Agricultural Development DivisionImpact

Changes in Mai7.e Technology

NSSA and ASA data sources provide maizetechnology information for the 1980s, but only

character in local maize or consumption penalties fordeficit producers Results should therefore be treatedwith caution.

20

case study information is available prior to thattime. The NSSA figures show that the percentof farmers growing hybrids and composites inMalawi was roughly equivalent at 5 percent in1980/81. In some ADDS sLlch as Mzuzu,Kasungu, and Salima, around 10 percent ofsmallholders grew composites Crable 5.8).21Since that tin~e the adoption rate for compos­ites appears toihave declined in the major maize­producing zones except for the [990/91 season,while hybrid lnaize adoption rates show a netincrease in all zonesY

The perio'd toward the mid- I980s appearsto have been rather sluggish, with the lowestadoption rates occurring in the 1986/87 season,the year following ADMARC's major financialcrisis that created problems in purchasing hy­brid maize from fam1ers. The last few years ofdata reveal perhaps the most rapid increase andthe highest cumulative adoption rates recorded.Because farmers sow both varieties and allo­cate fairly small proportions of their individualfarm area to hybrids, expansion of aggregatearea sown to hybrid maize has grown at equiva­lent rates but has reached lower cumulativepercentages (Table 5.9). Hybrid maize repre­sents a large percentage of aggregate maizeoutput, however. Because ofmeasurement prob­lems, the additional effect of composites onaggregate output through cross-pollination, aswell as the direct effect, cannot be determined_

One interesting socioeconomic highlight ofthese figures is that although Lilongwe ADDwas the principal focus of early promotional

2lSince composites are difficult to identify becauseof recycling, this figure can alternatively be interpretedas first-year composite seed. Recycling occurs whenfarmers retain grain from their harvest for usc as seed111 the following seasons. It is recommended that com­posites be recycled only once or tWice.

2.\ More recent aggregate adoption data for thesubsequent seasons (not shown here) shows a contin­ued upward trend. Data sources are described in Ap­pendix A. Hybrid area estimates reported from thesample surveys for the last few seasons are also consis­tent with estimates derived from seed sales.

Table 5.8. Varietal Adoption Characteristicsby Agricultural Development Division, 1980-1990

Agricultural Development Division

Blantyre Liwonde Lilongwe Kasungu Mzuzu + Salima

Percent of farmers sowing hybrid maize

1980/81 1 2 15 8 121985/86 1 12 27 241986/87 1 5 26 151987/88 4 9 201988/89 8 7 26 181989/90 14 22 33 381990/91 30 34 39 40 18

Percent of farmers sowing composite maize

1980/81 2 2 3 10 141985/86 2 1 1 31986/87 1 1 11987/88 2 0 21988/89 0 31989/90 1 2 21990/91 7 11 9

Percent of farmers sowing local maize

1980/81 98 98 99 951985/86 98 99 97 941986/87 99 99 95 991987/88 99 99 991988/89 99 97 991989/90 97 99 971990/91 98 96 99

+ Excludes Nhkata Bay, a cassava-producing zone

9

18

74

Source: National Sample Survey of Agriculture, 1980-81, National Statistical Office, Government ofMalawi; Annual Survey of Agriculture 1985-1989. Ministry of Agriculture. Government of Malawi; for

last two seasons, CIMMYT/MOA Maize Technology and Varietal Adoption Survey, 1989-91. and FoodSecurity and Nutntion Monitoring Reports 2 and 3, Ministry of Agriculture, 1990 and 1991.

efforts, by the 19805 both Kasungu and Mzuzu(excluding Nkhata Bay) ADDs had higher adop­tion rates but were far less studied. Until thelast two seasons, hybrid maize adoption inBlantyre and Liwonde ADDs was hardly dis­cernible. The geographical emphasis of socialresearch on parts of Lilongwe, Blantyre, andLiwonde ADDS probably contributed to the

21

misconception that hybrid maize adoption wasnegligible. In that context, perhaps the mostsurprising figures are those recently reportedfor Blantyre ADD.

Average maize yield figures illustrate dif­ferences among zones, not only because ofhigher adoption rates in Kasungu and Mzuzufor most years, but because hybrid maize yields

I

Table 5.9. Variety as Percent of Aggregate Maize Area ~d Aggregate Output,by Agricultural Development Division, 198 -1981

Agricultural Development Division

Blantyre ·liwonde Lilongwe Kasungu : Mzuzu + Salima

Hybrid maize as percent of aggregate maize area

1980/81 8 5 61985/86 1 7 15 16

1986/87 1 1 3 14 9 21987/88 2 2 7 11 13 21988/89 3 3 6 13 12 71989/90 6 5 1 1 13 22 101990/91 1 1 10 14 23 19 161991/92 11 1 1 20 19 19 161992/93 28 15 19 39 31 31

Composite maize as percent of aggregate maize area

1980/81 1 5 7 51985/86 3 1 1 2 1 11986/87 1 1 1 131987/88 1 1 1 161988/89 1 2 1 91989/90 1 3 2 101990/91 1 1 91991/92 2 61992/93 1

Hybrids as percent of aggregate maize output

1980/81 1 10 141985/86 2 15 24 351986/87 2 2 6 28 35 51987/88 2 4 16 22 26 71988/89 7 9 13 30 24 181989/90 18 16 26 44 47 281990/91 26 28 35 41 42 371991/92 22 39 47 40 52 401993/93 52 41 43 62 60 59

+ Excludes Nhkata Bay, a cassava-producing zone

Source: National Sample Survey of Agriculture, 1980-81, National Statistical Office, Government ofMalawi; Annual Survey of Agriculture 1985-1989, Ministry of Agriculture, Government of MalawI; forlast two seasons, CIMMYTIMOA Maize Technology and Varietal Adoption Survey, 1989-91; Govern­ment of Malawi, National Crop Estimates, 1985-1993.

22

1980/811985/861986/871987/881988/891989/901990/91

1980/811985/861986/871987/881988/891989/901990/91

1980/811985/861986/871987/881988/891989/901990/91

1980/811985/861986/871987/881988/891989/901990/91

Table 5.10. Mean Maize Yields by Variety,by Agricultural Development Division, 1980-1981

Agricultural Development Division

Blantyre Liwonde Lilongwe Kasungu 'Mzuzu'

Unfertilized local maize (t/ha)

1.0 0.8 1.0 1. 1 0.91.2 0.9 1.2 1.01.1 1.0 1 .1 1.01.2 1. 1 1.11.0 1. 1 1.3 1. 10.7 0.9 0.60.8 1.2 0.9

Fertilized Local Maize (t/ha)

1.6 1.1 1.8 1.7 1.41.2 1.4 1.7 1.41.8 1.6 1.7 1.31.9 1.6 1.71.5 1.7 2.1 1.51.2 1.4 1.21.4 1.6 1.6

Fertilized Hybrid Maize (t/ha)

2.1 2.1 3.5 3.0 3.02.8 2.6 3.0 3.33.0 3.2 2.7 3.22.3 3.7 3.22.1 3.3 3.2 3.22.2 3.0 2.93.1 2.7 3.6

All Maize (t/ha)

1.1 0.9 1.4 1.5 1.31.2 1.1 1.6 1.51.3 1.3 1.6 1.31.5 1.4 1.51.2 1.4 2.0 1.5

+ Excludes Nhkata Bay, a cassava-producing zone

Source: National Sample Survey of Agriculture, 1980-81, National Statistical Office, Government ofMalawi; Annual Survey of Agriculture 1985-1989, Ministry of Agriculture, Government of Malawi; forlast two seasons, CIMMYT/MOA Maize Technology and Varietal Adoption Survey, 1989-91.

23

Table 5.11. Illustrative Welfare Changes for Hybrid Maize Adopters in Blantyre,Kasungu, and Mzuzu Agricultural Development Divisions, 1989-1991

Characteristic Agricultural Development Division

Blantyre Kasungu Mzuzu1989-90 (MK) (MK) (MKI;Mean net returns per hectare

hybrid maize 381 533 455local maize 178 238 186

;Percent of total area sownhybrid maize 4 14 18local maize 92 67 65non-maize crops 4 19 17

1990-91Mean net returns per hectare

hybrid maize 531 593 651local maize 250 376 246

Average gross margins/hafor non-maize crops ' 355 830 369

Percent of total area sownhybrid maize 11 16 16local maize 86 54 66non-maize crops 3 30 18

Inter-season welfare effects + +

hybrid maize effect 2,192 -868 -2,025allocation effect -639 20,169 1,070interaction effect 56 -1,237 -162

net 1,609 18,064 -1,117per adopter household 39 347 -21

, Weighted by area; figures are overstated because labor costs are not included." Assumptions and construction described in text; figures are for survey farmers only and totals are

not expanded to survey zone.

Source: Maize Variety and Technology Adoption Survey, CIMMYT/MOA, 1989-90. N=420 farmers inBlantyre, Mzuzu and Kasungu Agricultural Development Divisions, with 140 farmers in each zone.

r

also appear to be lower 10 Blantyre (Table5.IO).2~

An E'tample ofSmallholder Welfare EfFects

One way of depicting the total economic effectof changes in area sown to hybrid maize is to

24Hybrid maize yield results for the years and zoneswith lower adoption rates may be a consequence ofsmall subsample sizes.

24

separate it into three components: (I) the per­centage change in hectares sown to hybrid maizevalued by the difference in net returns per hect­are from sowing hybrid maize rather than localvarieties; (2) the percentage change in hectaressown to nonmaize crops valued by the net valueof nonmaize crop returns; and (3) the percent­age change in hybrid maize area given no changein nonmaize area, valued by the difference innet returns per hectare from sowing hybrid

maize rather than local varieties. The first (hy­brid maize effect) is the simplest and mostobvious measure of farmer wei fare impact. Thesecond (allocation effect) expresses the impactof reallocating land from maize to nonmaizecrops, but carries the effects of other economicand technological variables than hybrid maizeadoption and will therefore tend to be over­stated. Unless there is a large increase in totalcultivated area, the first and second effectsshould have opposite signs because they aresubstitutes. The third (interaction effect) "cor­rects" the hybrid maize effect by controllingfor the effect of land reallocation, and willgenerally be fairly small in magnitude.

Results of calculations with the C[MMYTIMOA data are not realistic in absolute amountsbut provide an example of how hybrid maizeadoption affects farmers and agroeconomiczones differently (Table 5.11).25 When totalcultivated area is held constant, the area sownto hybrid maize increased and nonmaize areadecreased only in the Blantyre survey zone.Consequently, the only positive hybrid maizeeffect between the two years is found in thatzone. The negative effect of the small declinein hybrid maize area was greatest in Mzuzu,where the difference in net returns from grow­ing hybrid maize is also greatest. The associ­ated increase in nonmaize area is very large inKasungu, and combined with average nonmaizereturns that are several times as high because oftobacco revenues, the positive allocation effect

)'To separate and measure the effects, total culti­vated area must be assumed constant. A reasonableassumption in gross terms, actual cultivated area fig­ures do show some change over time - whieh weakensthe welfare calculations. Conventional pricing assump­tions were used, but could be modified to expressvarious farm household scenarios as described above.Observed yields were used rather than expected yields,but these were very close at the mean, especially forlocal maize. Net returns were calculated from theCIMMYT/MOA data, and include labor costs for maizebut not for non-maize crops, which further overstatesthe allocation effect.

25

dominates the total wei fare effect. The mean­ing of the relative figures is clear. [n Blantyreand MZUZLl, for di fferent reasons, maize is amore dominant crop and the hybrid maize ef­fect is correspondingly greater in magnitudethan the other two components. In Mzuzu, wherehybrid maize competes with other cash crops, adecline in hybrid maize area is likely to gener­ate a relatively large welfare loss to farmers. [nKasungu where per hectare returns to tobaccoare considerably higher than hybrid maize re­turns, the reallocation effect, to the extent thatit measures land released by cultivation of hy­brid maize instead of local maize, generates thelargest wei fare gain.

[n other words, crudely speaking, the great­est potential impact of hybrid maize adoptionin Blantyre is likely to be felt in a reduction ofthe household food deficits (especially if thehybrids are flinty);26 the largest effect on MzuZLlfarmers is in terms of cash crop production ofhybrid maize; and in Kasungu, the most signifi­cant aspect of adoption is the land it releases forthe production ofmore remunerative cash crops.The figures merely illustrate the notion that thewelfare impact of hybrid maize adoption inMalawi is likely to vary dramatically by region,or fam1ing system and farm household type.

)6As noted above, parts of the Blantyre survey zonehave characteristics that parallel the conditions de­scribed by Low (1986) for other zones in southernAfrica. The Low model offers two reasons why subsis­tence producers may adopt hybrid maize. Deficit house­holds will adopt hybrids if it is cheaper in labor time toproduce staple requirements through growing morehybrid maIze and purchasing less maize on the market- implying labor returns to own production are greaterthan returns in off-farm wage employment. Break-evenhouseholds will adopt hybrids if fewer labor units arcrequired to meet subsistence requirements and themarginal labor unit can be released to earn a higherreturn in off-farm employment than in farm production.Low emphasizes that, although hybrid maize adoptionmay result in a welfare improvement for adopting house­holds. in neither case would adoption result in tn­

creased commercial production and marketed output.

National '1pact

The generall approach used to assess the na­tional impact of maize research over time by nomeans provides an exhaustive account of re­search il11pa~t or rate of returns to research.Instead, it is' based on the useful notion that,beca use 0 f interactions and offsetti ng effectsamong underlying and exogenous economic

variables, trends in aggregate production fig­

ures often dipguise the extent of other changesinduced by shifts in technology. The purpose ofconstructing i various scenarios is to illustratehow certain key economic variables would haveevolved with and without maize research.

For Scenario I, or the "actual" case, expo­nential trends are fitted to maize yield, area,production and consumptIon (availability) datato smooth fluctuations resulting from climaticconditionsY Net imports is the estimated re­sidual of production less consumption andchange in stocks. Agricultural GOP and GOPseries are also fitted by exponential trends interms of 1978 Malawi kwacha.

In Scenario II, or "static yield," yields areheld constant at the 1961-65 average, maizearea changes according to the "actual" trend,and per caput consumption is held constant atthe 1961-65 average. Of particular importanceis the fact that, because per caput maize avail­ability exhibits a declining trend over time, the1961-65 average is slightly higher (230 kgs/person) than the average for the 1986-1990period (190 kgs/person). Net imports are thencalculated as in Scenario I, as the residual fromestimated figures. Agricultural GOP and GOP

27Since there arc no (hsccrnible distinct trends over

subsets of the data ye'lrs. exponential trends are closeto those produced by 5-year moving averages, althoughsmoother. Data for 1961 to 1990 IS from U.S. Depart­ment of Agriculture, Economic Research Service foryield, production, impo!1s, exports, consumption, andstocks variables. The GDr series and population seriesused to construct per caput consumption are from Pryor(1988). The real maize price series used to value pro­duction IS from Gulhati (1989).

26

series are tabulated by adding the real value ofmaize production estimated under Scenario IIto the "actual" agricultural GOP and GOP sc­ries from all nonmaize production.

Scenario II depicts the production, net Im­

ports, and GOP situation when farmers manageto use enough fertilizer to maintain maize yieldsdespite declining soil; fertility from maize

monocropping over an lextended time period.

No new varieties are rel~ased. The GOM has amajor policy goal of sustaining per caput maizeavailability at 230 kgs/person which is consid­ered the minimum tolerable level of consump­tion. Production shortfalls relati ve to consump­tion requirements result in increased net imports.Maize area expands to further dampen the ef­fects of declining soil fertility and temporarilybuoy national production levels, with deleteri­ous effects over the longer term because moremarginal lands are opened and the economybecomes more dependent on a single crop.Under Scenario II, production reaches an as­ymptote as the proportion of total cultivablearea sown to maize reaches I or all fannersapply fertilizer at their economic optimum,whichever occurs first.

Scenario III expresses "declining yield."Maize yields decrease at one percent per yearfrom the 1961-65 average, area expands at the"actual" rate, and per caput consumption isheld at the level consistent with food policygoals. Net imports and GDP figures are calcu­lated by the same method described in ScenarioII, with Scenario III production figures. In Sce­nario m, no fertilizer is used and no varietiesare released. Population pressure and consump­tion preferences slowly deplete the land re­source base with no offsetting technologicalchange. The production, net maize imports,and agricultural GOP results for the three scc­narios are shown graphically in Figures 5.1 to5.3.

The effects on total GOP (not shown here)are similar to those on agricultural GOP, al­though lesser in magnitude. In Figure 5.1, ac­tual yield trends combined with expansion of

..

5.1. Malawi Maize Production - Three Yield

)Ii

i

5.2. Malawi Net Maize Imports-Constant Per Capita Consumption

u...0..,

•.1

l ...f ...

1--...I.'

......n.t .1..

•_...

5.3. Malawi Agricultural GOP-Three Yield Scenarios

.", -r--------------->00.

>00

"""

.......,1«>uo""" ..f-c=r-~~_._-_,_-__.,_-~-~-~_,-......_-._._-...__~_4

'."'4 U17, ".77 ,.7. 1.7. f.a.a , ... , 19-A.:;! , ..... r ...... ,-...a \ .... ""1 ,'I,q , .....

27

hectares sown to maize causes national maizeproduction to roughly double since 1961. Thatincrease is approximately halved in the "staticyield" scenario, with no maize research andlimited use of fertilizer. In the "yield decline"scenario, maize production is nearly unchangedin 1990 from the 1961 level, and is kept at that'level only through continual expansion of maizearea. If maize area were held constant to ex­press a policy goal of at least some diversifica­tion of crop output (recall that 1961 maizealready occupied an estimated 66 to 75 percentof cultivated area), maizc production woulddecrease in Scenario [11.

If Malawi were autarkic (no trade), the re­sults of either Scenario [( or IlIon food secu­rity would be dramatic. Maize area would ex­pand quickly to maximum cultivable area andthere would be no means by which to sustainthe population. Prices would rise prohibitivelyand the GOM would need increasing funds tosubsidize consumer prices. To meet minimumconsumption needs, even if Malawi trades, theeffect of either static or declining yields is toincrease net imports six- and tenfold in 1990(Figure 5.2). Ifmaize area expanded more rap­idly to offset static or declining yields, the areadevoted to alternative cxport crops would di-

minish and Malawi's agriculture-l'ascd economywould gradually become unable 0 finance thevolume of imports. Even if 111, ize area ex­panded at the "actual" rate, agricultural GOPwould be cut by an average of5 percent perannum in Scenario II and 9 rerc~nt per annumin Scenario [II (Figure 5.3). Total GOP wouldbe reduced by 2 percent and nearly 4 perccntcach year, respectively. There would be norecourse for the GOM but greater indebted­ness, with little means for repaYlnent.

A more complete macroecohomie modelwould be necessary to generate reliable quanti­tative estimates of research impa~t for the vari­ous scenarios, but the essential point remainsclear in the case of Malawi. Without maizeresearch and at least gradual tcchnologicalchange, the nation's food security and macro­economic position would rapidly deteriorate.In an agriculture-based economy, and whenboth national agricultural production and indi­vidual producer livelihood is based on maize,maize research is critical. In that sense, thevalue of maize research cannot be overstated.The relevant policy issue is how to increasemaize research impact by speeding the technol­ogy adoption process.

6. Factors Affecting the Impact of MaizeResearch on Varietal Adoption

High labor to land ratios, the significance ofmaize in the cultural and economic setting, andthe: rdati vcly high ranking that Malawi enjoyswith respect to some major development indi­cators,"X suggest that conditions are favorablefor varietal adoption. As shown by the adoptionfigures in Section V, until recently, changes inadoption rates have been gradual and overalllevels fairly low. This section summarizes thecomplex of factors that are likely to have af­fected the extent and speed of the impact ofmaize research on varietal adoption in Malawi.

Varietal Preferences

Despite the many differences among the zonesin the CIMMYT/MOA survey, farmers statedalmost universally that they preferred localmaize for consumption and consume their ownor purchased local maize during more than sixmonths of a typical year. Only one or twohouseholds in the full sample of 420 reportedthat they have changed their consumption pref­erences toward hybrid maize, despite the factthat over three-quarters of households in allzones consume their own or purchased hybridmaize during the hungry season or immediately

28Sofranko and Fliegel (1989) concluded thatMalawi ranks fairly high relative to other African na­tions with respect to such indicators as the percentageof the smallholder faml population using credit (nowaround 25 percent and higher in major maize-produc­ing zones I: the extension staff/farmer ratio; and thenumber and spread of buying and selling points forinputs (since decreased with the curtailment ofADMARC activities). However, they also found thatthe level of credit (MK/member) is low and that thesame indicators also vary widely by region.

29

after harvest, in years of poor local maize har­vest or duress."'>

The consumption preference for local maizeis based primarily on its processing and storingcharacteristics, which reflect the flintiness ofthe varieties. The traditional processing methodused to produce the socially preferred, finewhite flour (ufi.l \1!oyera) involves multiple stagesand is labor intensive. Typically, the shelledmaize is dehulled with a mortar and pestle,winnowed, soaked for lactic fermentation, dried,and pounded again by mortar and pestle or,increasingly, by hammermill. Some of the dentyhybrids can be used to produce uJa woyera, buttheir relative softness leads to a lower flourextraction rate from shelled maize, and addi­tional sand and water are often needed to createa proper pounding medium.

Otherwise, hybrid maize can be processedmore quickly by hammermill and without lac­tic fermentation to produce a nutritious, al­though coarser and less prestigious flour(mgaiwa). During the colonial period, milledwhole grain meal was typically fed to laborersand the inmates of hospitals, schools, and pris­ons. Refined flour with the bran added (for lackof better term, "homemade" mgaiwa) was notconsidered a suitable food for men, and "evenwomen" were "often ashamed to admit that

29Maize-deficit households are obliged to consumehyhnd maize because most of the maize available onthe official market during the hungry season is hybrid.Local maize can sometimes be purchased or obtainedfrom other farmers through ganyu (labor exchange forseed or in-kind payment), but only a small percentageof farmers are surplus local maize producers, and evenfewer are willing to part with local maize. Instead, it isoften the poorer households who must selI a basket ora bag of local maIze to meet some pressing eash needs.

they eat it." Despite that fact, mgaiwa was saidto be "more sustaining than ordinary porridge,giving more strength for hoeing" (Williamson,1956).

Observers of hybrid maize adoption pat­terns in Malawi and other zones where localvarieties are flinty have often remarked thatwhen labor-savil~g mills are widely introducedin rural areas, hO~lseholds will find it optimal to

I

substitute milled:grain for the traditionally pre-ferred fine white flour that is so laboriouslyproduced by hand-pounding methods. Yet al­most all the women respondents in theCIMMYT/MOA survey have combined the tra­ditional with the modern method. They con­tinue to pound by hand, but where they used topound two to three times, they now substitutegrinding at the local mill for the final stage.The fact that women substitute grinding at thelocal mill only at the final stage of processingwas noted as long ago as 1959.30

Preparation of maize for consumption re­mains extremely labor-intensive. Williamson's(1956) estimate of labor time in preparationwas, for the first and second pounding only,about 12 to 13 hours, excluding dehusking,winnowing, soaking, washing, sieving, and re­lated minor processing steps. The processingmethods described by Williamson (1956) andby CIMMYT/MOA respondents (1991) werealmost identical, differing only in that today,the second or third stage of pounding is usuallyreplaced by milling. Even then, the time re­quired to walk to the mill and wait in line maynot be that much shorter than another stage ofhand-pounding. A conservative estimate of la­bor time for flour preparation might be about

lOElIis wrote that there had been "an enonnousincrease" in the number of small hammer mills inNyasaland in previous yr;ars and more people couldconsume whole-meal flour "If they wished." He ob­served generally that the women still separated themadeya (bran) from the IIlphale (broken grain) in themortar and took the IIlphale to the grinding mill to bemade into flour.

30

[5 hours for five days of flour (an averagefamily size).ll

Use of secondary products from the processmay still be important to many rural womenand m\IY contribute to their choice of poundingmethods. Williamson (1956) cited a number ofjoint products of the traditional processingmethod that are consumed by household mem­bers. When food supplies were plentiful, branwas fed to chickens, but during famine or thehungry season, more of the bran was consumedby household members. At any time, the branwas used to brew beer. Children and pregnantand lactating women often consumed by-prod­ucts of the pounding process, such as discardedwhole and broken grains, germ meal, or thefine bran mixed back into the refined whiteflour.

Despite the apparent strength of consumerpreferences, some aspects of maize consump­tion have changed slightly over time. Mosthybrid maize adopters in the CIMMYT/MOAsurvey stated that they occasionally consume aportion of their own hybrid maize harvest as ameans of bridging food requirements, althoughless so in Mzuzu zone. A large percentage ofadopters try to solve the problem by poundingdifferently, rather than by consuming mgaiwa.For example, women often pound hybrid maizewith larger amounts of sand and water, for ashorter period of time, and soak it for fewerdays. Some are more successful than others inproducing an acceptable ufa woyera. The Blan­tyre households seemed comparatively morewilling to consume mgaiwa, perhaps becausethey have less experience modifying their

\I Another indicator of the processing time is foundIII a report produced by the Agroeconomlc Survey(1982) In that report, of all staple foods in Malawi,maize has the highest value-added as a percent of thetotal value of the commodity. Using prices from asurvey of local markets, AES concluded that the con­version of maize grain through fennented broken maize(mphafe) to maize (lour (ufa) more than doubles thepnce.

pounding met1ds or possibly because a largerproportion of hem are food-deficit and havegrown accusto ed to consuming mgaiwa madefrom purchased maize in the hungry season.Most hybrid adopters consume their hybridproduction sogn after harvest rather than ap­plying actcllic,' the recommcnded insecticide.

Farmers' hybrid maize output is not an un­important factor in household food security,despite the prestige of ula woyera, the lowerextraction ratei from processing hybrid for lilawoyera, and Storage losses that occur whenactellic is not applied. For the majority of farmhouscholds wh~ purchase from ADMARC (theofficial marketing agency) rather than fromother farmers, the CIMMYT/MOA data alsoindicate that the modal consumption period forhybrid maize over the past few years has beenapproximately two months across strata.Whether farm households like it or not, theyhave often been obliged by their productionand market conditions to consume denty hybridmaIze.

Flinty hybrids can increase the area hybridmaize adopters allocate to hybrid maize if theyare more substitutable in consumption and stor­age than denty varieties. Those who are able toadopt but have not yet adopted may be morewilling to grow a flinty hybrid than a dentyhybrid. But flintier hybrids cannot relieve un­derlying expenditure constraints or inability toqualify for credit. Even those farmers who canafford to purchase inputs cannot be expected torelinquish their local sources of seed until theycan rely on marketing institutions for timely,certain delivery of quality seed meeting theirown specifications.

Whether or not varietal consumption pref­erences change over time will be affected notonly by the release of flintier hybrids but alsoby the quality and characteristics of local seed.Certainly in regions where a number of im­ported varieties have been introduced in thepast, cross-pollination has occurred. Wherefood-deficit households who are obliged toconsume their seed actually sow denty varieties

31

obtained as food, the degree of contaminationof local maize mllst be noticeable to farmers.The 1991-92 drought severely reduced and al­tered sources of local germplasm. Under theseconditions, over time, the polarity between theflintiness of local maize and dentiness of exist­ing hybrids becomes less clear. Following thedrought, the new semiflin~ hybrids may repre­sent some of the flintier rfaizc types availableto small farmers. I

The Role of Research32

Maize Research Themes and Priorities

The evolution of maize breeding themes inMalawi can be largely depicted in terms of twopoles: flint as opposed to dent character and, byassociation, the issue of hybrids versus open­pollinated varieties. Although, as in most breed­ing programs, yield may emerge most frequentlyas the foremost objective, the history of Mala­wi's program is one in which, at various pointsin time and varying degrees of intensity, otherbreeding criteria such as grain traits and grow­ing season have been considered. The concernfor flintiness has also created an alternativedefinition of yield-yield from the mortar.

JZMuch of the material in this section is drawn fromwritten communication or oral interviews with individu­als who have been involved with Malawi's maize researchprogram but who are not responsible forinterpretationsandconclusions. They include members ofthe Maize Commod­ity Team: B.T. Zambezi, Breeder, J.D. Kumwenda. V.H.Kabambe, W.O. Sakala, Agronomists, R.B. Jones, Agrono­mist and J. Wendt, Soil Scientist; as well as M. Collinson,Economist, fonnerlyCIMMYT/Nairobi; B Gelaw, MaizeBreeder, fonnerlyCIMMYT/Nairobi and nowCIMMYTIHarare; A. Hansen, anthropologist, fannedy with the Uni­versity of Florida project at Chitedzc Research Station;EJ.R. Hazcldcn, Executive Director, National Seed Com­pany of Malawi; P.W. Heisey, Regional Economist,CIMMYT/Malawi; G.Y. Mkamanga, fonnerCARO; B.R.Ndisale, Research Economist, Department of AgriculturalResearch; L. Ngwira, Deputy CARO; K. Short. Maize

Breeder, CIMMYTlHarare.

iFlint vs. Dent JFlint character has b en a recognized breedingobject in Malawi since Ellis began the maizebreeding program before independence, but withgreater or lesser imp011ance depending on thetime period. Above the themes of rust-resis­tance, good husk cover, and low-carried cobs,"the question of whether to breed for a flint ordent maize was of first importance" in the earlyyears of the prograrp (Ellis, 1959, p. 251).While he acknowledged that the dent commer­cial hybrids tested in Malawi produced highcob yields and shelling percentages, Ellis em­phasized that insect damage during storage andlocal preferences were of major concern. Asshown in the time chart in Table 4.1, the bestamong the first hybrids (LHll) and the syn­thetics (SV 17, SV28, SV37) he bred weresemi flints. Improving smallholder yield in con­sumption (net of crop damage, postharvestingand storage losses) rather than harvest yieldwas clearly what Ellis had in mind.

During the 1970s, Bolton appears to haveinterpreted the flint-dent question differently.In a 1974 article, he recognized the preferencefor white flint maize and the susceptibility ofthe dent SR52 (Rhodesian) to weevils in villagestorage conditions, but found that SR52 was"the highest yielding variety tested." By argu­ing that SR52 could be "successfully grown bycash crop farmers with good standards of crophusbandry" and "high fertilizer input," he im­plied that more well-endowed smallholders andestates should grow it as a cash crop (p. 108-10).When hybrid maize is produced as a cash cropand sold immediately after harvest, processingand storage losses do not reduce effective yield.To address the consumption needs of farmerswho could not afford the cash outlays to pur­chase seed in every season, Bolton bred semiflintcomposites. The reason for producing semi flintrather than flint OPVs was that flints tend toproduce small rounded kernels with low testweight. The idea behind OPVs was gradualchange, at lesser expense for farmers.

32

The late 1970s and early 1980s appear tohave been a period when new importance wasattached to the indigenous production of hy­brids, in part as a result of the continued popu­larity and yield advantages displayed by SR52in southern Africa. Large commercial farmerswho produced hybrid maize for sale or as foodfor laborers rather than for their awn consump­tion, demanded the importation! of the high­yielding, denty SR52. After borders were closedin Rhodesia with the Unilateral beclaration ofIndependence, SR52 was air-freighted intoMalawi at exorbitant cost. Pressure to replaceimports led to the release in Malawi of MH I2,based on Zambian SR52 material, which yieldedless than the Rhodesian version because thelines had lost their purity.

Dent character was necessarily associatedwith the demand for hybrids. Even in otherparts of the world where maize is used in hu­man consumption more than as animal feed,dent varieties have been preferred because theyare more suitable for large-scale processing byroller mills (Kydd, 1989). The fact that globalmaize-breeding activity has concentrated on denthybrids also limits the range ofgennplasm avail­able for breeders seeking flint materials. Tech­nology development in southern and easternAfrica has been restricted by the belief thatdent maizes have higher yield potential thanflints-in part a result of the early researchbreakthroughs with U.S.-bred dent materials(Blackie, 1989). While dent germplasm was"on the shelf," flint lines would have requiredseven years to develop.

Not only was breeding dent hybrids easier,cheaper, and faster, but there were argumentsfor promoting dent hybrids over flint varieties(Kydd, 1989). Although the colonial research­ers cited by Ellis and Williamson argued thatufa was more nutritious than mgaiwa becauseof lactic fermentation and other characteristics,most nutritionists since that time have insistedon the superiority of mgaiwa. A second argu­ment was that the prevalence of mechanicalmills in rural areas would change consumer

preferences because women would prefer toconserve labor time. Researchers were told thatthc urban roller mill operators, who were seenas the fastest growing component of the marketfor smallholder maizc surpluses, preferred dentsas being less injurious to their machinery. An­other argument was that the insecticide neces­sary to protect dent hybrids in storage is cheapcompared to the yield loss associated with flinthybrids. [n any case, some of the breeders postedto Malawi may have discounted complaints ofstorage losses because it is generally true thatHYVs have higher postharvest losses simply asa result of poor adaptation to traditional storagemethods (Lozano and Leopold, 1988).

In the early 1980s when the Malawian maizebreeders left for long-term training, lines forboth the dent hybrids MH15 and MH 16 and thesemi flint composites CCC and CCD had beenestablished. When they returned in 1986, pres­sure had accumulated, primarily on the part ofdonors, to produce flint hybrids. In the lateI980s, flint character became once again a prin­cipal breeding objective, for both hybrids andOPVs.

The earlier notion of promoting dent hybridmaize as a cash crop undoubtedly had implica­tions for which groups of smallholders adoptedhybrid maize (Gilbert et a\., 1982) and cumu­lative adoption rates. From farmers' perspec­tives, different decision-making criteria areassociated with cash crop and food crop pro­duction. From a development perspective, dif­ferent resource sets are required for profitableproduction of high-management cash crops.Promotion of hybrid maize as a cash crop ineffect reduced the ceiling adoption rate.

Hybrids vs. Open-Pollinated Varieties

Lipton (1988) has written that, although maizebreeding "successes" have depended mainly onhybrids, in much of sub-Saharan Africa com­posites and synthetics make more sense buthave been much less researched. In Malawi,both hybrids and OPVs were bred since the late

33

colonial period, although at cellain times influ­ential players in the breeding rocess (donors,an individual breeder, the nat ,onal seed com­pany) appear to have strongly favored eitherhybrids or OPVs. Some of the debate overOPVs and hybrids seems to have been associ­ated with the involvement of e'xpatriate breed­ers and external agencies, and has undoubtedlybeen counterproductive.

For example, when Bolton came to Chitedzein the 1970s, the popular view flmong develop­ment agencies was that hybrids were only ap­propriate for larger farmers and progressive,commercialized smallholders., Bolton empha­sized indigenous development of compositesover hybrids. In so doing, however, he alsodecided to replace rather than revive the latecolonial synthetics, whose lines had deterio­rated after independence. In one set of experi­mental results, he speci ficall y compares LH 11,SR52, Ellis' synthetics, his own CCA and'Malawianized' UCA. He concludes that notonly is SR52 "the best yielder" and that "allfour Malawi synthetic varieties...were inferiorto the new composites," but that at least two ofthe synthetics should be dropped from the seedmultiplication program (p. 106; III). Boltonseems to have recommended a complete shiftin the program from LH II to SR52 and fromsynthetics to composites. Since then, no syn­thetics have been released.

Until recently, the International Maize andWheat Improvement Center (CIMMYT) hasexclusively developed OPVs. CIMMYT's in­volvement with Malawi's maize breeding pro­gram began in the 1970s with a visit from ErnieSprague, Head ofthe Maize Program in Mexico,the transfer of subtropical and lowland tropicalOPV material, and some short-term training atCIMMYT headquarters in Mexico. By that time,however, the Malawi program was beginningto deemphasize composite development and waskeen on releasing their own hybrids to replaceSR52. CIMMYT posted a maize breeder inNairobi during the 1980s, but not until theHarare office and midaltitude station were es-

tablished in 1985 did the CIMMYT maize breed­ers become actively involved in the discussionof breeding strategies and in germplasm devel­opment for both OPVs and hybrids in Malawi.In fact, until the late 1980s when the work ofthe Harare breeders began to producemidaltitude and locally adapted materials, muchof CIMMYT's gerl11plasm was best suited forthe lowland tropics and they had less to offer in

terms of midaltitude (subtropical) material. A

number of CIMMYT lines have neverthelessbeen used in Malawi's program and, since thelate 1980s, Malawi has been considered a pri­ority for the CIMMYT Eastern and SouthernAfrica program. n

Even when the Malawi program produceddenty hybrids intended for cultivation as a cashcrop, semiflint OPVs continued to be producedand were viewed as the appropriate seed forsmall farmers. The problem is that the syntheticlines were abandoned and the adoption of com­posites has dwindled over time. UCA lodgedeasily and, although its shortcomings were elimi­nated with the release of CCC and CCO, therehas since been very little attempt to educatefarmers about OPVs or to promote and diffusethem as an alternative to hybrids. [n theC[MMYT/MOA survey, farmers recalled UCA,few had heard ofCCC or CCO, and even fewerknew the difference between a hybrid and acomposite.J4 Little composite seed is producedand little of it is demanded, but it is not in anycase accurate to say that farmers have rejectedthe new composites.

The intensity of research interest in the flinttrait has generally accompanied shifts in re­search focus between OPVs and hybrids be­cause commercial hybrids were denty. How­ever, a number of factors converged in support

HStreak-reslstant material contributed by liTA (an­other of the CGIAR Institutions) is likely to becomeincreasingly Important in the future.

J4For example, researchers recommend that hybridseed be bought each year while composites can begrown for 3 years between seed purchases.

34

of the in~usion of flintiness ~nto hybrid df'vel­opment III the late 1980s. fhcse factor in­cluded a strengthening of the Maize COlm,lOd­ity Team; assistance from CIMMYT's regionalbreeding program; and changes in the NationalSeed Company of Malawi's (NSCM) portfoliowhich accompanied the involvement ofCargill'sbreeders. Breeding for flintiness has alwaysbeen associated with OPVs, and it was in largepart the financial pressure exerted by donors

that resulted in the pursuit of flinty hybrips.

Length of Growing Season

Commercial hybrid seed in Southern Africahas a development cycle of more than 150 daysover much of the midaltitude range in which itis grown (Low and Waddington, 1989). Of thevarieties released by the Malawi program, MH 16and MH 18 have 125- to 130-day cycles andCCD has a 120-day cycle at approximately1000 meters above sea level. Among the hy­brids, only NSCM4 I is 120-day. In most cases,however, local maize matures more slowly thanthe improved varieties-a characteristic thathas added to the attraction of hybrids.

The Malawi maize breeding program has,over the years, released a few synthetics, com­posites, or hybrids with shorter growing sea­sons. These materials, produced primarily forthe lakeshore or lower-elevation environments,have greater probabilities of drought escape.Until recently, however, the maize program hasnot consciously addressed drought tolerance orthe related issue of heat stress as a breedingobjective. Now, a low-altitude hybrid program(under W.G. Nhlane) and a low-altitude OPVprogram (under E.M. Sibale) arc operated fromChitedze, with materials tested at Chitala andother sites in the lakeshore region and ShireValley. Concentration on early maturity andheat and drought resistance could increase adop­tion rates in some less favorable environmentsof Malawi. Early maturity and drought toler­ance could also provide more options for lateplanting in more favorable environments, a fact

that is not widely recognized. Drought and heattolerance are areas in which collaboration withCIMMYT might prove fruitful.

Soil Fertility

From the co[onial period to the present, themajor nonvar~eta[ research issue in Ma[awi hasprobably bee~ soil fertility or, more broadly,soil conservation. Given the pressurc on landand the predotninance of maize in the croppingsystem, it is evident that much of the arableland in Ma[awi is continuously cropped tomaize, a crop that extracts large amounts ofnutrients from the soil.

In general, assessing the impact of cropmanagement (e.g., nonbreeding) research ismore difficult than measuring the impact ofvarietal development. The yield figures cited inthis report appear to justify the assumption thatyields of unfertilized local maize have beendeclining over time. Application of inorganicfertilizer has therefore contributed to maizeproductivity despite nutrient/output price ratiosthat are quite high by world standards. Thisapplication has been encouraged both by pastresearch results and by extension efforts.

The continuing dominance of maize in thefarming system and high costs of imported fer­ti[izer has also spawned additional researchefforts. One line of attack is agroforestry, al­though there are considerable technical andmanagerial issues that must be resolved beforethis can be recommended as a large-scale so[u­tion. Other research seeks to improve the effi­ciency ofconventional fertilizers. To the extentthat improved maize varieties perform wellunder relatively low fertility conditions, and tothe extent that improved maize yields releaseland for other crops, breeding research can alsocontribute to alleviating soil fertility problems.

35

Human Resources. the Organization ofResearch. and Research Capacity

Discontinuities

A rllajor problem with the evolution of themaize program was that instead of cumu[at­ing a depth of expertise and a range ofgermp[asm, the program was beset bydiscontinuities and shifts in staff. Initially,the dependence on expatriate staff on fixed­term contracts and later, the lacunae causedby the departure of key Malawians for [ong­term training contributed to the scantiness ofhuman resources. Kydd characterizes the [atecolonial period with Ellis (1953 to the [ate1950s), the early 1970s (Bolton's period), andthe [ate [980s as peak productive periods inthe maize breeding program. The first slow­down of activities during the 1960s is easilyattributable to the change in administration.The second, in the late 1970s, Kydd blameson (1) donor misconception, based on overlyoptimistic reports from parts of LilongweAgricultural Deve[opment Division, thatadoption rates were high enough and Malawihad no "maize problem"; (2) World Bankpreoccupation with the then-popular notionof the Integrated Rural Development Projectover technology generation; and (3) mis­guid~d allocation of national research fundstoward less important crops such as rice andcotton.

Institution Building

On the other hand, the late 1970s and early1980s were a period during which there werechanges in the organization of maize research,adaptive research became a formal part of theresearch system, and the training of numbers ofresearchers culminated in improved maize re­search capacity. In-service training was a largepart ofCIMMYT's involvement with DAR and,at one time or another, most of DAR's maizescientists attended short-courses at CIMMYT

headquarters. Although often criticized, theUniversity of Florida project3s and the of1portu­nitics it rrovided were instrumental in chang­ing the organization of research and in long­term training of scientists.

Though a small part of the project, thet:1rming systems research (FSR) component hadactivities that were designed to improve re­search and extension co[~municationand even­tually led to the establishment of the AdaptiveResearch Teams (ARTS) in the AgriculturalDevelopment Divisions. At that time, the De­partment of Agricultural Research was reorga­nized into the interdisciplinary commodity teamstructure it has today. Another structural changeencouraged by the project was to set up a seriesof discussions in which researchers exchangedresults. However, the project took from one­third to one-hal f of the experienced researchersout of Malawi for training-at only one institu­tion.

In coordination with the economist fromCIMMYT, the social scientist on the team suc­ceeded in introducing a farming systems per­spective into crops research. Diagnostic sur­veys conducted at that time resulted in twonotions that would later become more widelyaccepted. One notion was the rediscovery ofthe cultural importance of local maize varietiesand the fact that, for subsistence farmers, yieldfrom the mortar is a better indicator of theeconomic value of a variety than grain yield.This rediscovery is today reflected in the maizebreeders' continued work with local maize col­lections. Research findings also suggested thatfertilizer use on local maize should be pro­moted to reduce the area required for staplefood production and release area for cash cropand hybrid maize cultivation. Considered as anundesirable alternative strategy at the time, fer­tilizer use is now recommended on local maize,even though hybrid maize is no longer viewed

35The Uni vcrsity of Florida project was the first ofa series of USAID projects aimed at strengtheningagricultural research.

36

solely as a cash crop.

Much of CIMMYT's involvement withMalawi's research program during the 1980swas also related to the establishment of theARTs and training of adaptive research person­nel by three CIMMYT Regional Economists.Their approach emphasized the need for socialscientists and on-farm research with a systemsperspective. Adaptive Research now has one ofthe most poorly staffed teams and one of thehighest attrition rates of the research units, butcertain key concepts such as interdisciplinarycommodity teams, the importance of consider­ing a wider range of breeding criteria, and acontinued emphasis on developing technologyfor local maize have probably resulted from thework of the 1980s.

From the beginning of Malawi's maizebreeding program until 1977 and intermittentlyuntil 1987, the post of Maize Breeder was heldby a succession of expatriates on fixed-termcontracts. In all other years, the same Malawianshave held positions as breeders-B.T. Zambezi,who has been with the program for over 20years and is now the only PhD on the staff, andW.G. Nhlaneand E.M. Sibale, who have workedwith the program for about 15 years and arenow completing their PhDs. Aside from theseindividuals, technicians are an often overlookedbut important source of continuity in the pro­gram. For example, when the three Malawianbreeders were abroad on training during the1980s, the technicians maintained the lines forMHI5 and MHI6, CCC and CCD, which weresubsequently released when the breeders re­turned. With respect to technical assistance,involvement with CIMMYT has also been asource of continuity in the program.

Based on a comparison from 1985/86 rela­tive to other commodity teams, the Maize Com­modity Team now ranks high in terms of yearsof experience per researcher, level of educa­tion, percent of researchers receiving promo­tions, and low attrition rates. Promotions havelargely been among professional rather thantechnical staff. What these indicators suggest is

I

that the level 9f commitment in tcrms of crc-dcntials, lengt~ of stay, incentives and profes­sional satisfaction is greater than it is amongmany of the other teams. Behind these figuresare, compared to previous decades, the strongfinancial comn)itment of the GOM, the WorldBank, and the Rockefeller Foundation to maizeresearch in Malawi.

Another recent organizational change onthe part of the Maize Commodity Team was thepreparation oqhe 1989 Maize Action Plan. thefirst of its type among the commodity teams.The development of a flint hybrid was desig­nated as a priority concern and, with the greaterinvolvement of CIMMYT in breeding strate­gies during the last few years, CIMMYT per­sonnel assisted in drafting the plan.

Among other factors inhibiting the progressof agricultural research systems in sub-SaharanAfrica, Lipton (1988) has cited the lack of"critical mass of scientists" and "inadequateintegration of economics and social analysisinto agricultural research." Although these criti­cisms probably still hold true in Malawi, theinstitution-building efforts of the 1980s havebegun to address these issues.

Input Distribution, Marketing, and PricePolicies

Seed Supply

At various points in time, seed quality, multi­plication, and distribution problems have inter­acted with other factors to inhibit farmer adop­tion of varietal releases. For example, Quintenand Sterkenburg (1975) reported that althoughLH II was a semiflint variety, it was not verypopular-partly because of a seed supply diffi­culty. The germination quality of one of thefirst large-scale seed crops was poor whichinfluenced the demand in later years, althoughthe quality improved. In 1971-72, the seed wassupplied to the local markets very late in theseason and farmers had already decided to grow

37

local varieties.ADMARC and the Ministry of Agriculture

were responsible for seed production until thelate 1970s when the National Seed Company ofMalawi (NSCM) was established. For the mostpart, seed distribution in the early years wasconfined to Lilongwe ~and DevelopmentProgramme (LLDP). [n th«e mid- [980s, NSCMobtained government c1ear~nee to pay royaltiesto Ciba-Geigy for the FI :material to produceNSCM41, a denty hybrid that is geneticallysimilar to R20 I (Zimbabwe) and processes rela­tively well. What is not clear is the role ofADMARC's financial problems in the mid­1980s slump in seed sales, or to what extentlow seed sales figures represented a produc­tion, distribution, or demand question. The rapidincrease in sales over the past few seasons sug­gests a latent excess demand for hybrid seed­so that in some years seed supply may haveactually been the limiting factor.

Breeding and seed production under rainfedconditions affect the speed of varietal releasesand seed supply. In Malawi, sufficient irrigatedland to complete two breeding cycles per yearcould have increased the flow of improvedmaterials through the research system, assum­ing adequate staff and other resources. Withoutirrigation, when farmers are contracted for thefinal stage of seed production, certain types ofhybrids that require synchronization (MH 15)are more difficult to produce. In a poor grow­ing season such as the 1991/92 season, substan­tial amounts of seed stock can be lost.

The costs of seed production also vary byhybrid type. A single cross hybrid such as MH 12is more expensive to produce than a three-waycross like NSCM41. [n general, the higher theyield potential the more expensive the seed, sothat, with a fixed budget, the varieties thatexhibit the greatest yield differential with re­spect to local maize can only be produced insmaller quantities. Nonconventional top-crosshybrids (MH 17 and MH 18) are generally lessexpensive to produce than conventional hy­brids, but are produced at the expense ofgreater

Figure L. 1. ADMARC Sales of Improved Maize Seed19b1/82-1989/90 and Preliminary 1990/91

6

Q

:.

f- ~

f-

I,f-

! ~7.

f- t

\

198J.f82 1983/84 1985/86 1987/S8 1989/901982/83 1984/85 1986/87 1988/89 1990f)!

CROP YEAR

~ HYBRID ~ COMPOSITE

yield variability and possibly lower averageyields in fanners' fields. Over three years ofbreeders' trials, however, MH 17 and MH 18yield did not differ statistically from those ofMHI2 and MHI6, respectively. In Malawi,where contract growers are often estates, grow­ing maize seed must be at least as profitable asproducing alternative crops-such as tobacco.

The varietal composition of seed supplyhas not always suited farmer preferences be­cause little was known about effective farmerdemand. In the CIMMYT/MOA survey, farm­ers often reported that they had no choice ofhybrid varieties from year to year-they sowedwhat was in sheds or what was provided by theclubs. The simplest criterion for varietal mix isunsold seed stocks, but detennining the "cor­rect" varietal mix when farmer demand is chang­ing takes time.

Most observers believe that NSCM, nowmajority-owned by Cargill,36 has adopted a more

J6Cargill, the largest privately-held company in theworld and one of the world's largest grain traders, alsohas substantial interests in seed production worldwide.

38

aggressive approach to seed production, seedprocurement from abroad, and seed sales dur­ing the past few seasons. Figure 6.1 shows themarked increase in sales of composite and hy­brid seed since the 1986/87 low.

Seed Difli/siol1

The Government of Malawi has promoted hy­brid seed as part of a seed-fertilizer packagethat is extended through formal credit clubswith subsidized credit and stringent repaymentrequirements. In the early 1980s, credit clubmembers composed only an estimated 10 to 15percent of the farm population (Kydd, 1989),although the percentage has grown consider­ably and is greater in the high-potential maizeproduction zones. In the survey zones, for ex­ample, frol11 a quarter to a third of farm opera­tors were club members in the second surveyseason (CIMMYT/MOA data). Although creditclub membership has facilitated adoption byrelieving seasonal cash flow problems, non­members also adopt hybrid seed and fertilizer

through careful organization of the cash re­sources they obtain from sales of alternativecrops or off-farm wage income. Credit clubmembership has generally been associated withgreater chances of personal attention from ex­tension workers.

[n thc past, the packages that were distrib­uted to club mcmbers were of a fixed size andcomposition. Credit club members sowed theseed variety that was provided in the packageand applied the type of fertilizer they receivedon one-acre (OA-hectare) plots. This diffusionmethod created a lumpiness in land allocationand curtai led fanners' experimentation and theirability to adapt the technology to their ownconditions. Currently, in recognition of the needto address a wider range of technological op­tions, the government has begun to providepackages of varying size and composition. Al­though hybrid seed was always promoted withfertilizer, recent research results indicate that,in some zones, hybrid varieties can yield aswell or better than local varieties with no fertil­izer. Of the two input costs, fertilizer is un­doubtedly the more limiting. Malawi is a land­locked country and fertilizer imports aretransported overland at high cost. In recentyears as part of donors' structural adjustmentprogram, fertilizer subsidies were graduallyreduced at the same time that internal strife inMozambique blocked Malawi's cheapest exter­nal transport route. Unwilling to pass all thecost burden to farmers, the government eventu­ally abandoned the subsidy removal.

Recent research has demonstrated thatMalawi hybrids perform well even under rela­tively low-input conditions and that for somefarmers, adopting hybrids without fertilizer maybe economical (Jones and Heisey, 1993). Iffam1ers are convinced they must grow hybridmaize with fertilizer, or if they are only permit­ted to purchase seed with fertilizer, their inabil­ity to pay for fertilizer inhibits their seed choice.On the other hand, the percentage of farmerswho usc fertilizer on local maize is higher thanthe percentage of credit recipients. Credit re-

I

cipients are a subset of fertilizer u~ers. Whenfarmers can afford fcrtilizer, the ipcrementalcost of seed is slight-i f that seed can be foundin local markets.

·Marketing System 17

Irregular marketing conditions have also im­peded the purchase of both seed and fertilizerby noncredit club members. Initially, fertilizerand seed in rural areas were sold I at officialADMARC outlets. These markets' were notevenly dispersed in all village ards, nor didthey always stock inputs. When ADMARCoperations began to incur heavy finariciallosses,many of these input sheds were closed. Bycontrast, inputs were delivered free to creditclub members. The dependence on credit as adiffusion mechanism undoubtedly slowed thedevelopment of private markets for inputs. Inisolated areas outside the credit system, seedand fertilizer are still not easily found.

Official maize output prices are announcedseasonally, and are panterritorial, uniform overthe harvest season, and equal for all maizevarieties. Although few price series exist, withmarket liberalization there is increasing evi­dence of price di fferentials between hybrid andlocal varieties and intraseason price variationon local markets. The difference in the wayfarm households value local and denty hybridmaize may appear in price differentials in localmarkets but is suppressed in the official price.Because ofconsumer preferences for local maizeand the credit repayment system, a higher pro­portion of hybrid maize circulates in officialmarkets. Local markets in many rural areas arealso likely to be thin, especially in certain sea­sons.

When the official prices capture little eco­nomic infom1ation, and private markets haveonly begun to operate, either observing truevaluations for maize or studying farmers' re-

\7Appendix B contains more details on the historyof maize marketing.

39

sponscs to these valuations is difficult. Econo­metric analysis of the CIMMYT/MOA datanevertheless shows that although the effects ofconventional price ratios and relative profit-

: ability variables on hybrid maize adoption arcweak, the same measures do significantly af­fect the level of nitrogen appl ied by farmers tolocal maize. In hybrid maize adoption. otherfactors such as the diffusion method (credit),consumer preferences (subsistence require­ments), and learning have probably played alarger role because ofcontrolled prices. A higherproportion of farmers purchase fertilizer withcash for their local maize and, when possible,prefer to sell their local varieties on local mar­kets where prices are generally higher and morevariable.

Promotional Efforts and Farmerlearning

Perhaps as a result of early fanner responses todenty varietal releases, hybrids have been gen­erally promoted as a cash crop. Over time. assuggested above, some changes in consump­tion patterns (if not consumption preferences)have undoubtedly occurred and even dentyhybrids have played a role in household foodsecurity. Especially in the Blantyre zone andsome parts of the Kasungu zone, CIMMYTIMOA survey farmers often ranked earlier ma­turity above yield as an advantageous charac­teristic of hybrid varieties. Now, especially withthe flinty hybrid releases, the importance ofproducing a more flexible promotional effortthat emphasizes food security as well as poten­tial cash income should attract the interest of abroader base of fanners.

Focussing on profitability of hybrid maize,combined with limiting its diffusion to creditclubs and emphasizing the importance of fol­lowing rigid recommendations, may have lim­ited the receptivity of large subsets of fanners,and even those capable of self-financing. Cul­turally, the term "local" is usually associated

40

I

with certain traits, such as flintiness and whitelness. Chil11{/f1ga cha makolo means maize 0

the ancestors, or a gift conferred by familie.through generations. Some researchers havesuggested that "local" is not so much a termreferring (0 grain characteristics, and that cJ/(lm(/ko!o docs not literally describe the origin ofthe seed. but that both are terms signifying theseed's institutional affiliation. 1x In some sense,"local" maize is the maize of rural people, or offarmers. By contrast, released varieties anJbrought to the locality from outside (formaIjinstitutions, whether these are national or inter~

national. Occasionally, survey farmers calledhybrids chimanga cha borna, which means, incommon parlance, "maize of the govemment."1'JSince hybrid seed is either purchased or pro­vided on credit which is repaid by selling atleast some of the harvest, many farmers prob­ably did not perceive that the seed was theirown or was produced with their interests inmind.

Extension messages with single themes wereundoubtedly useful in the early introductions,but over time may have discouraged farmerexperimentation that might have resulted inadoption and greater farmer benefits. 40 For ex-

lBBased on his farming systems research work dur­ing the 1980s, Hansen ([ 986) has described [ocal maizeas a folk category with two defining characteristics: (I)flintiness; and (2) the seed did not come directly fromthe government but was retained from the previousharvest or purchased on the local market. At the sametime, Hansen discovered a "[ocalization" process bywhich governmental origins were forgotten in areaswhere seed exchange and introduction programs hadbeen active. [n those areas, such as Lilongwe ADD, thelocal category also included the names of old releasedmaterial that had been recycled. In the CIMMYT/MOAsurvey the same phenomena were found in differentlocalities.

'"The origm of boma is British Overseas MilitaryAdministration, designating administrative bases Incolonial territories.

.oWriting about parts of Li[ongwe ADO in [987,CllIpande wrote that the least endowed fann house­holds (female-headed) did not even bother to ask for

ample, the emphasis on pure stand cultivationfor hybrids is now relaxing as field workersobserve that fanners in some zones have rea­sons for intercropping maize, whether it is ahybrid or a local variety. Smallholders whoboth consume and market crops have diverseobjectives, and producing hybrid maize underconditions that maY; not be agronomically opti­mal may nevertheletss be economically optimal

I

for them.On the other hand, continual exposure to

other farmers who grow hybrid maize and, morerecently, to radio messages that exhort farmersto grow hybrids has probably contributed to theupsurge in adoption, particularly in the SOLlth­ern Region. Analysis of the CIMMYT/MOAdata confirms that farmer experience with hy­brid varieties increases the probability of sow­ing hybrids in successive years. Once a "criticalmass" of hybrid maize growers accumulates ina given locality, the general level of knowledgeabout the varieties also increases. Those withlimited levels of working capital are more ableto experiment "passively" (by observation) than"actively" (by paying the costs of gaining in­fonnation from their own fields). Farmers whoobserve success and who have the resources toadopt can then adopt at faster rates and there isan increase in the slope of the aggregate di ffu­sion curve, as is evident in the figures from theSouthern Region.

credit because of their assessment of their land andlabor constraints--"they were afraid." In the CIMMYTIMOA survey, some households expressed the samesentiments with respect to growing fertilized hybndmaize on credit.

41

Farming Systems

Although maize dominates the farming sys­tems of all of the major maize-producing zonesof Malawi, essential agroeconomic differencesexist between zones that probably determinedi fferential adoption ceilings even when otherfactors such as consumption preferences anddi ffusion mechanisms are similar. For example,in parts of the Southern Region the importanceof off-farm income and relative wages in laborallocation decisions implies that hybrid maizewi II be adopted only when labor returns aregreater in hybrid maize production than in al­ternative income-earning activities. In Kasunguand parts of Lilongwe ADDs, by contrast, adop­tion rates and area sown to hybrid maize can beexpected to fluctuate in response to competi­tive conditions with alternative and more remu­nerative cash crops. The higher proportion offull-time fanners on the Central Lilongwe andKasungu plains and their relatively greater yieldpotential because of soils and rotations sug­gests that the ceiling adoption rates will behigher than in the South. In Rumphi and MzimbaDistricts of the Northem Region, the centralimportance of maize in the cropping system asboth a food and a cash crop suggests that adop­tion ceilings will be highest and cumulativelevels most stable in that zone, other factorsheld constant.

7. Conclusions

The evidence collected in this paper suggeststhe following conclusions regar1ing the historyand effectiveness of the nationa'l maize breed­

ing program in Malawi:

(I) Malawi's maize program, as compared toother conventional breeding programs inthe region and elsewhere, did incorporatesocioeconomic considerations into its breed­ing objectives. Since its inception the pro­gram has addressed the consumption pref­erences of small farmers which are relatedto the processing and storing characteristicsof flinty varieties, by breeding semi-flinthybrids or semi-flint open-pollinated vari­eties (OPVs).

(2) Until the recent development of the twosemi-flint hybrids MHl7 and MHI8, themajor underlying constraint on the speed ofrelease of flinty varieties had been lack ofsuitable flint germplasm. Inbred lines de­veloped from local flinty materials are tootall and have too long a growing seaSOn.Exotic flint germplasm was difficult to lo­cate because the focus of most maize breed­ing efforts in other parts of the world hadbeen denty varieties.

(3) Item (2), when combined wi th thediscontinuities in senior staffing and finan­cial support through the 1970s, led to coun­terproductive conflicts in breeding objec­ti ves and swings in emphasis amongsynthetics, composites and hybrids. At criti­cal points in the early years of the program,a weak decision-making structure and theneed for more trained Malawians in deci­sion-making positions severely curtailed

42

germplasm development.

(4) Flintiness is only one of many breedingfactors and socioeconomic factors that haveaffected the impact of varietal innovationson adoption rates.

lessons learned

Specifically, the above conclusions are associ­ated with the following "lessons" for breedingprograms with that operate in similar condi­tions:

• The flint maize preferences of farmers ledto complexity in breeding objectives. Themajor constraint to breeding popular flinthybrid varieties was not that breeders ig­nored the signi ficance of flint character, butthat there was limited local and exotic flintgermplasm that was also high-yielding, shortin stature, and shorter in growing season.Each of Malawi's major breeders, in oneway or another addressed a concern for"yield from the mortar," either by attempt­ing to breed a semi-flint hybrid or a semi­flint OPV.

• In the early (post-independence) years ofthe program when varieties were distrib­uted primarily in the Lilongwe area, theeffective demand for hybrid seed was foundamong commercial farmers whose foremostconcerns were harvest yield and productionfor sale. To mcct the perccived demands oftwo groups ofclients--eommercial farmersand subsistence farmers-the program pur­sued the dualistic strategy of importing thehigh-yielding, denty SR52 from Zimbabwe

for cash crop producti~n and developingflinty OPYs for smallh lders. The need torcplace hybrid seed imp rts because ofhighcosts led them to the development of dentyindigenous hybrids. Breeding denty (ratherthan flinty) hybrids was,expedient and wasa first step in indigenou~ varietal diversifi­

cation.• Although development and importation of

denty hybrids and their promotion as a cashcrop effectively reduce9 the ceiling adop­tion rate by focusing 0n larger or morewell-endowed producer~, the breeding pro­gram always worked with OPY alternativesdesigned to meet the 'maize subsistenceneeds of smallholders. Two problems af­fected the progress of the OPY program:(I) discontinuity in breeders; and (2) a lim­ited range of high-yielding, mid-altitudematerial suitable for developing Malawianlines. An example of (I) is the deteriorationof the synthetic lines bred by Ellis and theirsubsequent rejection by Bolton. An exampleof (2) is that although CIMMYT breederssent mid-altitude (at the time, "sub-tropi­cal") materials to Malawi in the 1970s and1980s, their more attractive materials werenot developed until the mid-altitude stationwas established in Harare in 1985.

• For OPYs to have been successful (inMalawi they have been popular to a moder­ate extent and over brief periods in selectedlocalities), they needed yield, disease-resis­tance, drought-resistance or early maturity,in addition to flintiness. The history ofOPYsuccesses shows that both OPYs as well ashybrids need to be "spectacular." OPYs canbe high-yielding, however, and the argu­ment that only hybrids will work in Malawiis unfounded. In any case, OPY develop­ment is of continued importance in breed­ing lincs for kernel texture and other desir­able characteristics to usc in the hybridprogram, and in maintaining a varietal port­folio.

• For either hybrids or OPVs to have been

43

adopted at a steadier and faster rate wouldhave required more of a commitment toseed production and distribution. Althoughit may be true that the involvement of aprivate seed company can provide a keyimpetus at certain stages of the breedingprocess, in most success stories the role ofprivate companies ill seed distribution hasbeen even greater than their role i~ breed­ing. On the other hand, private seed:compa­nies are not usually as interested irt OPYs.To guarantee that OPYs are given? chancewith farmers, a conscious public sector ef­fort is ncedcd to distribute the seed widelyand to educate farmers about the relativeadvantages and disadvantages associatedwith OPYs and hybrids.

• Flintiness is not the only issue affectingresearch impact. The diffusion mechanism(limited to credit clubs and packages offixed composition and size), the varietalcomposition of seed supply, the economicrisk of taking fertilizer and seed on credit(or of allocating land away from subsis­tence production), and farmer learning(which takes time and accumulation in alocality) influence farmer adoption deci­sIOns.

• Flintiness is also not the only importanttrait that affects varietal adoption. Otherimportant breeding issues involve, for ex­ample, plant stature and length of growingseason.

• Even the discontinuities in funding, staff­ing, and breeding objectives that were re­lated to the turnover of expatriate breedersand ebb and flow of financial support wouldnot have jeopardized the program if therehad been more senior Malawian breedersbefore the mid-1970s. Since then, althoughthe three Malawian senior breeders havetaken over decision-making responsibility,overseas training has caused some disrup­tions. The program will soon have threePhD-trained breeders with lengthy experi­ence-but there is no "younger generation"

of breeders in line to follow them. Thesheer number, and not the quality of thepersonnel has been a problem. At this criti­cal juncture in the breeding program whcnthe impact of recent varietal releases isbecoming apparent, the need for a newgcneration of breeders to sustain varietaldevelopment cannot be overstated. Thc ex­perience of the maize program has shownthat the next generation of breeders is usu­ally best drawn from promotions within thesystem, from technical to professional of­ficer.

• The need for socioeconomic contributionsto the maize program has been recognizedsince the early 1980s, but the capacity forsocioeconomic research has not been SLlC­

cessfully institutionalized.• In a nation where maize is of such critical

socioeconomic importance, the issue is notwhether maize research should be fundedbut how to improve the efficiency of maizeresearch through addressing some of theabove concerns. As demonstrated clearly inthe Malawi case, the impact of maize re­search should also be viewed in terms ofthe welfare loss associated with no maizeresearch.

Windows of Creativity

The recent release of MH I7 and MH 18 byMalawi's national research team is an exampleof how the scientific creativity of several indi­viduals has coincided with certain conditions togenerate the potential for rapid technologicalchange. The new hybrids are the first semi­flints developed since the colonial period andhave the processing and storage traits valued by

I

small farmers as well as the yields that compareto the denty hybrids previollsly grown as cash

,crops (Smale et a!., 1993). The speed of their. release (only three years after the initiation ofthe semi-flint hybrid program in 1987) can be

~ attributed in part to the convergence of factors,. including (I) the idea of brecding a top-crossrather than a conventional hybrid; (2) the com­fortable working relationship with CIMMVT'sregional breeders that enabled the Malawi team

i to identify appropriate parent material in Popu­'Iat[on 32; and, most importantly, (3) the years:ofdevelopment and maintenance of parent lines,by technicians and breeders as they graduallyaccumulated germplasm and experience. Thework of the three senior breeders, B.T. Zambezi,E.M. Sibale, and G. Nhlane, was publicly rec­ognized for the first time when they receivedthe MASTA (Malawi Award for Scientific andTechnical Achievement) from the Governmentof Malawi for the new hybrids. Additional donorsupport to the maize program may have facili­tated the progress of the maize team by en­abling its members to obtain advanced degreesand pursue their research with fewer opera­tional constraints. However, without the dedi­cation of the breeders to their work duringmore difficult years, the breakthrough wouldnot have occurred so rapidly. Concurrently,adoption rates for denty hybrids have been ris­ing as weather conditions underscore the yieldadvantages of the shorter-season hybrids overfarmers' varieties, and as the quantities of seedproduced and marketed have increased. Thescientific breakthrough, combined with thegrowing receptivity of farmers to hybrids andgradual improvement in seed production andmarketing have created a situation that is ripefor major technological change in Malawi'sfarming communities.

44

References

Anderson, D., "Fluctuations of Maize andGroundnuts Yields in the Lilongwe LandDevelopment Programme." World BankDevelopment Policy Staff, Washington:Thc World Bank, 1975.

Agroeconomic Survey, "An Analysis of Agri­cultural Producer Prices in SelectedLocal Markets in Malawi." Lilongwe:Ministry of Agriculture, 1979 and 1982.

Blackie, M.J., Maize in East and Southern Af~

rica. Mimeo. Lilongwe: RockefellerFoundation, 1989.

Blackie, M.J., "Maize, Food Self-Sufficiencyand Policy in East and Southern Af­rica." Food Policy 15 (1990): 383-394.

Bolton, A., "Response of Maize Varieties inMalawi to Various Environmental Fac­tors." Proceedings of the 5th East Afri­can Cereals Research Conference.Lilongwe: March 10-15,1974.

Bowbrick, P., "An Economic Analysis of theImpact of Private Traders on Agricul­tural Marketing." Mimeo. Lilongwe:Planning Division, Ministry of Agricul­ture, 1988.

Bulla, a.M., Input Demand and Supply Re­sponse Behavior of MalawianSmallholder Farmers. MS thesis.Urbana-Champaign: University of Illi­nois, 1990.

Brown, P., "Maize Growing in Nyasaland I.Distribution and Cultivation." EmpireJournal ofExperimental Agriculture 31(1963): 71-82.

Brown, P., "Maize Growing in Nyasaland II.Fertilizer Requirements." ExperimentalAgriculture 2 (1966): 49-60.

Chipande, G.H.R., "Innovation AdoptionAmong Female-Headed Households:

45

The Case of Malawi. " Development andChange 18 (1987).

Christiansen, R.E. and L.A. Stackhouse, "ThePrivatization of Agricultural Trading inMalawi." World Development 17, 5(1989): 729-740.

Christiansen, R.E. and V. Roy Southworth,"Agricultural Pricing and MarketingPolicy in Malawi: Implications for aDevelopment Strategy." Prepared for theSymposium of Agricultural Policies forGrowth and Development sponsored bythe Government of Malawi, Mangochi,1988.

Darrah, L.L. and L.H. Penny, "Altitude andEnvironmental Responses in the 1970­71 Eastern African Maize Variety Trial."Proceedings of the 5th East AfricanCereals Research Conference. Lilongwe:March 10-15,1974.

Ellis, R.T., "The Food Properties of Flint andDent Maize." The East African Agricul­tural Journal 24, 4 (1959): 251-253.

Food and Agricultural Organization of theUnited Nations, Food Balance Sheets1967 to 1977, Rome: FAO, 1980.

Ferguson, A.E., A.V. Millard and S.W. Khaila,"Crop Improvement Programmes andNutrition in Malawi: Exploring theLinks." Food and Nutrition Bulletin 12(1990): 273-278.

Gilbert, E. et al., A Review of the AgriculturalResearch System of Malawi. Report tothe Government of Malawi. The Hague:International Service for National Agri­cultural Research (ISNAR), 1982.

Gulhati, R., Malawi: Promising Reforms. BadLuck. Economic Development Institute(EDI) development policy case series,

Analytical case studies, nO.3. Washing­ton, D.C.: The World Bank, 1989.

Hansen, A., "Farming Systems Research inPhalombe, Malawi: The Limited Utilityof High Yielding Varieties," in J.R.Jones and B.J. Wallace, Social Sciencesand Farming ~)~\'stems Research: Meth­odological Per.spectives in AgriculturalDevelopment, Boulder: Westview Press,

1986.Hansen, A., Department of Anthropology,

University of Florida, Gainesville, let­ter to R. Tripp of CIMMYT EconomicsProgram, February 21, 1990.

Heisey, P.W., Regional Economist, CIMMYT,letter to D. Winkelmann, Director Gen­eral of CIMMYT, Mexico, March 4,1992.

Hirschmann, D., "Malawi's 'Captured' Peas­antry: An Empirical Analysis." Journalof Developing Areas 24 (1990): 467­488.

Hirschmann, D. and M. Vaughan, "Food Pro­duction and Income Generation in aMatrilineal Society: Rural women inZomba, Malawi." Journal of SouthernAfrican Studies 10, 1 (1983): 86-99.

Humphrey, D.H., "Malawi's Economic Progressand Prospects," Eastern African Eco­nomic Review, 3 (1974): 35-47.

International Maize and Wheat ImprovementCenter (CIMMYT), 1989/90 CIMMYTWorld Maize Facts and Trends: Realiz­ing the Potential of Maize in Sub-Sa­haran A/i-ica, Mexico, D.F., 1990.

Jere, O.A.H., Evaluation ofSmallholder Agri­cultural Technology in Malawi. MSthesis. East Lansing: Michigan StateUniversity, 1990.

Jones, R.B., and P.W. Heisey, "PreliminaryResults from the MOA/UNDP/FAOFertilizer Programme 1989-1992."Maize Commodity Team and MOAIUNDP/FAO Draft Working Paper.Lilongwe: Malawi Ministry of Agricul­ture, 1993.

46

Kadyampakeni, J., "Pricing Policies in Africawith Special Reference to AgriculturalDevelopment in Malawi." World De­velopment 16, I ( (1988): 1299-1315.

Kaluwa, B.M., "Private Traders' Response toFood Marketing Liberalization: MarketIntegration and Economic Efficiency."Paper presented at the National Work­shop on Food Security, Socio-Economic

Status and Nutrition of the Rural Popu­lation in Malawi. Zomba: ChancellorCollege, 1990.

Kandoole, B., B. Kaluwa and S. Buccola, "Mar­ket Liberalisation and Food Security,"in Southern A/i-ica: Food Security PolicyOptions, ed. M. Rukuni and R.M.Bernsten, University of ZimbabwelMichigan State University Food Secu­rity Research in Southern AfricanProject. Harare: Department of Agri­cultural Economics and Extension, Uni­versity of Zimbabwe, 1988.

Kettlewell, R.W., "Agricultural Change inNyasaland: 1945-1960." Food ResearchInstitute Studies 5 (1965): 229-285.

Kirchner, J., 1. Singh, and L. Squire, "Agricul­tural Pricing and Marketing Policies inMalawi: A Multi-Market Analysis."Draft CPD Discussion Paper No. 1985­17. Washingtion, D.C.: The WorldBank, 1988.

Kydd, J., "Maize Research in Malawi: Lessonsfrom Failure." Journal of InternationalDevelopment 1 (1989): 112-144.

Kydd, 1. and R.E. Christiansen, "StructuralChange in Malawi since Independence:Consequences of a Development Strat­egy Based on Large-Scale Agriculture."World Development 10,5 (1982): 377­396.

Lavers, G., "Marketing and Price Policy: AnEconomic Analysis of the Maize Sec­tor." Mimeo. Lilongwe: Planning Divi­sion, Malawi Ministry of Agriculture,1988.

Lele, LJ., "Sources of Growth in East African

Agriculture." The World Bank 1:'coJlomic!?eviel1' 3, 1 (1989a): 119-144.

Lcle, U., "Structural Adjustment, AgriculturalDevelopment and the Poor: Lessonsfrom the Malawian Experience."MADIA Discussion Paper 9. Washing­ton, D.C.: The World Bank, 19WJb.

Lcle, U., W.I-1. Kinsey, and A.O. Obeya, "Build­ing Agricultural Research Capacity inAfrica: Policy Lessons from the MAtlACountries." Paper prepared for the jbintTAC/CGIAR meeting, June 21, 1989.

Lilongwe Land Development Programme, "Re­port on the Survey of Potential Adopt­ers of Improved Maize." Lilongwe,1971.

Lipton, M., "The Place of Agricultural Re­search in the Development of Sub-Sa­haran Africa." World Development 16,10 (1988): 1231-\257.

Low, A., Agricultural Development in South­ern Africa: Farm Household Econom­ics and the Food Crisis, London: JamesCurry, 1986.

Low, A.C., and S.R. Waddington, "On-FannResearch on Maize Production Tech­nology for Smallholder Farmers inSouthern Africa: Current Achievementsand Future Prospects." Paper preparedfor the Third Eastern and Southern Af­rican Regional Maize Workshop, Nai­robi, 17-23 September, 1989. Harare:CIMMYT Southern African RegionalProgram, 1989.

Lozano, J.L., and A.C. Leopold, "The Physiol­ogy and Genetics of Seed Aging." Re­cent Advances in the Conservation andUtifitization orGenetic Resources. Pro­ceedings oj' the Glohaf MaizeGermplasm Work'>hop. Mexico, D.F.:CIMMYT, March 1988.

Malawi Ministry of Agriculture, "Food Secu­rity Situation in the ADDs--Proceed­ings of a National Meeting," FSNMReports 2 and 3, Food Security andNutrition Monitoring Project, Lilongwe,

47

1990 and 1991.Ministry of Economic Affairs, Federation of

Rhodesia and Nyasaland, Report on (//1

Economic Survey (~r Nyasaland. 1958-59. Salisbury. ,

Peters, P.E., "Cash Cropping and' Food Produc­tion Among Smallholders in Malawi."Mimeo. Cambridge: Harvard Institutefor International Development, 1988.

Peters, P.E. and M.G. Herrera, Cash Cropping,Food Security and Nutrition: The E./~

fects ofAgricultllral CommercializationAmong Smallholders in Malawi. Pre­pared for the Agency for InternationalDevelopment. Cambridge: Harvard In­stitute for International Development,1989.

Pryor, F.L., Income Distribution and EconomicDevelopment in Malawi: Some Histori­cal Statistics. World Bank DiscussionPaper No.36. Washington, D.C.: TheWorld Bank, 1988.

Pryor, F.L., "Changes in Income Distributionin Poor Agricultural Nations: Malawiand Madagascar." Economic Develop­ment and Cultural Change (1990): 23­45.

Quinten, K. and 1. Sterkenburg, "Marketing ofSmallholder Agricultural Produce inMalawi: The Production, Consumptionand Marketing of Maize." AES No.17,Vol.II. Lilongwe: Ministry of Agricul­ture, 1975.

Roberts, DJ., "Comparative Maize Yield Study1970171." Mimeo. Lilongwe: CentralRegion Agricultural Office, 1972.

Sahn, D.E. and J. Arulpragasam, "The Stagna­tion of Smallholder Agriculture inMalawi." Food Policy (l991): 219-234.

Schulten, G. and D. Westwood, "Grain StorageProject 1969-1972." Zomba: ExtensionAids, Ministry of AgrIculture, 1972.

Smale, M., Z.H.W. Kaunda, H.L. Makina,M.M.M.K. Mkandawire, M.N.S.Msowoya, D.J.E.K. Mwole, and P.W.Heisey, Chimanga Cha Makolo. Hy-

hrids, and Composites. lin Ana~"sis orF'armas' Adoption or Maize Techllol­

ogy ill Malawi. /989-9/. CIMM YTEconomics Working Paper 91/04,Mexico, D.f.: CIMMYT, 1991.

Smale, M., "Risk, Disaster Avoidance, andFarmcr Experimentation: TheMicroeconomics of HYV Adoption inMalawi." PhD dissertation. CollegePark; University of Maryland, 1992.

Smale, M., Z.H.W. Kaunda, H.L. Makina, andM.M.M.K. Mkandawire, "Farmers'

48

Evaluation of Newly Released MaizeCultivars in Malawi: A Comparison ofLocal Maize, Semi-Flint and Dent Hy­brids." Lilongwc and Harare: CIMMYT,1993.

Sofranko, A.J. and F.C. Fliegel, "Malawi'sAgricultural Development: A SuccessStory?" Agricultural Economics 3€1989): 99-113.

William~on,J., Useful Plants ofMala wi. Zomba:Government Printer, 1956 (reprinted1972).

I

Appen1ix A

Principal Data Sources

A. CIMMYT/MOA Maize Varietal andTechnology Adoption Survey

The data collection effort was funded by theIntcrnational Maize and Whcat ImprovementCenter (CIMMYT) and implemented with thesupport of the Department of Agricultural Re­search and the Planning Division, Ministry ofAgriculture. Entitled the CIMMYT/MOA MaizeTechnology and Varietal Adoption Survey(MVTS). the field research was conducted in 3of the 5 major maize-producing AgriculturalDevelopment Divisions (ADDs) of Malawi, toa subset of households (420) participating inthe Annual Survey of Agriculture (ASA) dur­ing two cropping seasons.

Following the recommendations of the De­partment of Agricultural Research, segments ofBlantyre, Kasungu, and Mzuzu ADDs werechosen as representative of contrasting agro­ecological and economic characteristics foundamong major maize-producing regions. Vari­ables hypothesized to affect maize technologyadoption differ sharply between the zones, andthe 3 zones constitute the strata for the MVTS.

Within each ADD, households were selectedfrom the multi-stage, stratified cluster samplingframe designed by the National Statistical Of­fice for the ASA. The survey households fonna statistical sample drawn with equal probabil­ity of selection within each of the 3 zones andvarying probability of selection between zones.Statistical statements generated from the dataare broadly representative of fann householdslocated in the major maize-producing, higherpotential adoption areas of Malawi.

The MVTS was designed as a module at­tached to the ASA and covering a subset ofhouseholds included in the 1989/90 national

49

sampling frame. In consultation with Ministryof Agriculture officials, this design was chosenbecause the ASA households are selected withprobability sampling procedures, the ASA ques­t,ionnaires elicit extensive agronomic data relatedto maize technology, and the ASA enumerators~re fully-trained, professional field investigatorsthat reside in survey villages. The MVTS in­cluded questions designed to provide more de­tai led varietal information, information onfanner perceptions and experience with variet­ies, and supplementary wage and price infonna­tion used in valuing costs of production, output,and household income. The variables measuredin this report therefore represent a combinationof those assembled from the routine ASA data,additional variables collected in the MVTS, andvariables composed by transfonning and com­bining the complementary data sets.

B. Food Security and NutritionMonitoring Project

The Food Security and Nutritional Monitoringproject, implemented in the Ministry of Agri­culture as a collaborative effort between thePlanning Division and Food and Nutrition Unit,has as its objective to provide periodic infor­mation that can be used to monitor the house­hold food security position and nutritional sta­tus of the population in all 8 AgriculturalDevelopment Divisions. The sample frame forthe FSNM survey is also the ASA frame, but 10out of the 20 households in each EnumerationArea (EA) are interviewed for a total of about2500 households. The questionnaire consists of"Food Security," "Expenditure," and "Nutri­tion" modules.

c. Annual Survey of Agriculture

Originally designed for the 1980/81 NationalSample Survey of Agriculture, the ASA sam­pling frame is a stratified cluster design inwhich the nation's area is divided into approxi­mately 200 agroecological strata, and each stra­tum (primary sampling unit) is composed of a

varying number of EAs or sampling clusterscomposed of roughly equal populations. Thestrata are contained within 8 Agricultural De­velopment Divisions (ADDs), which are ad­ministrative and development units responsiblefor implementing and evaluating the RuralDevelopment Projects (ROPs) in their geo­graphical zone. Evaluation Units in the ADDsimplement the ASA, as well as other smaller­scale surveys requested by the Ministry ofAgriculture.

Since 1985/86, Evaluation Officers for theASA have implemented a rotating EA sampledesigned in that year by the National StatisticalOffice. In each EA the enumerator responsiblefor collecting the ASA data lists all householdsat the beginning of the cropping season (200­300), and from the list frame a systematic ran­dom sample of 20 households is drawn. Selec­tion of strata with probability proportionate tosize of population, and subsampling of clustersof about equal size ensures that the overall

50

!

probability of includi~g a household In thesample is the same fot all households in thepopulation. More dens~ly populated ADDs andstrata have a larger number of EAs, sampleEAs, and sample households. Changes in popu­lation have eroded the self-weighting designover time with subsequent difficulties in devel­oping weighting schemes for computations ofaggregates, but the 1989/90 sample was basedon a revised stratification and more recent 1987Population Census fig*es.

The ASA questionnaires consist of 4 mod-i

ules, or schedules. The household fonn con-tains questions on the demographic composi­tion ofthe household, hours and type ofoff-fannemployment, hours and type of hired farm la­bor, the value of remittances, and quantity oflivestock owned by the household. Data aregathered in 4 visits covering the calendar year.The garden worksheet is used for measurementof all fields and plots operated by the house­hold. The plot survey records basic agronomicinformation for each plot cultivated by thehousehold as well as objective yield measure­ments from yield subplots, and is administeredintermittently throughout the cropping season.The operator form consists of questions aboutuse of credit and extension services by thehousehold, and is addressed to the operator(s)following harvest.

... '\<

Appendis B

Economic and Policy Context ofSmallholder Maize Production

A. The Pre-Independence Period

Malawi's pre-independence economy has beencharacterized as consisting of plantation,smallholder, and labor reserve sub-economies(Kydd and Christiansen, 1982). Reliant on worldmarket opportunities, the plantation sub­economy was founded successively on coffee,cotton, tobacco, and tea. Encouraged by thecolonial administration through land alienation,hut taxes, and tenancy arrangements, its socio­economic impact was heaviest in the SouthernRegion and in some parts of Central Region. I

The labor reserve economy supplied labor forthe mines and European commercial agricul­ture in South Africa and Rhodesia through theearly 1970s. Both the colonial administrationand the estate economy created markets formaize and other smallholder foodcrops andKydd and Christiansen assert that, because ofthe weakness ofthe estate economy, smallholdercash crop production was officially supportedfrom the 1920s to the I960s.

An overt colonial policy of promotingsmallholder cash crop production is also docu­mented by Kettlewell (1965). Before WorldWar I contributions of the colonial administra­tion consisted of limited specialist advice forthe plantation crops. After World War I, for

'Combined with high population densities, landalienation for estate production restricted land avail­able for smallholder cultivation. A village headman inChiradzulu is reported to have said in 1937: "As I look(0 Crown Land, I sec no land remaining where so manyof my people could go. I look in the air, I see I cannotfly there, then [ come to the point that I say what is theusc ofliving." (Public Record Office, London. CO 5251165. Minutes of the Chiradzulu District Native Asso­

ciation.)

51

"agricultural development" purposes, the ad­ministration expanded field services tosmallholder production of tobacco, cotton, andrice. The Nyasaland administration appears tohave devoted few resources to general hus­bandry until the Great Famine of 1948-49.Subsequently, various soil conservation meth­ods, early preparation and planting, and im­proved processing and preparation of crops formarket were recommended for small-scale cul­ti vators. Kettlewell describes the overall pol icyat independence as the pursuit of "gradual im­provement for the mass of cultivators" while"concentrating on the most progressive indi­viduals" (p.243). Of relevance to current ef­forts to promote fertilized hybrid maize is thefact that subsidized fertilizer was available to atleast a subset of farmers in Nyasaland from1952 onwards.2

Kettlewell refers particularly to maize whendiscussing the changes in pricing policy thatwere instituted after the Great Famine. In thefollowing season the Maize Control Board es­tablished formal control over maize marketingand set a guaranteed price at twice the previouslevel. In 1952, the Maize Control Board wasreconstituted as the Produce Marketing Boardand was authorized to purchase other crops andexport surpluses to Europe. The new policy ap­pears to have increased the importance of maizeas a cash crop, especially in the Central Region.Maize acreage increased and the marketed re­sponse for maize was so great that maize waseventually exported at unremunerative worldprices. From 1957 the administration pursued a

~In the CIMMYT/MOA survey, a number of theolder farmers claimed that they first used fertilizer in

the pre-independence period.

policy of minimum uniform pre-planting pricestied to world market prices, and price stabiliza­

tion through utilization of buffer stocks(Kandoole et aI., 1988). The o~iective was torestrict maize production to local needs plus afamine reserve, encouraging maize as a cashcrop only in t!lOSC areas most suitable for itsproduction. The marketed surplus nevcrthelessremained ex4ssive. The administration re­sponded by relax ing the monopoly of the mar­keting board tind reduced its role to that of aprocurement agent for government departments,commercial orders, industrial and urban mar­kets, and a strategic reserve. Maize was no longerproduced for export, although the famine reservecould be exported each year after the new har­vest was assured (Kettlewell, 1963; Kandoolcet aI., 1988; Brown, 1963).

B. The 1970s: Emphasis on EstateAgriculture

As in other post-independence' African econo­mies, the major features of Malawi's pricingsystem during the 1970s included governmentadministration of both input-output and pro­ducer-consumer price ratios through a market­ing board with a quasilegal monopsony andmonopoly over sales of agricultural inputs andcommodities. The policy concems of the GaMincluded regulating maize prices so that theywere high enough to stimulate producers butlow enough for urban consumers, taxing exportcrops for government revenues, subsidizingimports like fertilizer and fuel and, reflectingequity considerations, establishing a pan-terri­torial pricing regime (Kirchner et al., 1985).

Aggregate figures show high growth ratesfor estate production during the 1970s (lele,1989a; Kydd and Christiansen, 1982;Humphrey, 1975; Thomas, 1975). The estateshare of exports grew from 32 percent in 1967to 65 percent in 1979 and 80 percent in 1981­82 (Kandoole et al., 1988). The nation's im­pressive performance in ternlS of GDP growth

52

was offset by evidence of distributional prob­lems and declining real income of rural house­

holds (Kydd and Christiansen, 1982: Pryor,(990). Malawi's agricultural economy duringthe 1970s has been termed "dualistic," with thcmajority of farmers llsing low levels of tech­nology to produce for their own domestic mar­ket and a minority using higher levels of tech­nology to produce for the export market onestates. The prevailing view is that a consciousstrategy of promoting estate production to fuelthe national economy and generate much-neededrevenues divided agriculture into prosperingestates that were given preference in the pro­duction and sale of major export crops, andsmallholders producing mostly local maize forsubsistence.

Consequently, despite substantial invest­ments in the smallholder sector by donors andthe government, and Malawi's relatively supe­rior record in the implementation of rural de­velopment projects (Lele, 1989a; Sofranko andFliegel, 1989), marketed output of mostsmallholder crops and per caput maize produc­tion appear to have stagnated or declined dur­ing the 1970s (Lele, 1989b; Kydd andChristiansen, 1982.)3 Flagging internal effectivedemand and widening income differentials mayexplain why Malawi remained a net exporter ofmaize over the period. Lele writes that althoughan estate bias may have been essential in stimu­lating growth with limited national resources,"the quick resumption of overall growth inMalawi may now be constrained by the extremepoverty of most of its populace" (Lele, 1989b).

One of the major policy instruments of theGOM during the 1970s was the official market-

)The reported figures for cash crops appear fairlyInconclUSIve Estimated trends for smallholder outputare statIstIcally weak. Between competing cash cropssuch as tobacco and groundnuts, output responses toprice changes are expected to express the opposite signand the combined results arc usually ambiguous. Givenpopulation trends, per caput maize production docs notappear to have kept pace with population growth in thatperiod.

ing agency whose role expanded over time.After irjdependence the Farmer's MarketingBoard (;MB) was responsible for marketing,processing, disposing of agricultural products,subsidizing agricultural inputs, and providingadequate price stability in order to protect farm­ers ITon} world price fluctuations. [n 1971 theFM B became the Agricultural Development andMarketi ng Corporation (A DM A RC).ADMARC was responsible for the buying, stor­ing, processing and adapting for sale, distribut­ing, insl!ring, advertising, and transporting allproducts grown for sale on customary lands.ADMAr~C played a stabilization role by pro­viding storage facilities for food reserves and afood security role by transporting maize intodeficit areas during the hungry season. Thegovernment reimbursed ADMARC for the dif­ference between the prices charged to consum­ers and the cost recovery price, and sellingprices were nearly always below cost recoverylevels (Kandoole et aI., 1989).

ADMARC was also viewed as contributingto the country's broader development strategyby investing the difference between the pricereceived in the international markets for exportcrops and the price paid to producers, in estateexpansion (Christiansen and Stackhouse, 1989).Smallholder agriculture was implicitly taxedby maintaining a gap between producer pricespaid to smallholders and international pricesearned for their produce (Christiansen andSouthworth, 1988.).4 Estate producers, on theother hand, sold their products by auction onthe world market and remained largely untaxed

'Thc diffcrentlal between producer and interna­tional priccs for export crops consists of proccssingcharges, marketing costs, differences expressing ex­changc rate disequillbriulll, and the proportion held bymarketing agcnts abovc these costs. Smallholders havetendcd to recclvc a relatively small proportion of thcrevenucs obtaIned by I\DMI\RC from the final sales ofthei r output. ProduccrJi ntcrnational price ratios for to­bacco, cotton, and groundnuts, for examplc, have beenlower than similar ratios cited for Kenya and Tanzania(Lclc, 19X%).

53

(Lelc, 1989b). As part or its development roleADMARC also subsidizcd farm inputs, includ­ing seed and fcrtilizcr, which has been an im­portant aspect of the MOA Rural DevelopmcntPlan to encourage smallholdcrs to increase yieldsand release land for cash crop cultivation with­out violating self-sufficiency objectives.

With the restrictions imposed on Asian trad­ers during the 1970s and cross-border tradeopportunities limite~ by the weak economies ofneighboring countri;es, ADMARC increasinglydominated agricultural marketing and, in mostyears, effectively enjoyed monopsonistic/ mo­nopolist status (Christiansen and Southworth,1988.) Although African traders have been ex­empted from restrictions against trading pro­duce since 1')57, producer-consumer pricemargins have been so small that it has not beenworthwhile for large traders to participate(Kandoole et a!., 1988).

With maize5, however, the volume mar­keted is typically only a fraction of what isproduced. Unofficial markets exist alongsideofficial markets when the announced prices donot reflect true supply and demand conditions- as in the case of maize, most of which hasbeen marketed by small traders or through di­rect fanner-consumer or farmer-farmer trans­actions (Kandoole et a!., 1988). OneAgroeconomic Survey report states that al­though no quantitative data exists, private trad­ers are believed to have been responsible forthe bulk of inter-district and intra-district localmaize trade and have played an important rolein levelling deficits and surpluses (979). Themost probable result of the pricing policiesfollowed during the 1970s was an increase inmaize subsistence production relative to therroduction of smallholder export crops.

Kydd and Christiansen also argue that, as aconsequence of low cash crop prices forsmallholders during the 1970s, rates ofretum tolabor engaged in smallholder agriculture declined.

'Most of which was local maize or improved OPVs(synthetics and compositcs) in the 19705.

I

The labor-intensivity ofestate *riculture attractedlabor from small farms. The fxport of labor toSouth Africa and Rhodesia also:pcaked in the early1970s, and returning migrants were absorbed intothe growing estate sector as low-paid agriculturallabor rather than into smaIlhold~ragricul ture. Lele( 1989b) states that the shortage of land in thesmallholder sector (pariicularly in the South), dis­criminatory price and land policies, and the returnof migrants from Zimbabwe and South Africahave tended to increase wag~ employment andtenancy on the estates. Hirschmann and Vaughan(1983) cite inter-ccnsal figureS for the South andZomba that indicate a substantial shift of male la­bor into full or part-year wage employment andan increase in the proportion of individuals work­ing on their own holdings who were women.11

A socio-political interpretation of the ef­fects of strategies pursued by the GOM in the1970s is found in Hirschmann (1990).Hirschmann contends that the economic growthpolicies pursued by the GOM in the 1970sreduced the choices available to subsistencefarmers, their autonomy, their capacity to inno­vate, and their nutritional security. Smallholderoptions were limited by policies such asADMARC's monopoly over sales and purchaseof smallholder crops, di fferential rights to growand sell export crops, and the design of thecredit system, combined with the reduction inthe opportunities for migrant contract work.Pryor concludes that the increasing inequalityof income in Malawi from 1968 to 1986 re­flected growing di fferentials within the ruralsector rather than between the urban and rural

6Kadyampakeni offers a dissenting view by argu­ing that the estate and smallholder sectors are not dual

but symbiotic. The estates have been the greatest em­

ployers of wage labor in Malawi, providing part-time

jobs WIth supplementary income to smallholders.Kadyampakeni claims that disguised unemploymentexplains why men from the very early colonial periodsought work as migrant laborers. Since 1974 when theyhave worked for a shorter time at closer proximity, theyhave been able to contribute more substantially to pro­duction and livelihood on their small holdings.

S4

sectors. Although Malawi was one of the fewAfrican countries to consciously pursue a de­velopment strategy favoring the agriculturalsector, the government provided the greatestassistance to the richest smallholders and thepublic and private estates. According to Pryor,"it was the wager on the strong that .led towidening rural income di ffcrcntiation" ~p.30).

I

C. The 1980s: Renewed Emphasis onSmallholder Production

In 1979 Malawi's economy suffered from sev­eral "external shocks." These included the sec­ond oil price increase, decreased tobacco prices,a drought which resulted in food imports, andthe war in Mozambique which led to increasedinput transport costs. The combined result wasan increase in the current account deficit anddebt service ratio, requiring the nation to pro­duce a larger volume of exports to maintain itsreal income (Lele, 1989b).

[n response to these problems, the struc­tural adjustment programs conceptualized dur­ing the early 1980s and enforced by donors,emphasized the redress of account imbalancesthrough increasing output (rather than contract­ing demand), reforming trade policies, liberal­izing markets and reforming parastatals, andgradually moving toward border pricing. Ratherthan fiscal and monetary restraint, the loansextended financing to sustain both recurrentand development budgets. For example, the1981 SAL [ was designed to diversify foreignexchange sources and promote smallholder pro­duction for export. In addition to this objective,the 1983 SAL rr stipulated input price changessuch as phased removal of the fertilizer subsidyand shift toward high-analysis fertilizers (Sahnand Arulpragasam, 1991).7

'Largely because of increased transport costs, thesubsidy removal coincided with rapidly rising fertilizerprices. Unwilling to pass these costs on to smallholders,the GOM subsequently abandoned the program.

SAL 1-[11 had as conditions reforms in thestructure and operations of ADMARC. Finan­cial problems resulting from excessive staffing,costly external and domestic transport and, in1985, a liquidity crisis that resulted in AOMARCborrowing to finance crop purchases, also en­couraged the government to pursue market pri­vatization initiatives. The goal of the privatiza­tion program is for ADMARC to provide pricesupport by acting as a buyer of/ast resort untilthe private sector assumes a greater share ofmarketing activity. As a first step, farmers'clubs, traders, and other private sector opera­tors were asked to perform secondary market­ing operations by offering di fferential pricesbetween ADMARC's primary marketing fa­cilities and smaller, more isolated buying points(Christiansen and Stackhouse, 1989).

The current account and budget deficitsappear to have declined as a percent of GOP,but aggregate figures suggest that neither estatenor smallholder production demonstrated a sup­ply response (Lele, 1989b). Several authorshave therefore concluded that, although thepolicy emphasis on pricing changes have beennecessary, they are not sufficient to "resuscitatesmallholder production (Lele, 1989b; Sahn andArulpragasam, 1991; Christiansen andSouthworth, 1988). As pricing reforms haveprogressed, the importance of non-price factorsin stimulating smallholder foodcrop and exportproduction has again become evident to do­nors.

Sahn and Arulpragasam also question theefficacy of the pricing reforms. Output priceincreases for tobacco and maize were "erraticand reactive," and real prices of smallholdercrops such as maize, tobacco, groundnuts, andcotton actually declined. After the drought thegovernment raised the official maize producerprice by 68 percent in 1981/82, resulting in agrowing maize surplus and maize exports un­dertaken at a loss. Under the SALs in 1981 and1983, donors realigned smallholder producerprices away from maize toward groundnuts,tobacco, and cotton.

55

!

Further, they argue that the dejree of im­plicit taxation of smallholder expo crops fellbecause of falling world commo ity pricesrather than higher domestic prices. Consumerprices were subsidized, but the index of theofficial consumer price to the official ruralminimum wage fluctuated over the 1980s andwas generally high, eroding the purchasingpower of food-deficit rural households.

Whether open market access (market liber­alization) implies greater food secur~ty has alsobeen questioned (Sahn and Arulpragasam, 1991;Bowbrick, 1988; Lavers, 1988). Open marketretail prices are consistently higher thanAOMARC prices in any season, and expresswide seasonal variation. Poorer consumers tendto be farther from the ADMARC distributionsites and farm households that produce a smallvolume of maize output relative to their needsare less likely to benefit from interseasonalprice differentials. ADMARC outposts havedi fficulty maintaining supplies during the peakdemand period that precedes the maize harvest,causing the real cost of obtaining maize to risefar higher than the nominal price. In at leastsome zones, private traders are an insignificantsource of food supplies during the hungry sea­son because they sell their grain immediatelyafter purchase at harvest time (Kaluwa, 1990).From a small farmer's viewpoint, the most sa­lient result may be that he or she can no longersell harvested maize to meet immediate cashneeds, transfer the costs of maize storage toADMARC, and purchase the maize neededduring the hungry season (Agroeconomic Sur­vey, 1979; Bowbrick, 1988).8 For many farmhouseholds, ADMARC essentially provided acredit facility with better tem1S than can befound in their villages (Lavers, (988).

~ln the past, farmers could sell denty hybrid outputto ADMARC at harvest and purchase local maize whentheir local maize stocks were depleted, effectively trans­ferring the private costs of unstorable denty hybrids toADMARC and exchanging them for the preferred va­riety at a lower price than they would have to pay onlocal markets.

Christiansen and Stackhouse report that avariety of other administrative and logisticalproblems have plagued the privatization scheme.The possibility of large storage losses throughimproper handling of maize by traders and thequestion of ADM ARC's inability to purchaseenough maize to insure national food securityhave also been raised. They contend that thecost-trimming program of eliminating "redun­dant" markets on the criterion of volume ofsales is unsound from a development perspec­tive. A market may be low volume but servicean area that is developing its potential to pro­duce cash crops, and for which no other marketcatchment exists. A case in point might besome areas in the more sparsely populatedNorthern District where hybrid maize is a com­petitive cash crop for smallholders.

Most conclude that what is needed to stimu­late a marketed output response by smallholdersis a nexus of technical changes, adjustments inthe differential rights over the production andexport of tobacco, improvements in the cover­age of the credit system, control over diversionof land from smallholders, integration of prod­uct and input market markets, and a period ofprice and supply stabilization. Among all theseauthors, one of the major points used to demon­strate inefficacy of refom1s is that maize outputper caput has stagnated during the period ofprice reform except for the 1981/82 season inwhich maize prices were abruptly increasedrelati ve to other smallholder crop prices. In­creases in output have resulted from the expan­sion of maize into the last remaining cultivable

56

area rather than improved yields. They con­clude that only land-augmenting technicalchange (seed-fertilizer transformation) can gen­erate price-responsiveness in the aggregate sup~

ply curvc.Thc fundamental problcm with this percep­

tion is that, becausc maize is produced in ahousehold production process as the primarystarchy staple, an incrcase in price can lead toa decrease in marketed maize and a weI fareloss for households who are net buyers of maize(food-deficit households). To assure their sub­sistence requirements under conditions of pro­duction and marketing uncertainty, farmers mayactually sow a greater proportion of their farmarea in maize as the price increases. Even aftermarket liberalization, prices convey little rel­evant economic information for decision-mak­ing and farmers may not appear highly price­responsive even for hybrid maize which remainsmore of a cash crop than local maize (Smale,1992). The quality and reliability of rural mar­keting infrastructure, although high by sub­Saharan standards (Sofranko and Fliegel, 1989),affects aggregate response to price incentives(Sahn and Arulpragasam, 1991).

Limited changes in marketed maize outputor even in actual maize production per caputmay mask underlying technical changes withinregions and farm households. The second prob­lem with the findings reported in the citedsources is that, because of their emphasis onnational data, technical changes that have oc­curred in maize production are less perceptible.