agricultural extension - world...

2000

The World Bank

www.worldbank.org/html/oed Washington, D.C.

W O R L D B A N K O P E R A T I O N S E V A L U A T I O N D E P A R T M E N T

Madhur Gautam

Agricultural ExtensionThe Kenya Experience

An Impact Evaluation

Copyright © 2000

The International Bank for Reconstruction

and Development/THE WORLD BANK

1818 H Street, N.W.

Washington, D.C. 20433, U.S.A.

All rights reserved

Manufactured in the United States of America

First edition June 2000

The opinions expressed in this report do not necessarily represent the views of the World Bank or itsmember governments. The World Bank does not guarantee the accuracy of the data included in thispublication and accepts no responsibility whatsoever for any consequence of their use. Theboundaries, colors, denominations, and other information shown on any map in this volume do notimply on the part of the World Bank Group any judgment on the legal status of any territory or theendorsement or acceptance of such boundaries.

The material in this publication is copyrighted. The World Bank encourages dissemination of its workand will normally grant permission promptly. Permission to photocopy items for internal or personaluse, for the internal or personal use of specific clients, or for educational classroom use is granted bythe World Bank, provided that the appropriate fee is paid directly to the Copyright Clearance Center,Inc., 222 Rosewood Drive, Danvers, MA 01923, U.S.A., telephone 978-750-8400, fax 978-750-4470. Please contact the Copyright Clearance Center before photocopying items. For permission toreprint individual articles or chapters, please fax your request with complete information to theRepublication Department, Copyright Clearance Center, fax 978-750-4470.

All other queries on rights and licenses should be addressed to the Office of the Publisher, World Bank,at the address above, or faxed to 202-522-2422.

Photo credits: cover, page 10, and page 28, Curt Carnemark.

ISBN 0-8213-4758-6

Library of Congress Cataloging-in-Publication Data

Gautam, Madhur, 1960–

Agricultural extension : the Kenya experience / Madhur Gautam.

p. cm. — (Operations evaluation study)

Includes bibliographical references.

ISBN 0-8213-4758-6

1. Agricultural extension work—Kenya. I. Title. II. World Bank operations evaluation study.

S544.5.K4 G38 2000

630’.715’096862—dc21 00-032087

Printed on recycled paper.

C o n t e n t s

iii

v Acknowledgmentsvii Foreword, Prefacio, Préfacexi Executive Summary, Resumen, Résumé Analytiquexxi Abbreviations and Acronyms

1 1. Introduction2 Study Objective2 The National Extension Projects3 Study Strategy and Outline

7 2. Institutional Development7 Before8 And After10 Conclusions

13 3. Beneficiary Assessment13 Welfare and Productivity13 Access to Services14 Interactions Between Farmers and Agents14 The Relevance of Agents’ Advice15 Farmers’ Priorities15 Conclusions

17 4. Efficacy: The Quantity and Quality of Contact17 Program Design17 Access to Information18 The Contact Farmer Approach19 System Performance19 Biases20 Farmer Assessment of Extension Advice20 The Suspension of NEP II’s Disbursements20 Conclusions

23 5. Outcomes: Farmer Awareness and Adoption23 Awareness24 Adoption25 Statistical Tests of Extension’s Impact25 Conclusions

27 6. Results I: Farmer Efficiency and Productivity Change28 Relative Efficiency29 Productivity Change30 Conclusions

31 7. Results II: Production Effects of Extension31 A Fixed-Effects Approach32 Findings32 Conclusions

iv

A g r i c u l t u r a l E x t e n s i o n : T h e K e n y a E x p e r i e n c e

33 8. Client Focus: Farmer Valuation of Extension Benefits33 Desired Frequency of Visits34 Willingness to Pay34 Factors That Influence Willingness to Pay34 Conclusions

37 9. Conclusions and Lessons37 Lessons and Recommendations

Annexes39 Annex A. Background41 Annex B. The Design of NEP I and NEP II43 Annex C. Institutional Features47 Annex D. Focus Group Briefing on NEP’s Impact49 Annex E. Awareness and Adoption of Extension Messages51 Annex F. The Contingent Valuation Method and Its Application53 Annex G. Comments from the Government of Kenya57 Annex H. OED’s Response to Borrower Comments59 Annex I. Household Survey Data

61 Endnotes

65 Bibliography

Tables8 2.1 NEP I and II: Borrower Perspectives14 3.1 Changes in Farmers’ Access to Services and Service Quality

Figures3 Figure 1.1 Stylized Impact Model18 Figure 4.1 The Availability of Information:

Now, Relative to 10–15 Years Ago18 Figure 4.2 Normal Place and Frequency of Extension Agent–Contact

Farmer Meetings19 Figure 4.3 Extension Staff Allocation and Poverty by District20 Figure 4.4 Impact of the Suspension of NEP II Disbursements24 Figure 5.1 Farmers’ Awareness of Simple-to-Complex Messages, 1982–9725 Figure 5.2 Farmers’ Adoption of Extension Recommendations, 1982–9729 Figure 6.1 Productivity Change, 1982–97: Trends in Cluster Averages

Boxes39 Box A.1 The Difficulty of Drawing Inferences from Field Visits

v

AcknowledgmentsThis study presents the main findings of an evaluationof the impact of the World Bank’s support of projectsfor agricultural extension services in Kenya during1982–98. Madhur Gautam, Evaluation Officer, pre-pared this report; it was peer reviewed by GershonFeder, John Heath, and David Nielson. WilliamHurlbut provided editorial assistance. Helen Claverie,Constance Frye, Maisha Hyman, and Janet Wyseprovided administrative assistance at various stages ofthe evaluation. Detailed analyses of the Bank’s agricul-tural extension work in Kenya are contained in the sixworking papers prepared in support of this study (seethe Bibliography).

Madhur Gautam performed the overall analysis,with contributions from Jock Anderson in WorkingPaper 3, Jonathan Alevy in Working Paper 4, and AmyGautam in Working Paper 6. The Tegemeo Institute ofEgerton University undertook the household survey anddata processing under the overall supervision of GemArgwings-Kodhek. The team included supervisorsThomas Awuor and Francis Karin and assistant super-visors Betty Kanunga and Bridget Ochieng, whileJoshua Ariga performed the editing. Enumerators wereStephen Irungu, Joshua Kerima, Isaack Maritim, SalimMgallah, Geoffrey Oganga, Daniel Ouma, TonySisule, Milka Wangare, Alex Wanjohi, and LucyWanyanga. Willis Oluoch-Kosura of the University ofNairobi conducted the staff survey and data analysis,and Lutta Muhammad of the Kenya AgriculturalResearch Institute at the Katumani Research Stationcompiled the inventory of technology and extensionmessages. Allan Musamali and Martin Mutuku of theMinistry of Agriculture compiled the time series dataon extension services in the study districts. The AfricaTechnical Department (ATD) household survey data for1990 were graciously made available by Robert

Evenson, the Rural Household Budget Survey (RHBS)data for 1982 by Steven Block.

So many people in Kenya contributed to thisevaluation that it is not possible to acknowledge themall individually. The many Kenyan farmers whopatiently participated in the household survey and itspretests, and those subjected to the many field visitsover the course of this evaluation made invaluablecontributions. The support and assistance of the Minis-try of Agriculture, particularly of the extension staff atKilimo House, in the study districts and in otherdistricts visited were instrumental as the evaluationwas being conducted. Generous contributions weremade by many staff of the Kenya Agricultural ResearchInstitute and the International Research Centers inNairobi, as well as by members of the donor commu-nity and civil society in Kenya. Finally, the author isgrateful to his many colleagues in the resident missionand at the Bank’s headquarters for the support theyprovided.

This study was published in the Partnerships andKnowledge Group by the Dissemination and Outreachunit, directed by Elizabeth Campbell-Pagé (task teamleader), including Caroline McEuen (editor), KathyStrauss and Aichin Jones (graphics and layout), DianaQualls (editorial assistant), and Juicy Qureishi-Huq(administrative assistant).

Director-General, Operations Evaluation: Robert Picciotto

Director, Operations Evaluation Department: Elizabeth McAllister

Manager, Sector and Thematic Evaluations: Gregory Ingram

Task Manager: Madhur Gautam

vii

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

ISPREFACIO PRÉFACEFOREWORD

F o r e w o r d

The Training and Visit (T&V)system of management was intro-duced in Kenya by the World Bankin 1982 as a pilot, financed by theSecond Integrated AgriculturalDevelopment Project. It has sincebeen supported by the NationalExtension Project (NEP I), approvedin 1983, and the Second NationalExtension Project (NEP II), whichbecame effective in 1991.

The Bank projects had twoobjectives: institutional develop-ment of the extension service andsustained increases in agricul-tural productivity. The effective-ness of the extension approachadopted by the projects has beena subject of debate because of theperceived high cost and an appar-ent lack of impact on agriculturalproduction.

This debate has been part of abroader disagreement on the effec-tiveness of the T&V approach toextension. The arguments havefocused on its efficacy relative toalternative mechanisms for deliv-ering extension advice. While it isgenerally agreed that the T&Vsystem is costly, the controversycenters on its impact on agricul-tural production. Despite theintensity of the debate, however,there have been very few attemptsto rigorously establish the impactof T&V projects.

This evaluation adopted atheory-based approach to gather acredible body of empirical evi-dence to rigorously establish theimpact of the projects. Following aresults-based management frame-

En 1982, la Banque mondiale amis en place au Kenya le système degestion basé sur la formation et lesvisites. À ses débuts, ce projet piloteétait financé par le deuxième projet dedéveloppement agricole intégré.Depuis, il a bénéficié de l’aide du projetNEP I (approuvé en 1983) et du ProjetNEP II (mis en œuvre en 1991).

Les projets de la Banque avaientdeux objectifs : la consolidationinstitutionnelle des services dedéveloppement et l’augmentationsoutenue de la productivité agricole.L’efficacité des services dedéveloppement est cependantcontroversée. Certains trouvent qu’ilscoûtent fort cher pour le peu d’impactqu’ils ont sur la production agricole.

Cette controverse fait partie d’undébat plus large sur l’efficacité auKenya du système formation-visites.Les discussions portent largement surleur efficacité comparée à celle d’autressystèmes d’encadrement. Alors qu’ons’accorde généralement pour dire quele système formation-visites coûte cher,la controverse porte essentiellementsur leurs répercussions sur laproduction agricole. Malgré l’intensitéde la discussion, on ne s’est guèreefforcé d’évaluer avec rigueur l’impactdes projets de type formation-visites.

La présente évaluation a adoptéune approche théorique afin de réunirdes preuves scientifiques quidémontreront avec rigueur lesrépercussions des projets. En utilisantun cadre de gestion axé sur lesrésultats, on y évalue les principauxindicateurs des résultats à diversesétapes du projet afin de déterminer lerendement du système de

El sistema de gerencia que sebasa en la formación y las visitas (elsistema T&V) fue introducido enKenia por el Banco Mundial en 1982como práctico financiado por elSegundo Proyecto Integrado para elDesarrollo de la Agricultura. Desdeentonces, ha sido sostenido por elProyecto Nacional de Extensión (elNEP I), aprobado en 1983, y elSegundo Proyecto Nacional deExtensión (el NEP II), que entró enefecto en 1991.

Los proyectos del Banco tuvierondos objetivos: el desarrolloinstitucional del servicio de extensióny unos incrementos sostenidos de laproductividad agrícola. Se hadiscutido la eficacia del enfoque sobrela extensión adoptado por estosproyectos a causa de sus altos costospercibidos y su falta de tener un efectoapreciable en la producción agrícola.

Este debate forma parte de undesacuerdo más amplio cuyo sujeto esla eficacia del uso del sistema T&V enla extensión. Los argumentos se hanenfocado en la eficacia de este sistemarelativa a los mecanismos alternativospara entregar avisos de extensión.Mientras que todo el mundo está deacuerdo en que el sistema T&V escaro, la controversia se concentra ensu efecto en la producción agrícola.Sin embargo, a pesar de la vehemenciade este debate, pocos ensayos se hanhecho para rigurosamente determinarel efecto de los proyectos T&V.

Esta evaluación adoptó unenfoque basado en la teoría pararecoger una colleción convincente deevidencia empírica para rigurosamenteestablecer el efecto de estos proyectos.

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS


viii

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS

work, key indicators areevaluated at the variousstages of the results chain toassess the performance of theKenyan extension system.

The evaluation findsthat the projects have hadlimited institutional devel-

opment impact. They have hadsome beneficial impacts in in-creased geographical coverage,improved research-extension link-ages (albeit belated), andimproved staff quality throughtraining. Overall, however, thisevaluation found the currentextension system to be ineffectiveand inefficient in delivering theneeded services to farmers. Theinstitutional design has lacked afocus on farmer empowerment. Assuch, inappropriate incentiveshave resulted in a lack of account-ability or responsiveness to theclients’ needs. Most important,the system is not financiallysustainable.

A distinction needs to be main-tained between the relevance ofextension services per se and therelevance of the projects’ design.The rationale for providing exten-sion services is still relevant; how-ever, several features of theprojects’ design proved to be inap-propriate. The available evidencesuggests that the extensionapproach applied in Kenya was notefficacious. The physical impact ofthe extension services also cannotbe established with the currentdata. While it is likely that therewas a positive impact on farmerproductivity and efficiency in theinitial years of NEP I, the benefitsappear to have been short-lived, aseven the 1990 data do not indicateany significant impact. This evalu-ation thus could not establish a

développement kenyan.Les résultats de l’évaluation

indiquent que les projets ont eudes répercussions limitées sur laconsolidation institutionnelle.Par contre, ils ont eu des effetspositifs : ils ont étendu leurcouverture géographique,

resserré (bien que tardivement) lesliens entre le développement et larecherche, et amélioré la qualité dupersonnel par la formation. Mis à partces quelques points positifs,l’évaluation a conclu que les systèmesde développement actuels ne sont niefficaces ni rentables à offrir auxpaysans les services dont ils ontbesoin. Cependant, la conceptioninstitutionnelle, qui reflète les objectifsdes projets, ne tenait pas suffisammentcompte de la nécessité d’augmenterl’autonomie des paysans. De plus, desincitatifs inadaptés aux besoins ontengendré un manque deresponsabilisation et de sensibilisationaux besoins des clients. Enfin, cesystème n’est ni viable ni rentable.

Il ne faut pas confondre lapertinence des services dedéveloppement avec celle de laconception des projets. Par exemple,l’offre de services de développementaux petits exploitants du Kenya sejustifie toujours. Toutefois, lesdonnées recueillies donnent à penserque l’approche adoptée manqued’efficacité. Ces mêmes données nepermettent pas de déterminer lesrépercussions physiques des servicesde développement. Même s’il estprobable que le NEP I ait eu, à sesdébuts, des répercussions positives surla productivité et l’efficacité desfermiers, cela n’a pas duré. En effet, lesdonnées recueillies en 1990n’indiquent pas de progrès notables.La présente évaluation n’indiqueaucune répercussion appréciable surles résultats ni sur les indicateurs de

Siguiendo un estilo de gerenciaque se basa en los resultados, seevalúan los indicadores claves avarias etapas de la cadena deresultados para valorar elfuncionamiento del sistema deextensión en Kenia.

La evaluación halla que losproyectos tuvieron un efecto limitadoen el desarrollo de las instituciones.Tuvieron algunos efectos beneficialesen cuanto al alcance geográfico másamplio, los enlaces más fuertes entre laextensión y la investigación (aunquecon retraso), y la mejor calidad delpersonal gracias a la formación. Sinembargo, los resultados de estaevaluación indicaron que, enconjunto, el sistema de extensióncorriente es ineficaz e incapable deentregar a los campesinos los serviciosque necesitan. Le ha faltado al diseñoinstitucional un efoque en apoderar alos agricultores. Como tanto,incentivos inapropiados resultaron enuna carencia de contabilidad osensibilidad a las necesidades de losclientes. Más importante, el sistema nose sostiene financieramente.

Hay que hacer una claradiferenciación entre la relevancia de lossistemas de extensión de por sí y lapertinencia del diseño de losproyectos. Aunque todavía vale lapena ofrecer servicios de extensión,algunas características del diseño delos proyectos resultaron serinapropiadas. La evidencia disponiblesugiere que el enfoque en la extensiónque se llevaba a cabo en Kenia no fueefectivo. Tampoco se puede establecerel efecto físico de los servicios deextensión utilizando los datoscorrientes. Mientras es probable que elNEP I rindió algunos beneficiostempranos, no parecen haber duradomucho, y aún los datos de 1990 noindican ningún efecto significativo.Como consecuencia, esta evaluación

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS

ix

significant impact for eitherthe key outcome or the re-sults indicators; nor could itestablish a positive rate ofreturn to the expenditures onagricultural extension.

Five main lessons emergedfrom this evaluation. First, there is aneed for more efficient targeting ofextension services to focus on groupsand areas where the marginal impactis likely to be the greatest. This, inturn, calls for a more flexible and a“smart” system that can identify gapsbetween average and best practices,and allocate scarce resources morerationally.

Second, to guide the “smart”system and to target better, it isnecessary to have timely flows ofrelevant management informationand continuous evaluation to pro-vide in-time feedback. Hence, thereis a need for a reliable monitoringand evaluation system.

Third, commensurate with thepace of technology generation andthe demand for advice, the inten-sity of the extension service needsto be suited for particular circum-stances. In some areas, a leanerand less-intensive presence (inextension staff per farm house-hold) would allow a widergeographical coverage and maybe more cost-effective. This callsfor a more demand-driven andresponsive delivery system.

Fourth, a blanket approachusing a uniform methodology inall circumstances is also unlikelyto be effective. A more pluralisticapproach that exploits the syner-gies among the extension service,low-cost modern communications,demonstrations, printed media,and partnerships with civil societyand the private sector needs to bedeveloped.

F o r e w o r d

résultats. Elle n’a pas non plusengendré un taux de rendementpositif dans les dépenses dedéveloppement agricole.

Cinq grandes leçonsressortent de cette évaluation.

Premièrement, il estnécessaire de mieux cibler les zones etles groupes où l’impact marginal estsusceptible d’être le plus important.D’où le besoin d’un système plussouple et plus « ingénieux », capabled’identifier le fossé qui sépare lesmeilleures pratiques de cellescourantes et d’allouer de façonrationnelle les ressources limitées.

Deuxièmement, pour bienorienter le système « ingénieux » etmieux cibler l’action, les services dedéveloppement doivent disposerrégulièrement d’information degestion pertinente et procéder à uneévaluation continue pour fournir de larétroaction au moment opportun. Ilest donc indispensable de mettre enplace un système d’évaluation et decontrôle fiable.

Troisièmement, le service dedéveloppement doit s’adapter auxcirconstances particulières en tenantcompte du rythme du développementtechnologique et de la demande deconseils. Dans certaines régions, uneprésence réduite et modérée (nombred’agents de développement par ferme)étendrait la couverture géographiqueet peut-être la rentabilité. Celanécessite un système mieux adapté etaxé sur la demande.

Quatrièmement, une approchegénérale qui utilisera la mêmeméthodologie quelles que soient lescirconstances a peu de chance d’êtreefficace. Une approche imprégnée depluralisme pour exploiter les synergiesqui existent entre les services dedéveloppement, les moyens decommunication modernes peucoûteux, l’organisation de

no podía establecer ningúnefecto significativo para losindicadores claves del desenlaceo de los resultados, ni tampocoera capaz de establecer una tasapositiva de rendimiento conrespeto a los gastos para laextensión agrícola.

Cinco lecciones principales seaprendieron de esta evaluación. Laprimera lección es la necesidad deseleccionar más eficientemente lasáreas y los grupos para los cuales elimpacto marginal de los servicios deextensión va a ser el más importante.Para esto, se necesita un sistema másflexible e “inteligente” que puedaidentificar las diferencias entre lasmejores prácticas y las prácticasmedias y asignar los recursos escasosde una manera más racional.

Segundo, para guiar el sistema“inteligente” hacia una selección máseficiente, se necesita un flujo oportunode datos relevantes para la gerencia yuna evaluación contínua que puedaproveer información de vuelta enbuena hora. Para esto es necesario unsistema fiable de monitoreo yevaluación.

Tercero, proporcionado con elpaso de la generación de la tecnologíay la demanda de avisos, la intensidaddel servicio de extensión tiene queadaptarse a las circunstanciasparticulares. En algunas áreas, unapresencia menos fuerte e intensa (entérminos del número de personal deextensión por cada granja) permitiríaun alcance geográfico más amplio ypuede ser más rentable. Esto exigiríaun sistema de entrega más sensitivo yempujado por la demanda.

Cuarto, un enfoque general quese fija de una metodología uniformepara todas las situaciones tampoco vaa ser efectivo. Hay que desarrollar unenfoque de naturaleza más pluralistaque se aprovecha de las sinergias entre

x


Robert Picciotto

Director-General, Operations Evaluation Department

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

ISAnd finally, the service

delivery mechanism needs tofully incorporate client focus.This requires consideringalternative options such ascost-sharing, farmer organi-zations, decentralization, and

the like as an integral part of thedelivery mechanism.

démonstrations, le recours auxjournaux et la création departenariats avec la société civileet le secteur privé doit être miseau point.

Cinquièmement, lemécanisme d’offre de services

doit intégrer des moyens pour lespaysans de se faire entendre. Celaexige d’envisager d’autres optionscomme le partage des coûts,l’incitation à l’établissementd’organismes de paysans et ladécentralisation

el servicio de extensión, lascomunicaciones modernas debajo costo, las demostraciones,la prensa, y las asociaciones conla sociedad civil y el sectorprivado.

Finalmente, el mecanismopara entregar los servicios tiene

que incorporar un enfoque en elcliente. Esto requiere la consideraciónde alternativos tales como larepartición de costos, el apoyo a losorganismos que representan a loscampesinos, y la decentralización,entre otros, como parte integral delmecanismo de entrega.

xi

RESUMEN RÉSUMÉANALYTIQUE

EXECUTIVESUMMARY

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS

E x e c u t i v e S u m m a r y

Kenya’s extension servicesadopted the training and visit (T&V)management system in 1982, and theWorld Bank supported the systemthrough the first and second NationalExtension Projects, NEP I and NEP II.The projects had two objectives: theinstitutional development of theextension services and sustainedincreases in agricultural productivity.

The effectiveness of the exten-sion services has been subject to

debate, however, because of its per-ceived high costs and apparent lackof impact on agricultural produc-tion. This debate is part of a broaderdisagreement about the effectivenessof the T&V approach. Argumentshave focused largely on its efficacyrelative to alternative mechanismsfor delivering extension advice.While it is generally agreed that theT&V system is costly, the contro-

Les services d’extension duKenya ont adopté, en 1982, unsystème de gestion basé sur laformation et les visites. La Banquemondiale a appuyé ce nouveausystème dans le cadre des deuxpremiers projets, NEP I et NEP II, quivisent la consolidation institutionnelledes services d’extension etl’augmentation soutenue de laproductivité agricole.

L’efficacité des services d’extensionest cependant controversée. Certainstrouvent qu’ils coûtent fort cher etqu’ils ont peu d’impact sur laproduction agricole. Ce débat faitpartie d’un différend plus large surl’efficacité au Kenya des systèmesformation–visites. Les discussionsportent largement sur leur efficacitécomparée à celle d’autres systèmesd’encadrement. Alors qu’on convientgénéralement que le systèmeformation–visites coûte cher, lesretombées des gros investissementsdes pays emprunteurs et, parconséquent, leurs répercussions sur laproduction agricole animent lacontroverse. Malgré l’intensité de ladiscussion et les prêts importants de laBanque, on ne s’est guère efforcéd’évaluer avec rigueur l’impact desprojets basés sur la formation et lesvisites.

En adoptant une approchethéorique, cette étude a réuni de façonméthodique des renseignements quiont permis de déterminer l’impactprobable des projets. En utilisant uncadre de gestion axé sur les résultats,l’étude a essayé de relier les résultatsobservés dans les champs des fermiers

En 1982, los servicios deextensión de Kenia adoptaron unsistema de gerencia basado en laformación y las visitas (“training andvisit” o “T&V”), y el Banco apoyó aeste sistema en el transcurso de losproyectos nacionales de extensión, elNEP I y el NEP II. Estos proyectostuvieron dos objetivos: el desarrolloinstitucional de los servicios deextensión y unos incrementossostenidos de la productividad agrícola.

Sin embargo, se ha discutido laeficacia de los servicios de extensiónporque existe la percepción de quecuestan mucho y tienen poco efecto en laproducción agrícola. Este debate formaparte de un desacuerdo más ampliocuyo sujeto es la eficacia del uso delsistema T&V en Kenia. Los argumentosse han enfocado sobre todo en laeficacia de este sistema relativa a losmecanismos alternativos para entregar

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS


xii

versy centers on returns toborrowing countries’ high lev-els of investment, and henceon its impact on agriculturalproduction. Despite the inten-sity of the debate and the highvolume of the Bank’s lending,

very few attempts have been made torigorously determine the impact ofT&V projects.

Adopting a theory-basedapproach, this study systemati-cally gathered a credible body ofevidence to establish the likelyimpact of the projects. Followinga results-based managementframework, the study sought torelate the results observed in thefarmers’ fields to the projects’inputs. In addition, intermediateoutput and outcome indicatorswere measured in order to assessthe performance of the extensionsystem along the results chain toconfirm the potential for impact.

Principal FindingsInstitutional DevelopmentInstitutional development under NEPI and NEP II was limited. NEP Iintroduced T&V as “first and fore-most a management system.” After17 years, however, the effectivenessof extension services does not appearto have improved appreciably. Astrategic vision for the future develop-ment of the extension system is lack-ing, and management continues to beweak, with virtually nonexistent in-formation systems. Inadequate finan-cial management disrupts the timelyflow of operational funds. The pro-longed ineffectiveness of the exten-sion services recently led Kenya, withthe help of the Bank and other donors,to try to rationalize them with alter-native approaches.

The projects, having establisheda national system based on the T&V

aux intrants des projets. Deplus, des indicateursintermédiaires de la productionet des résultats ont été mesurésafin d’évaluer le rendement dusystème d’extension parrapport aux résultats.

Constatations principalesConsolidation institutionnelleLes projets NEP I et NEP II ont aboutià une consolidation institutionnellelimitée. Le projet NEP I a introduit leconcept formation–visites commeétant « d’abord et avant tout unsystème de gestion ». Dix-sept ansplus tard, l’efficacité de ces servicesd’extension ne semblent pas avoir faitde progrès notables. Il leur manqueune vision stratégique pour l’avenir, etleur gestion demeure insuffisante etpresque dépourvue de systèmesd’information. Une mauvaise gestionfinancière empêche les capitauxd’exploitation d’être disponibles àtemps. L’inefficacité persistante desservices d’extension pousse le Kenya,appuyé par la Banque et d’autresbailleurs de fonds, à tenter de lesrationaliser en adoptant d’autresméthodes.

Une fois lancés à l’échellenationale grâce au système de gestionformation–visites, les projets ontétendu leur couverture géographique,resserré (bien que tardivement) lesliens entre l’extension et la recherche,et amélioré la qualité du personnel parla formation. Cependant, laconception institutionnelle, qui reflèteles objectifs des projets, ne tenait passuffisamment compte de la nécessitéd’augmenter l’autonomie des paysans.De plus, des incitatifs inadaptés auxbesoins ont engendré un manque deresponsabilisation et de sensibilisationaux besoins des clients. Enfin, lastructure hiérarchique de ces projetsdécourageait l’innovation, le

avisos de extensión. Mientras quetodo el mundo está de acuerdo enque el sistema T&V es caro, lacontroversia se concentra en elrendimiento de las altas tasas deinversión de los países que pidenpréstamos y así en su efecto en laproducción agrícola. A pesar de

la vehemencia del debate y el granvolumen de los préstamos concedidospor el Banco, pocos ensayos se hanhecho para rigurosamente determinar elefecto de los proyectos T&V.

Adoptando un enfoque basadoen la teoría, este estudiosistemáticamente recogió una colleciónde evidencia empírica para establecer elefecto probable de estos proyectos.Siguiendo un estilo de gerencia que sebasa en los resultados, el estudiointentó relacionar los resultadosobservados en los campos de losagricultores a los ingresos de losproyectos. Además, se midieron losindicadores intermedios de productosy resultados para evaluar elfuncionamiento del sistema deextensión a lo largo de la cadena deresultados para confirmar el potencialdel sistema de tener un impacto.

Conclusiones principalesDesarrollo institucionalEl desarrollo institucional bajo losNEP I y II fue limitado. El NEP Iintrodujo el sistema T&V como “unsistema de gerencia ante todo.” Sinembargo, después de 17 años, laeficacia de los servicios de extensiónno parece haberse mejoradoapreciablemente. Se falta una visiónestratégica para el desarrollo futurodel sistema de extensión, y la gestióncontinúa a ser débil, con una carenciacasi total de sistemas de información.La ineficiencia prolongada de losservicios de extensión recientementemotivó a Kenia, con la ayuda delBanco y de otros donantes, de tratar

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS

xiii


system of management, didincrease geographical cover-age, improve the linksbetween extension and re-search (although belatedly),and raise staff quality throughtraining. But the institutional

design, reflecting the projects’ objec-tives, lacked a focus on the criticalissue of empowering farmers, withinappropriate incentives leading to alack of accountability or responsive-ness to the clients’ needs. Further-more, the projects’ hierarchical struc-ture yielded disincentives forinnovation, partnerships, and effi-ciency, and the top-down extensionsystem has been supply-driven andnonparticipatory.

SustainabilityThe system is neither financiallysustainable nor cost-effective. It issignificantly more costly, and nomore efficient, than the system itreplaced. The government’s alloca-tions for extension, as for otherpublic expenditures, continue todecline, leaving the system heavilydependent on donor funding.

An overwhelming proportion(80 percent) of the operational bud-get is consumed by staff salaries. Asa result, many of the problems thatlimited the effectiveness of the pre-vious system have persisted, andstaff have reverted to the methods ofdissemination used earlier.

The projects’ approach of high-intensity contact with a limitednumber of farmers has been costlyand unwarranted, given the inad-equate stock of messages for dis-semination and the slow pace ingenerating new technology.

RelevanceA distinction is needed between therelevance of extension services and

partenariat et l’efficacité, et lemodèle d’extension à partir dusommet était axé sur l’offre etnon sur la participation.

ViabilitéCe système n’est ni viable ni

rentable. Il est beaucoup plus coûteuxet pas plus efficace que le système qu’ila remplacé. L’État continue de réduireles fonds qu’il alloue à l’extension, demême que ses autres dépenses, rendantainsi le système très dépendant desbailleurs de fonds.

Les salaires du personnelabsorbent la majeure partie (80 %) dubudget de fonctionnement. Parconséquent, nombre de problèmes quilimitaient l’efficacité du systèmeprécédent subsistent, et le personnel enest revenu aux anciennes méthodes dediffusion de l’information.

Les contacts fréquents avec unnombre limité de paysans coûtent cheret ne sont pas justifiés, car lesmessages à transmettre sont peunombreux, et le rythme d’introductiondes nouvelles technologies, lent.

PertinenceIl ne faut pas confondre la pertinencedes services d’extension avec celle de laconception des projets. Par exemple,l’offre de services d’extension auxpetits exploitants du Kenya se justifietoujours. Par contre, plusieurs aspectsde la conception des projets se sontrévélés inadéquats. Les paysansapprécient l’accès à ces services aupoint d’être disposés à payer pour,mais la demande n’est pas satisfaite.Les solutions de rechange aux servicespublics d’extension sont rares, bienqu’une offre se dessine de ce côté.

La plupart des paysans, y comprisceux qui assurent la liaison avec lesservices d’extension, ne veulent pasrencontrer les agents trop souvent, cequi porte à s’interroger sur la

de racionalizarlos utilizandoenfoques alternativos.

Después de establecer unsistema nacional basado en elsistema de gerencia T&V, losproyectos sí aumentaron elalcance geográfico, fortalecieronlos enlaces entre la extensión y la

investigación (aunque con retraso), yperfeccionaron al personal por medio dela formación. Pero al diseñoinstitucional, que reflejaba los objetivosde los proyectos, le faltaba el enfoque enel asunto crítico de apoderar a loscampesinos; además, los proyectosproporcionaban incentivosinapropiados que condujeron a unafalta de contabilidad o sensibilidad a lasnecesidades de los clientes. Además, laestructura jerárquica de los proyectosno favoreció a las asociaciones, lainovación, o la eficiencia, y el sistema deextensión, que funciona de arriba enabajo, ha sido impulsado por elaprovisionamiento y no ha sidoparticipatorio.

SostenibilidadEl sistema no es ni financieramentesostenible ni rentable. Essignificativamente más caro, pero sinser más eficiente, que el sistema quereemplazó. Las afijaciones delgobierno para la extensión, como laspara otros gastos públicos, continúana ir en declive, dejando el sistemaexcesivamente dependiente delfinanciamiento por donantes.

Un porcentaje asombrante (unos80%) del presupuesto operacional esconsumido por los sueldos del personal.Como resultado, muchos de losproblemas que limitaron la eficacia delsistema anterior han persistido, y elpersonal ha revertido a los métodos dediseminación que se empleaban antes.

La estrategia de contactosintensos y frecuentes con un númerolimitado de campesinos que

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS


xiv

the relevance of project de-sign. While providing exten-sion services to smallholdersstill has a rationale in Kenya,several features of theprojects’ design have provedinappropriate. The farmers

value access to extension servicesenough to be willing to pay forthem, but their demand is unmet.The alternatives to government-provided extension are few atpresent, although some providersare emerging.

Most farmers, including the con-tact farmers, do not want to meetextension agents often, which ques-tions the relevance of a biweekly—oreven monthly—visit schedule. Eventhe staff are wary of the repetitivenessand ineffectiveness of the visits.

Blanket coverage of the major-ity of the production areas with asingle approach and standard mes-sages has proven inefficient andunproductive. Limited experiencefrom pilot initiatives under NEP IIconfirms the potential usefulness ofalternative and more responsiveapproaches.

EfficacyThe farmers did not have adequateaccess to extension advice in 1982,and it appears that they do not haveit now. All methods of applying theavailable data indicate that thecurrent institutional arrangementsfor delivering services are ineffec-tive. This is partially a reflection ofthe poor enabling environment inwhich the extension services haveoperated in recent years. The poorperformance, however, is notentirely a function of these externalfactors, as evidence suggests thatthe current situation differs littlefrom that of 1990.

Both qualitative and quantita-

pertinence de rencontresbimensuelles, voire mensuelles.Même les agents des servicesremettent en question lecaractère répétitif et l’inefficacitéde leurs visites.

L’utilisation pour presquetoutes les zones de production d’uneapproche générale et de messagespasse-partout se révèle inefficace etimproductive. Les quelquesexpériences tirées d’initiatives pilotesdu projet NEP II confirment lanécessité d’adapter les actions auxbesoins.

EfficacitéEn 1982, les paysans ne pouvaient pasaccéder aux services d’extensioncomme ils l’auraient dû, et c’est encorele cas aujourd’hui. Toutes les donnéesaccessibles démontrent l’inefficacitédes dispositions institutionnellesactuelles pour fournir des services auxpaysans. Cela reflète en partie lecontexte peu favorable dans lequel lesservices d’extension ont dûfonctionner ces dernières années.Toutefois, leurs piètres résultats nedécoulent pas uniquement de facteursexternes, puisque tout indique que lasituation a très peu changé depuis1990.

Les évaluations qualitatives etquantitatives indiquent que le systèmed’extension ne fournit pas auxpaysans le type d’information qu’ilsveulent. Elles montrent aussi que lesconseils prodigués à nombre d’entreeux sont discutables. Les servicesd’extension continuent de privilégier ladiffusion de messages agronomiquessimples portant notamment sur lemaïs. Leurs activités dans le cadre desprojets NEP I et NEP II n’encouragentguère la prise de conscience despaysans et l’adoption par ceux-ci desrecommandations. L’incapacité à leuroffrir des conseils plus élaborés et

practicaban los proyectos hasido cara e injustificada, dado elinventario inadecuado demensajes para diseminar y elpaso lento de la creación denueva tecnología.

PertinenciaHay que distinguir entre la relevanciade los servicos de extensión y lapertinencia del diseño de losproyectos. Aunque es todavía lógicoofrecer servicios de extensión a losminifundistas en Kenia, se ha probadola impropriedad de algunascaracterísticas del diseño de losproyectos. Tanto valen los graneros elacceso a los servicios de extensión queestán dispuestos a subvenirlos, perono se satisface la demanda.Actualmente, hay pocos alternativos ala extensión proporcionada por elgobierno, aunque algunosproveedores alternativos estáncomenzando a aparecer.

Visto que la mayoría de losagricultores, incluso los campesinos quese contactaron, no quieren reunirse conlos agentes de extensión muy a menudo,la relevancia de un programa de visitasque tienen lugar cada dos semanas, oaún cada mes, es dudosa. Aún elpersonal se da cuenta de la repetitividady la ineficiencia de estas visitas.

La tentativa de proveer un alcancegeneral para la mayoría de las areas deproducción utilizando un sóloenfoque y mensajes convencionales haresultado ser ineficaz e improductiva.La experiencia limitada de lasiniciativas pilotas del NEP II confirmala utilidad potencial de enfoquesalternativos más sensibles.

EficaciaLos campesinos no tuvieron accesoadecuado a los avisos de la extensión en1982, y parece que todavía no lo tienen.Todos los métodos de aplicar los datos

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS

xv


tive assessments indicatethat the extension system hasnot responded with the kindsof information the farmerswant, and the relevance ofthe advice it offers to abroad range of farmers is

questionable. The focus of the ex-tension services has remained ondisseminating simple agronomicand maize-related messages. Ex-tension activities under NEP I andNEP II had little influence on theevolution of patterns of awarenessand the farmers’ adoption of rec-ommendations. The failure to de-liver more advanced and context-specific advice reduced thecost-effectiveness of the main fea-ture of the projects’ design—that is,face-to-face extension. This wasparticularly true for NEP II. Whenthis project began, most farmerswere known to have alreadyadopted the simpler messagesabout maize.

A significant finding is that avery large proportion of those whoare aware of the messages on even themore complex practices have adoptedthem. Thus, although factors notrelated to extension, including a fre-quently cited lack of financialresources or access to credit, may beimportant in preventing farmers fromadopting certain complex practices(for example, using fertilizers andpesticides), it is evident that the lackof information continues to be animportant constraint.

Access to existing services islimited, especially for the poor andthe less educated, despite the overallincrease in coverage under NEP Iand II. Most farmers also report thatthe availability of information andthe quality of services have declinedsince the early 1980s.

Progress toward gender equity

adaptés à leur environnementréduit la rentabilité des projets,notamment des rencontresindividuelles avec les paysans.C’est particulièrement vrai pourle projet NEP II. Quand ceprogramme a été lancé, on

savait que la plupart des paysansavaient déjà assimilé les messages lesplus simples concernant le maïs.

Constatation significative, la trèsgrande majorité des paysans qui ontreçu des messages, même sur lesméthodes plus complexes, les ontadoptées. Bien que des facteursétrangers aux services d’extension,comme le manque de ressourcesfinancières ou l’incapacité d’obtenirdes crédits, fréquemment mentionnéspar les paysans, puissent les empêcherd’adopter certaines méthodescomplexes (p. ex. l’emploi d’engrais etde pesticides), il va de soi que lemanque d’information reste unobstacle important.

Malgré les progrès réalisés en cequi concerne l’accès aux services, cedernier demeure limité, surtout pourles paysans pauvres et peu éduqués.Les paysans affirment quel’information est plus rare et que lesservices sont de moindre qualitédepuis le début des années 80.

Les progrès pour l’égalité dessexes sont mitigés. Les préjugés dontétaient victimes les paysannes autrefoisont disparu, mais certains préjugésdemeurent quand vient le moment dechoisir des contacts dans lapopulation paysanne. La proportionde femmes parmi les agentsd’extension sur le terrain estsensiblement la même depuis 1982.

La pénétration des projets restebien inférieure aux niveaux prévus aumoment de leur conception, carseulement 7 % des paysans qui y sontassociés et 2 % de l’ensemble despaysans rencontrent régulièrement les

disponibles indican que losarreglos institucionales corrientespara suministrar estos serviciosson inefectivos. Esto es en parteun reflejo de las pocasposibilidades de apoderarse queexisten en el ambiente en que losservicios de extensión han

operado en años recientes. Sin embargo,este pobre funcionamiento no se debetotalmente a estos factores externos,porque la evidencia sugiere que lasituación presente no es muy diferentede la de 1990.

Ambas las evaluaciones cualitativasy cuantitativas indican que el sistema deextensión no ha respondido con el tipode información que quieren loscampesinos, y la pertinencia de losavisos que se les ofrecen a un rangoamplio de agricultores es dudosa. Losservicios de extensión continúan aenfocarse en diseminar mensajes simplesy agronómicos que muchas veces tratandel maíz. Las actividades de extensiónbajo los NEP I y II tuvieron pocainfluencia en la evolución de los modelosde conocimiento y la adoptación por loscampesinos de sus recomendaciones. Suinhabilidad de ofrecer avisos másavanzados y específicos dentro delcontexto redujo la rentabilidad de lacaracterística principal del diseño delproyecto¾es decir, la extensión cara acara¾sobre todo en el caso del NEP II.Cuando se inició este proyecto, se sabíaque la mayoría de los campesinos yahabía adoptado los mensajes mássencillos sobre el maíz.

Un hallazgo importante es que unaproporción muy grande de loscampesinos que se han dado cuenta delos mensajes, o aún de las prácticas máscomplejas, los ha adoptado. Así, aunquefactores que no tienen nada que ver conla extensión, tales como lafrecuentemente citada falta de recursosfinancieros o acceso al crédito, puedenser importantes en impedir a los

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS


xvi

has been mixed. The earlierbias against women farmershas been rectified, but somebias persists in the selection ofcontact farmers. The propor-tion of women field-extensionagents has remained largely

unchanged since 1982.Outreach is well below the levels

anticipated in the project design—only about 7 percent of the contactfarmers and 2 percent of all farmersmeet regularly with extension agentsin settings envisioned in the design.

EfficiencyOverall, a positive rate of return onexpenditures for extension cannot beestablished. It is likely that NEP Iyielded some early benefits, but theyappear to have been short-lived, asthe data do not indicate any signifi-cant impact, even by 1990.

Farmers’ technical efficiencyhas improved somewhat since1982, but overall efficiency con-tinues to be low. According to thedata, extension services have hada very small positive impact onthe level of technical efficiency,but the level of confidence in theresult is low. The data also indi-cate that extension services havehad no discernible impact on thelevel of economic efficiency.Farmers’ low economic efficiencysuggests the potential for realizingsignificant savings by simplymoving to a more economical mixof inputs under current marketconditions.

The current data do not indi-cate a significant impact of thesupply of extension services on pro-ductivity at the farm level. What thedata do indicate is that extensionresources have been allocated ineffi-ciently and poorly targeted. Growthin agricultural production has been

agents d’extension au cours dessessions prévues dans le cadredes projets.

RentabilitéDans l’ensemble, on ne peutétablir un taux de rendement

positif pour les dépenses courantesdes projets d’extension. Il se peut quele projet NEP I ait récolté quelquesprofits au début, mais cela n’a pasduré. Les données ne révèlent aucunimpact significatif, même depuis 1990.

L’efficacité technique des paysanss’est quelque peu améliorée depuis1982, mais le rendement d’ensemblereste faible. Selon les données, lesservices d’extension ont eu un trèsléger effet positif sur l’efficacitétechnique, mais le degré de confianceest faible. D’après ces mêmes données,les services d’extension n’ont euaucune influence visible sur l’efficacitééconomique. Le faible rendementéconomique des paysans laissesupposer qu’il suffirait d’adopter unemeilleure combinaison d’intrants pourréaliser des économies considérablesdans les conditions actuelles dumarché.

Les données ne permettent pas dedire si l’offre des services d’extension aeu une influence significative sur laproductivité des fermes. Il estcependant possible de soulignerl’inefficacité de l’allocation desressources de ces services et le mauvaischoix de leurs objectifs. La productionagricole affiche une croissance plusforte dans les zones jusque là moinsproductives, tandis que la mise enplace d’agents d’extension favorise leszones plus productives. Les servicesd’extension ont peut-être aidé àtransmettre des messagestechnologiques simples dans les zonesauparavant mal desservies et moinsproductives, mais les données actuellesne permettent pas de l’affirmer.

campesinos de adoptar ciertasprácticas complejas (por ejemplo,el uso de los fertilizantes y laspesticidas), es evidente que la faltade información continúa a seruna limitación importante.

El acceso a los serviciosexistentes es limitado, sobre

todo para los pobres y losineducados, aunque su alcance hacrecido. Los agricultores informan quela disponibilidad de información y lacalidad de los servicios son peores quea principios de los años ochenta.

El progreso hacia la igualdad delos géneros no ha sido totalmentepositivo. El perjuicio anterior contralas campesinas se ha remendado, peropersiste este perjuicio en la selección delos agricultores que se contactan. Laproporción de agentes de extensión enel campo que son mujeres hacambiado poco desde 1982.

El alcance es bastante debajo delos niveles anticipados en el diseño delproyecto: sólo unos 7% de loscampesinos contactados y unos 2% detodos los agricultores se reunenregularmente con los agentes deextensión bajo las condiciones que seconceptualizaron en el diseño.

EficienciaEn conjunto, no se puede estableceruna tasa positiva de rendimiento encuanto a los gastos para la extensión.Es probable que el NEP I rindióalgunos beneficios tempranos, peroparece que no duraron mucho, vistoque los datos no indican ningún efectosignificativo aún en 1990.

La eficiencia técnica de loscampesinos se ha mejorado algo desde1982, pero su eficiencia en generalcontinúa a ser baja. Según los datos, losservicios de extensión han tenido unefecto positivo muy pequeño en el nivelde eficiencia técnica, pero hay un bajonivel de confianza en los resultados.

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS

xvii


higher in the previously less-productive areas, while theplacement of extension staffhas favored the more produc-tive areas. Extension mighthave helped spread simpletechnological messages to the

formerly underserved and less-pro-ductive areas, but this cannot beconclusively determined with thecurrent data.

A significant proportion offarmers is willing to pay for exten-sion services, which indicates thatthey value the advice they receive.But the perceived value, as revealedby what the farmers are willing topay, is well below what the govern-ment currently spends on extensionservices per farm.

Principal RecommendationsThe lessons emerging from thisstudy that can be applied to thedesign of future agricultural exten-sion projects suggest the followingrecommendations.

TargetingThe first lesson is the need for moreefficient targeting of extensionservices to areas and groups wherethe marginal impact is likely to bethe greatest. This calls for a moreflexible, “smart” system that canidentify the gaps between best prac-tices and average practices andallocate scarce resources more ra-tionally. Further, the farmers whoare selected for interaction shouldbe more representative of the localsocioeconomic environment, whichwill ensure the delivery of morerelevant advice to the various cat-egories of farmers.

Information SystemsTargeting calls for appropriate flowsof timely and reliable information,

Une grande proportion depaysans est prête à payer pourles services d’extension,montrant ainsi leur intérêt pourles conseils qui leur sontprodigués. Mais la valeur qu’ilsy attachent, comme l’indique les

montants qu’ils sont prêts à payer,reste cependant bien inférieure à quel’État investit par ferme dans lesservices d’extension.

Recommandations principalesLes leçons que l’on peut tirer de laprésente étude, et dont on devrait tenircompte dans la conception des futursprojets d’extension agricole,débouchent sur les recommandationssuivantes.

Choix des objectifsLa première leçon porte sur lanécessité de mieux cibler les zones etles groupes où l’impact des servicesd’extension devrait être plus fort.D’où le besoin d’un système plussouple et plus « ingénieux », capablede déterminer le fossé qui sépare lesmeilleures pratiques des pratiquescourantes et d’allouer de façonrationnelle les ressources limitées. Enoutre, les services d’extensiondevraient sélectionner des paysansplus représentatifs du milieusocioéconomique local, de manière àpouvoir fournir des conseils mieuxadaptés aux catégories de paysans.

Systèmes d’informationAfin de mieux cibler leur action, lesservices d’extension doivent fournirau bon moment une informationfiable, ce qui souligne l’importance dumonitorage et de l’évaluation. Une desgrandes leçons à tirer de l’expériencekenyane est qu’il faut cerner lesdemandes des paysans et leur offrirdes services adaptés au contextetechnologique et économique de leur

Estos datos también indican quelos servicios de extensión no hantenido ningún efecto perceptibleen el nivel de la eficienciaeconómica. La baja eficienciaeconómica de los campesinossugiere la posibilidad de realizarahorros significantes sólo por

medio de introducir un conjunto deingresos que sería más económico bajolas condiciones del mercado presente.

Los datos corrientes no indicanningún efecto importante de la oferta dela extensión en la productividad al nivelde la granja. Lo que sí indican es que losrecursos de la extensión han sidoasignados de manera ineficiente aobjetivos malamente escogidos. Elincremento en la producción agrícola hasido más grande en las areas que anteseran menos productivas, mientras que lacolocación del personal de la extensiónha favorecido las areas más productivas.Es posible que la extensión hayaayudado a diseminar simples mensajestécnicos a las areas menos productivasque antes no se servían bien, pero estono se puede determinar conclusivamentea base de los datos disponibles.

Una proporción significativa de loscampesinos está dispuesta a pagar porlos servicios de extensión, lo cual indicaque valen los avisos que reciben. Pero elvalor percibido, cuando se representapor la suma que los campesinos estándispuestos a pagar, es bastante menosque lo que el gobierno desembolsa porgranja para los servicios de extensión.

Recomendaciones principalesLas lecciones que se han aprendido deeste estudio y que se pueden aplicar aldiseño de proyectos futuros deextensión agrícola sugieren lasrecomendaciones siguientes.

SelecciónLa primera lección es la necesidad deseleccionar más eficientemente las areas

EN

GL

IS

H

ES

PA

NO

L

FR

AN

CA

IS


xviii

which brings the focus to moni-toring and evaluation (M&E).An important lesson of theKenyan experience is that farm-ers’ demands should be identi-fied and that services should betailored to suit local technologi-

cal and economic conditions and cir-cumstances. M&E is also critical foridentifying the gaps and guiding the“smart” system, as needed, towardmore efficient targeting.

IntensityBased on their experience, farmershave clearly indicated that they donot want to see extension agents toooften, and there are not enough newtechnological recommendations tosustain a highly intense visit sched-ule. It would be more cost-effectiveto establish a leaner and less-inten-sive presence with wider coverage.It is conceivable that improvingservice quality could increasedemand. This would amplify theimportance of a responsive anddynamic delivery system.

PluralismDelivering standard messages bymeans of a single or uniform method-ology is likely to limit the effective-ness and efficiency of extension ser-vices. Younger, more educatedfarmers are taking over from theirparents. Radio programs that dis-seminate new information are popu-lar, and alternative providers arebeginning to emerge in rural Kenya.Exploiting low-cost modern commu-nications, demonstrations, print me-dia, and partnerships with civil soci-ety and the private sector couldbe cost-effective. This would leverageresources to increase outreach, andwould be likely to have a greaterimpact with the same or reduceddemand on government resources.

région. Le monitorage etl’évaluation sont aussiindispensables pour déterminerles lacunes et orienter le système« ingénieux » vers un meilleurchoix de ses cibles, en fonctiondes besoins.

IntensitéEn se basant sur leur expérience, lespaysans ont clairement exprimé qu’ilsne veulent pas rencontrer tropsouvent les agents des servicesd’extension, notamment parce qu’iln’y a pas assez de nouvellestechnologies pour justifier des visitesrapprochées. Il serait plus rentabled’instaurer un système qui allégerait laprésence et qui couvrirait des zonesplus vastes. On peut supposer quel’amélioration de la qualité des servicesaugmenterait la demande. Celapermettrait d’accroître l’importancedes services et de mettre en place unsystème dynamique et axé sur lesbesoins de la population.

PluralismeDes services d’extension quitransmettent des messages standardsinspirés d’une méthodologie uniqueou uniforme risquent d’être moinsefficaces. Des paysans plus jeunes etplus instruits remplacentprogressivement leurs parents. Lesémissions radio qui diffusent del’information nouvelle sontpopulaires, et d’autres fournisseurs deservices font peu à peu leur apparitiondans le Kenya rural. Il pourraits’avérer rentable d’exploiter lescommunications modernes à bas prix,d’organiser des démonstrations, derecourir aux journaux et de créer despartenariats avec la société civile et lesecteur privé. Ainsi, on multiplierait lesressources et étendrait la portée desactivités, ce qui accroîtraitprobablement l’impact des ressources

y los grupos para los cuales elimpacto marginal de los serviciosde extensión va a ser el másimportante. Para esto, se necesitaun sistema más flexible e“inteligente” que puedaidentificar las diferencias entre lasmejores prácticas y las prácticas

medias y asignar los recursos escasos deuna manera más racional. Además, loscampesinos que se escogen para lainteracción deben ser másrepresentativos del ambientesocioeconómico local, lo cual aseguraráel suministro de avisos más relevantes alas varias categorías de agricultores.

Sistemas de informaciónPara la selección se necesita un flujo dedatos oportunos y fiables, lo que sehace pensar en el monitoreo y laevaluación (M&E). Una lecciónimportante de la experiencia de Keniaes que las demandas de los campesinostienen que identificarse y que losservicios se deben hacer a la medida delas condiciones y circunstancias localestecnológicas y económicas. El M&E escrítico también para identificar lasfaltas y guiar el sistema “inteligente,”cuando sea necesario, hacia unaselección más eficiente.

IntensidadA base de su experiencia, loscampesinos han claramente indicadoque no quieren ver a los agentes deextensión con demasiada frecuencia, yno hay bastantes nuevasrecomendaciones técnicas parasostener un programa muy intensivode visitas. Sería más rentable estableceruna presencia menos fuerte e intensacon un alcance más amplio. Es posibleque un adelanto en la calidad de losservicios pueda incrementar lademanda. Esto agrandaría laimportancia de un sistema de entregamás sensitivo y dinámico.

xix


Client FocusThe central focus of the insti-tutional design should be toempower farmers. An effec-tive way to incorporate clientfocus is to integrate alterna-tive means of giving farmers

a voice—such as cost-sharing, fos-tering farmers’ organizations, anddecentralization—into the deliverymechanisms. Cost recovery (even ifonly partial), in particular, offersseveral advantages. It provides ap-propriate incentives, addressing theissues of accountability and qualitycontrol; it makes service moredemand-driven and responsive; itprovides some budgetary respite;and it encourages alternative pro-viders. Such institutional arrange-ments remain unexplored in Kenya.

EN

GL

IS

H

ES

PA

NO

L de l’État, qui pourraient peut-être même être réduites.

Priorité aux clientsLes services d’extensiondevraient viser l’autonomie despaysans. Une bonne façon d’y

parvenir consiste à intégrer auxmécanismes d’offre des services desmoyens permettant aux paysans de sefaire entendre – partage des coûts,incitation à l’établissementd’organismes de paysans etdécentralisation. Le recouvrement descoûts (même partiel) présenteplusieurs avantages. Il sert d’incitatif etintroduit la responsabilisation et lecontrôle de la qualité. Il axe davantageles services sur la demande, procureun certain répit budgétaire etencourage d’autres fournisseurs deservices. De tels arrangementsinstitutionnels demeurent inexplorésau Kenya.

PluralismoLa entrega de mensajesconvencionales por medio deun método único o uniformeprobablemente va a limitar laeficiencia y la eficacia de losservicios de extensión. Loscampesinos más jóvenes y

educados se están encargando de lastareas de sus padres. Programas de laradio que diseminan nuevainformación son populares, y losproveedores alternativos empiezan aaparecer en la Kenia rural. Laexplotación de comunicacionesmodernas de bajo costo,demostraciones, la prensa, yasociaciones con la sociedad civil y elsector privado puede ser rentable. Asíse aprovecharía de los recursosdisponibles para el alcance para tenerun máximo efecto con la misma o aúnmenos demanda para los recursosgubernamentales.

Enfoque en el clienteEl enfoque central del diseñoinstitucional debe ser de apoderar a loscampesinos. Una manera efectiva deincorporar un enfoque en el cliente es laintegración dentro de los mecanismos deentrega de métodos alternativos paradar una voz a los agricultores, métodostales como la repartición de costos, elapoyo a los organismos que representana los campesinos, y la decentralización.En particular, la recuperación de loscostos (aunque sea solamente en parte)ofrece varias ventajas. Provee incentivosapropiados, dirigiéndose a los temas decontabilidad y control de la calidad;resulta en un servicio más sensitivo yempujado por la demanda; ofrece unatregua presupuestaria; y estimula a losproveedores alternativos. Arreglosinstitucionales de este tipo se quedan aexplorar en Kenia.

FR

AN

CA

IS

xxi

ABBREVIATIONS AND ACRONYMS

AIC – Agriculture Information CenterASIP – Agricultural Sector Investment ProgramATD – Africa Technical DepartmentCVM – Contingent valuation methodDEA – Data envelopment analysisFEW – Frontline extension workerFTC – Farmer Training CenterGDP – Gross domestic productIDA – International Development AssociationIFAD – International Fund for Agricultural DevelopmentKARI – Kenya Agricultural Research InstituteKP – Kenya pounds (currency)Ksh – Kenya shillings (currency)MALDM – Ministry of Agriculture, Livestock Development, and MarketingM&E – Monitoring and evaluationNASSEP – National Sample Survey and Evaluation ProgramNEP – National Extension ProjectNGO – Nongovernmental organizationOED – Operations Evaluation DepartmentPPA – Participatory poverty assessmentRHBS – Rural Household Budget SurveySAR – Staff appraisal reportSMS – Subject matter specialistT&V – Training and visitWTP – Willingness to pay

1

T

11Introduction

he focus on agricultural extension in Kenya is rooted in agriculture’s strategic

importance in development. With 70 percent of the population living in rural areas,

almost half in absolute poverty, the centrality of rural development to any strategy for

poverty alleviation is clear. Critical to rural development is the performance of agriculture, both

because of its direct contribution to the national economy (28 percent of gross domestic product;

60 percent of export earnings; and 80 percent of national employment) and because of its

indirect multiplier effects. Within agriculture, 81 percentof the farmers are smallholders (owning less than 2hectares), which makes it imperative to improve small-holder productivity.

Accordingly, the government of Kenya has longhad agricultural extension on its development agenda.With World Bank support, Kenya adopted a T&Vsystem of management for its extension services in1982 as part of its growth strategy. The Bank thenfinanced the extension system with NEP I (1983–91)and NEP II (1991–98).

The performance of Kenya’s extension system iscontroversial, and this debate is part of a broaderdisagreement about the cost-effectiveness of the T&Vapproach. Debate has focused largely on the efficacy ofthe mechanisms for delivering extension advice(Picciotto and Anderson 1997). It is generally agreed thatthe T&V system is costly. The controversy centers on thereturns to the high levels of investment by borrowercountries in the T&V system, and hence on its impact onagricultural production (Purcell and Anderson 1997).(The general nature of the debate is discussed in AnnexA.) Despite the debate’s intensity, the importance ofagricultural extension in the Bank’s development strat-

egy for Africa, and the large investment made, therehave been few attempts to rigorously measure theimpact of T&V extension.

In the Kenyan context, the debate has been elevatedby the estimate of veryhigh returns to T&Vextension by a Bank Af-rica Technical Depart-ment (ATD) study(Bindlish and Evenson1993, 1997), on the onehand, and the lack ofvisible results on theground, on the other.1

Agriculture’s poor per-formance in Kenya in recent years, the country’sdeclining budgetary resources, and efforts to rationalizethe structure of the Ministry of Agriculture, LivestockDevelopment, and Marketing (MALDM) have increas-ingly called into question the effectiveness of extensionservices.2

Notwithstanding the ATD’s estimates of highmarginal returns to extension, an Operations Evalua-tion Department (OED) review of NEP I raised

The performance ofKenya’s extensionsystem is controversial,mirroring the broaderdisagreement about thecost-effectiveness ofthe T&V approach.

2


questions about the effectiveness of the T&V approachin Kenya. The review concluded that NEP I had somebeneficial aspects but several operational deficiencies,and, most important, that it was not financiallysustainable. With little evidence to suggest any sig-nificant impact on agricultural growth to justifyT&V’s high fiscal costs, OED questioned both theappropriateness of the extension format developed asa uniform approach throughout Kenya’s croppingareas and the overall outcome of the project. Theensuing discussions with the Africa Region led toOED’s final rating of the project outcome as margin-ally satisfactory. The Africa Region contested therating, and the disagreement has persisted. This studywas undertaken to inform the discussion.

The discontent with the current system of extensionis not limited to the Bank. Criticism is sharpest amongthe donor and nongovernmental organization (NGO)

communities, which seekreform toward a demand-driven system that betteraccommodates alternativeways of reaching farmers.The Kenyan governmentis also wary of continuingwith the large allocationof resources to extension,given agriculture’s poorperformance since thestart of the 1980s.

It is not clear, however,whether the poor record of

agriculture necessarily reflects the performance of exten-sion; it is possible that agriculture could have done worsewere it not for extension. Establishing the impact ofextension from the sector’s aggregate performance isdifficult because, much like casual observations from fieldvisits, it lacks an appropriate counterfactual. The prob-lem can be demonstrated with maize yields.3 Between1970 and 1989, the annual growth rate of maize yields inKenya was 4.7 percent. This rate rises to 5.4 percent whenrainfall is factored in.4 Comparing the periods before andafter 1982—that is, before and after T&V—growthslowed from 6.3 percent to 5 percent, a statisticallysignificant difference. The difference, however, ceases tobe significant when rainfall is factored in.

Given the strong and divergent opinions on theperceived performance of NEP I and NEP II and the lackof evidence to determine their impact, this study takes an

objective, empirical approach.5 Most of the conclusionsare based on the results of a 1997 OED household surveyand a survey of the extension staff, supplemented bysecondary data and information from several recentstudies by MALDM. OED’s household survey (see AnnexI) covered the same population surveyed for the 1990 ATDstudy, which used a subsample of the 1982 RuralHousehold Budget Survey (RHBS) data. In 1997, inter-viewers revisited as many of the respondents as could becontacted in the clusters that had been sampled by theATD study. This evaluation thus has the advantage ofbaseline data, even if they are somewhat limited. Todissociate the OED survey from the government extensionservice and the Bank, it was implemented by the TegemeoInstitute of Egerton University in Kenya.

Study ObjectiveThe goal of this study was to make an empiricalassessment of the impact of the NEP I and NEP IIprojects in Kenya. Following a theory-based evaluationapproach, it combined qualitative and quantitativemethods to arrive at a credible body of evidence on theprojects’ likely impact. In so doing, it provides anindependent review of earlier findings.

To allow for appropriate policy conclusions, thestudy distinguishes clearly between the impact of exten-sion and the impact of a particular system of extension.This has implications for interpreting the results toilluminate the three key aspects of the study: the rel-evance, efficacy, and efficiency of the Kenyan extensionsystem. In Kenya, impact evaluation is complicated bythe introduction of the T&V system on a national scale,which precluded a with-and-without comparison. Thesystem was also introduced rapidly, over the course ofonly 3 years, so that now, after some 15 years, the dataallow only limited before-and-after comparisons.

The National Extension ProjectsBefore NEP I and NEP II, the traditional system ofextension in Kenya suffered from several weaknesses.6

To overcome them, the T&V system of extension wasintroduced with the intent of providing “competent,well-informed village-level extension workers who willvisit farmers frequently and regularly with relevanttechnical messages and bring farmers’ problems toresearch” (Benor and Baxter 1984). (The design fea-tures of the T&V system implemented by NEP I andNEP II are described in Annex B.) T&V was introducedas a brief pilot project in two districts in 1982. Starting

The discontent withthe current system of

extension is notlimited to the Bank.

Criticism is sharpestamong the donor and

nongovernmentalorganization

communities.

3

I n t r o d u c t i o n

FIGURE 1.1. STYLIZED IMPACT MODEL

Technology Generation Knowledge Delivery Impact

Householdobjectives Output

RiskCredit

Food security

Activities

Demonstrations/field trials

Field days

Media: audio, video, print

Public extension service

Farmer organizations, NGOs

Private sector: input suppliers,processors, consultants

ProductivityEfficiency

InstitutionaldevelopmentSustainabilityEfficacyPlurality

Research-extension links

RecommendationsTrainingFeedback

Land quality

Inputs

Farm Decisionmaking

Output

AwarenessAdoption

AccessContactDistribution

Outcomes Results

Friends, neighbors,innovative farmers

Basicresearch

AdaptiveresearchSpillovers

Indigenoussystems

Labor Prices

EducationInfrastructure

Weatherand pests

HHwelfare

Note: Broken line: feedback from farmers to decisionmakers. Solid line: delivery of time-bound “messages” by extension workers fromresearchers to farmers. HH: household.

in 1983 with NEP I, it was rapidly expanded to coverabout 90 percent of Kenya’s arable land.

The objective of NEP I was to achieve sustainedincreases in agricultural production in 30 of Kenya’s41 districts, covering all medium- and high-potentialarable areas. The approach entailed reorganizingand strengthening extension services through theadoption of T&V over a period of three years andimproving the link between research and extension.NEP I was considered an institution-building project,and the Bank expected to provide external assistancefor 10 to 15 years to ensure that the necessaryinstitutional reforms and improvements in staff skillswere made.

In 1991, NEP II succeeded NEP I. The objective ofNEP II was to stimulate the development and adoptionof technical packages that would enable smallholders

to increase their productivity and incomes. In additionto continuing to support the work of NEP I, NEP IIintroduced T&V to 6 new areas, and when 4 of theoriginal districts were split, T&V extension wasapplied in 40 of Kenya’s 45 districts. The project wasdesigned to consolidate and fortify the gains madeunder NEP I, to increase direct contact with farmers, toimprove the relevance of extension information andtechnologies, to upgrade the skills of staff and farmers,and to introduce pilot innovations into the extensionsystem.

Study Strategy and OutlineThe measurement of the impact of the two projectsfocused primarily on their key objectives—institutionaldevelopment and sustained increases in agriculturalproductivity. While important economic, social, and

4


environmental impacts are implicit in the projects’rationale, they were not explicitly stated as majorobjectives. But the key social issues of gender and thedistribution of benefits are naturally addressed becauseof the large proportion of Kenyan farmers who arewomen or smallholders.

The study’s theory-based approach used the styl-ized causal flow model depicted in figure 1.1. Themodel reflects the reliance of NEP I and NEP II onextension agents successfully delivering time-bound“messages” (indicated by solid arrows) from research-

ers to farmers.7 The taskof delivering feedbackfrom the farmers to theresearchers (indicated bybroken arrows) was alsoto be carried out by theextension workers. Asimplemented, the pro-gram was decidedlynonparticipatory—therewas no pretense ofinvolving farmers in thedevelopment of technol-ogy or the messages, norwas there an opportunityfor the farmers to selecttopics of interest to them(MALDM 1997b).

The impact of the projects can beassessed at several points along the continuum, fromthe knowledge complex to the final change in welfare.Following the results-based management framework,this evaluation sought to relate the results observed inthe farmers’ fields back to project inputs. In addition,intermediate output and outcome indicators weremeasured to assess the performance of the extensionsystem along the results chain to confirm the potentialfor impact. Accordingly, the five boxes along thebottom of the model present the key indicativemeasures corresponding to inputs, activities, outputs,outcomes, and results. The evaluation strategy was touse a mix of qualitative and quantitative methods tomeasure and analyze these key indicators. No attemptwas made to study the projects’ impact on householdwelfare, because it was likely that it had been affectedby a number of factors beyond the scope ofextension’s activities. The outline of this volumereflects the study approach.

Institutional DevelopmentThe focus of institutional impact is on the left half ofthe model in figure 1.1, specifically on the institutionalarrangements, or input, used by NEP I and II to delivertheir output. The analysis is presented in Chapter 2,and combines secondary data with the findings ofseveral recent reviews of extension in Kenya and resultsfrom the OED staff survey.

Beneficiary AssessmentAn important element of the strategy was to obtainparticipatory beneficiary assessments of extension ser-vice. Although the beneficiary assessments for thisevaluation were not conducted, findings from tworecent participatory assessments in Kenya are perti-nent, and these, along with the farmers assessmentsfrom the OED household survey, are discussed inChapter 3.

Efficacy: The Quantity and Quality of ContactThe debate on the effectiveness of T&V in Kenya hascentered primarily on the “real” side of the equation—that is, T&V’s impact on agricultural productivity atthe farm level. This impact is evaluated following thestylized flow expressed in the model in figure 1.1. Thefirst part is an analysis of the outreach and the qualityof interaction between extension agents and farmers,which is discussed in Chapter 4.

OutcomesNext on the continuum leading to ultimate results is themeasurement of the proximate outcomes of extensionefforts, which indicates potential for impact. Thisentailed measures of farmer awareness and adoption ofextension outputs (that is, its technological recommen-dations, or “messages”). These measures, and how theyrelate to the supply of extension services, are discussedin Chapter 5.

ResultsDetermining the physical impact, or results on theground, required relating the supply of extensionservices to changes in productivity and efficiency atthe farm level. To do this, the evaluation followedtwo lines of inquiry. One was a nonparametricmeasurement of farmers’ efficiency in production andhow farm productivity changed over time. (This isdiscussed in Chapter 6.) The other was an econometricmeasurement of the impact of extension on farm

As implemented, theprogram was

decidedlynonparticipatory—

there was no pretenseof involving farmers in

the development oftechnology or the

messages, nor wasthere an opportunity

for the farmers toselect topics of

interest to them.

5

I n t r o d u c t i o n

production (see Chapter 7). These estimates reveal thereturns to investment in extension and are needed toestablish its efficiency.

Willingness to PayThis analysis revisits the farmers with a new method-ological tool—the contingent valuation method, whichis used to directly elicit their willingness to pay forextension services (see Chapter 8). The findings fromthis exercise have implications for the relevance of thedesign of past projects and provide insights for futureproject design.

7

22

A

Institutional Development

key objective of both NEP I and NEP II was the development of organizational and

institutional arrangements for the efficient and effective delivery of agricultural services

to Kenyan smallholders. In 1982, it was expected that the extension system would

need 15 years to mature. The period has now passed, and it is natural to ask how the extension

system is performing in comparison with the system it replaced. The discussion in this chapter is

based on reviews undertaken or commissioned by MALDM over the past two to three years in

an attempt to develop a national policy and a frame-work to rationalize the current extension service.1

These reviews are supplemented with conclusions fromdiscussions with extension and ministry staff, focusgroups, individual district-level staff, and former exten-sion staff familiar with the early days of NEP I andwith findings from the 1997 staff survey conducted byOED for this study.

It is instructive to review the before-and-afterpictures of Kenya’s extension system.

BeforeAgricultural extension in Kenya dates back to the early1900s. Several approaches were tried, including indi-vidual visits, group methods, unified extension, farmmanagement, integrated development, and specializedcommodity extension programs. Except for the last,none endured. Nevertheless, the traditional extensionsystem was highly successful in the dissemination ofhybrid maize technology. Starting in 1965 with anationwide program of demonstrations and field days,hybrid maize was promoted so successfully that by1977, 50 percent of even smallholders were growinghybrid maize (Johnson and others 1980).

By 1982, the year before NEP I was introduced,several disparate and uncoordinated donor-supportedefforts were in the field. The various extension arrange-ments lacked a consistent national strategy and wereessentially ad hoc project components. This prolifera-tion of programs was viewed as expensive, inefficient,and largely ineffective.

The extension services did, however, have awell-defined line of command, from the director ofagriculture down to the field-level, frontline exten-sion worker (FEW). The staff numbers were seen asadequate, but the services were thought to be per-forming well below their potential (World Bank staffappraisal reports).2

There were severalreasons for this: thejunior staff were notwell trained; therewas a lack of fieldemphasis; and FEWsvisited few farms,and most of their vis-its were to progressive farmers, who representedabout 10 percent of all smallholders. Other problems

The traditional extensionsystem was highlysuccessful in thedissemination of hybridmaize technology.

8


TABLE 2.1. NEP I AND II: BORROWER PERSPECTIVESWeak points Strong points

• Very broad objectives • Wide coverage• Strong top-down • Coverage of all types of

planning farmers• Target not specific • Strong staff training• Weak farmer • Development of

participation professionalism at• Low staff motivation district officer level• Weak monitoring and • Strong presence of FEWs

evaluation • Procurement of• Supply-driven messages transport equipment• Donor-dependent and office• Little flexibility accommodations• Little accountability

Source: MALDM 1997a, p. 14.

included insufficient operational funds, which lim-ited mobility (except for projects and programsfunded by donors), resulting in lax supervision andsupport of the field staff, and generally inadequateallowances for the FEWs, which left them to travelon foot and unable to cover a large area.

The extension system in place before theprojects were implemented concentrated its effortson male farmers, even though almost a third of thefarmers or farm operators were women. However,women’s groups had started to proliferate, and thegovernment, recognizing their importance, was tak-ing steps to integrate women into the extensionservice.

The link between research and extension wasweak, but it was recognized that the research staff

members were more spe-cialized and betterequipped to analyzefarmers’ problems thanthe specialists workingwith extension. Themain vehicles for dis-seminating technicalknowledge to extensionstaff were field days;research station bulle-tins; Agriculture Infor-mation Center (AIC)publications on recom-mended practices; and

barazas, public meetings called by the chief, thelocation administrative officer, or local extension staff.The Banks’ appraisal of NEP I, however, was realistic.It warned against expecting major increases in produc-tion as in the past, because new technological develop-ments were not available to promote rapid growth. Thefocus was thus to rely on smallholder intensificationand effective resource conservation.

And AfterAnnex C details some of the extension services’ specificinstitutional features. Table 2.1 summarizes the analy-sis of the approach to agricultural extension of NEP Iand NEP II from the perspective of Kenyanpolicymakers (MALDM 1997a). Annex D presents abriefing prepared by the staff of one of the districts thatwas visited for this study for a focus group discussion; itsummarizes the generally held views about NEP I and

NEP II in Kenya. The rest of the chapter describes themain conclusions on institutional impact that emergefrom this study’s analysis.

Organizational StructureNEP I and NEP II succeeded in establishing an integratednational extension system. Although the existing organi-zational structure was consistent with the single line ofcommand envisaged in the projects’ design, the manage-ment of the system has been weak, an outcome of poorproject implementation arrangements. Inadequate finan-cial management compounds the problem of limitedbudgetary resources, disrupting the timely flow of opera-tional funds. Kenya is also still developing a nationalpolicy on agricultural extension. While a qualitativeanalysis of the extension approach before NEP I similar tothat shown in table 2.1 is not available, the institutionalcharacteristics of the previous system, summarized in thepreceding section, provide some points of comparison.

On the positive side, the infusion of large sums ofdevelopment and operational funds in the early yearsof NEP I generated unprecedented energy in the system.Increased staff training, new vehicles and office equip-ment, and the new institutional paradigm representedby the extension system lifted the morale of the fieldstaff. The link with research was weak, but sufficient toensure an adequate flow of simple agronomic messagesto the farmers. The increased activity is generallybelieved to have been beneficial, and training im-proved staff skills.

The focus was on maize and simple agronomicmessages. While it is likely that this initially had some

The extension systemin place before the

projects wereimplemented

concentrated its effortson male farmers, eventhough almost a thirdof the farmers or farm

operators werewomen.

9

I n s t i t u t i o n a l D e v e l o p m e n t

positive impact in areas previously not covered byextension, the continued narrow focus over timereduced incremental benefits. With sufficient fundsflowing, services expanded rapidly and staff numbersincreased, which increased outreach to uncovered areasand unserved categories of farmers.3 This helpedreduce some of the biases of the previous system—thoseagainst women, younger farmers, and farmers livingfar from access roads. But new biases, in favor of themore educated and more productive areas, were intro-duced. The allocation of resources also favored areaswith lower poverty levels, and there does not appear tohave been any improvement in staff productivity asmeasured by the number of contacts per staff member.

With staffing rising to unsustainable levels and thecost of extension high, operational budgets proved inad-equate. At the end of NEP II, a number of the problemsafflicting the previous system continued, and their impacton the effectiveness of services has been significant. Thestaff survey showed that more than half of those whowere in service in 1982 felt that the system was lesseffective at the time of the survey (1997) than it had beenin 1982.4 Less than a third considered it to be moreeffective (the rest thought it had remained the same). Thesystem is also more expensive now, but no more efficient.Insufficient operational funds mean limited transportfacilities and inadequate staff allowances, restrictingmobility and reducing supervision. Slow progress in thegeneration of technology and limited adaptive researchhave diminished the number of new messages and havemade training sessions and field visits repetitive andunproductive. As a result, staff motivation and moralehave declined substantially since the initial boost in theearly years of NEP I.

Institutional DevelopmentThe institutional aspects of extension have also beenpoorly developed. As implemented, both NEP I and NEPII adopted a top-down, supply-driven approach, andlacked a focus on the critical issue of farmer empower-ment. While this may have been useful in the early,formative years of the program, change was small andslow, even during NEP II, despite the explicit intent of itsdesign. More important, the primary client, the farmer,still has little or no voice. While most staff believe thatboth they and the farmers have a say in the developmentof messages, a majority have also noted that districtofficers determine the topics of the training sessions. Thekind of information that farmers want—advice on com-

plex practices—and what extension effectively delivers—simple agronomic messages—are mismatched. The dem-onstrations that the farmers want and the home visits thatthe extension agents prefer are also methodologicallymismatched. These disconnects manifest the lack of clientfocus and responsiveness of the extension service.

The poor functioning of the contact-farmer andcontact-group approaches, together with inadequatemessages and the lack of operational funds, has sent theextension agents back to the old system of disseminat-ing messages through barazas. A large number of fieldstaff are also working with alternative providers,NGOs, the private sector, and other projects. But theextension service hasnot yet attempted to in-stitutionalize theselinks to make the sys-tem more effective andefficient.

A key feature, theincentive structure, hasbeen given inadequateattention in the institu-tional design of NEP Iand NEP II. FEWs haveno accountability to the farmers. The control mecha-nism adopted in NEP I and NEP II, following thestandard T&V approach, was designed to ensure thatFEWs would follow a prescribed route to regularlymeet with a fixed number of contact farmers. Supervi-sion is reduced to ensuring compliance with thespecified route. There is no focus on the quality of therelationship between FEWs and farmers, and FEWsare, for the most part, messengers.

Financial SustainabilityThe most problematic feature of the Kenyan system is thatit is not financially sustainable. Compared with theprevious system, T&V is significantly more expensive. In1982, Kenya was spending the equivalent of US$3.92 perhousehold in 1991 constant dollars (Bindlish and Evenson1993, estimates for their study districts). Current estimates(using national data that do not appear to differ muchfrom estimates for the same study districts) suggest anexpenditure of US$13.29 in 1991 dollars, or US$15.11 in1997 dollars. That is, in real terms extension serviceexpenditures have gone up more than 300 percent, whileincreasing fiscal difficulties have led to declining opera-tional budgets.

As implemented, bothNEP I and NEP IIadopted a top-down,supply-drivenapproach, and lacked afocus on the criticalissue of farmerempowerment.

10


The rise in costs reflects the system’s design, withits focus on a high frequency of field visits and trainingsessions, and blanket coverage of most of the country’sarable area according to a standard FEW-to-farm ratio.While it may be argued that the high costs stem fromincreases in staff numbers, and that NEP I and NEP IIdid not cause these increases, it should be noted thatcurrent staffing levels are below the standards consid-ered acceptable for the two projects. Even at currentlevels, the operational budget is highly inadequate.

ConclusionsThe limited impact of the two projects on the institutionaldevelopment of Kenya’s extension services is evident. Atthe end of NEP II, the service lacks a strategic vision,appears to have had no appreciable improvement in itseffectiveness, and suffers from weak management. Thereis virtually no management information system. Thebenefits include wider coverage, improved research-extension links, and improved staff quality throughtraining.

While the general principles underlying T&V arerelevant for any institutional design, the highly struc-tured extension approach that NEP I and NEP IIadopted has been neither effective nor sustainable.5 It

has shown little flex-ibility, even thoughNEP II’s design soughtto introduce some plu-rality to improve sys-tem functioning. Froman institutional perspec-tive, responsiveness andaccountability usuallycannot be bureaucrati-cally imposed. Theprojects’ design failedto incorporate appro-priate incentives for ef-fective service delivery.

Searching forAlternativesThe prolonged ineffec-tiveness and lack of fis-

cal sustainability of the extension service have turned thegovernment toward alternative approaches. It has takenimportant steps recently to adopt a farming-systemsapproach to extension, improve research-extension links,

rationalize the deployment of frontline extension staff,and experiment with alternative modes of delivery suitedto local circumstances. In conjunction with other donors,the Bank has supported the bulk of this work, indicatingthat the Bank is also being responsive. Resulting changeson the ground, however, are yet to be seen.

When alternatives for the future are considered,appropriate exit mechanisms need to be incorpo-rated into the institutional design. To do this, thenature of the benefits, the kinds of services, and theefficiency of their delivery mechanism need to beconsidered. Most agricultural information yieldsbenefits in the form of private returns to farmers,whether embodied as a technological characteristicof goods, such as a fertilizer of high quality or anew type, or in less concrete form, such as informa-tion that improves farmers’ management skills.6

Several types of information also have benefitsexternal to the farmers, raising their overall orsocial returns, as others learn by indirect informa-tion or observation. These externalities provide therationale for public support for extension activities.Nevertheless, as long as farmers experience private

Women working in the field, Kenya.

Government has takenimportant steps

recently to adopt afarming-systems

approach to extension,improve research-

extension links,rationalize the

deployment of frontlineextension staff, and

experiment withalternative modes of

delivery suited to localcircumstances.

11

I n s t i t u t i o n a l D e v e l o p m e n t

benefits, they should be willing to pay for theinformation, suggesting scope for cost recovery.(The findings reported in Chapter 8 support this.)

Delivery OptionsThe kinds of information most closely associated withextension services are simple agronomic messages,which have the characteristics of genuine public goods,since they are easily passed around. But these messagesare also the kind of information that can be deliveredwell through the more cost-effective means of radiobroadcasts, pamphlets, and public meetings. The printmedia, in particular, have significant potential as theyounger, more educated farmers take over from theirparents. Given the nature of the information that mostextension agents have to offer, the decision of mostagents to return to the baraza appears to be rational. Forother kinds of extension information, more detailed andpersonal advice is required, such as addressing afarmer’s specific pest problems or the optimal type andquantity of fertilizer for local soil conditions. Since thereturns to such information are also largely private,there is significant scope for cost recovery or privateprovision. Further, since the greater use of such inputsalso benefits the private input suppliers, they have anincentive to deliver such information. (That this trendmay already be under way, although proceeding slowly,is evident from the findings reported in Chapter 5.)

IncentivesA number of institutional options—including decen-tralization, using farmers’ organizations and those ofcivil society, outsourcing extension services, and cost-sharing—could sharpen system focus on clients. Thecentral aim of the institutional design should be toempower farmers, which requires the inclusion ofappropriate incentives.

Incentives can be addressed directly by building insome degree of commercialization. It is often argued thatextension services are a public good and that mostfarmers, particularly poor subsistence farmers, may not

be willing to pay for them. This hypothesis is difficult totest, however, because public extension has historicallybeen provided without charge, which tends to crowd outalternative providers. Still, in the Kenyan communitiesthat extension does not reach or where the service isinadequate—mostly low-potential and poorer areas—NGOs are active, and some charge a fee, althoughindirectly, as a membership fee for a group or club.

Commercialization does not necessarily imply adirect cash payment for advice or doing away withpublic extension alto-gether. The lattermay be neither fea-sible nor desirable, atleast in the interest ofequity and broadercoverage, in the fore-seeable future. Creat-ing partnerships withlocal suppliers, pro-viding training sessions for them (although they arelikely to be able to afford fees for training), andimproving the infrastructure for input delivery are alsoways of commercializing information delivery.

Recovering costs through direct charges may notbe as far-fetched as is often believed; farmers appear tobe willing to pay for extension advice. Even if directcharges are only nominal or partial, they have severaladvantages in correcting for the shortcomings of thecurrent system. They provide the right incentives foragents to deliver the advice that farmers want; makeagents accountable to the farmers; build in a genuinequality control mechanism; alleviate budgetary con-straints (at least somewhat); and encourage other,notably private, services. Commercialization does notnecessarily mean totally privatizing services. Ways ofintroducing commercial elements into the provision ofpublic service and various forms of partnershipbetween civil society and public and private entitiescan be profitably exploited. Such partnerships arealready emerging in Kenya.

Recovering costs throughdirect charges may not beas far-fetched as is oftenbelieved; farmers appearto be willing to pay forextension advice.

13

A

Beneficiary Assessment

33

participatory beneficiary assessment was included in the study to hear directly from the

farmers about their access to extension services, the quality and relevance of the advice

they receive, and their suggestions for the future.1 In the spirit of collaboration, it was

agreed that MALDM would conduct the beneficiary assessment with active involvement from

OED and the Bank’s Africa Region. After the initial preparations had been made, however,

MALDM decided unilaterally, for unknown reasons, not to pursue the assessment.

This study draws instead on two independentassessments. One was a beneficiary assessment con-ducted for a separate OED study (Actionaid Kenya1997, done for an OED NGO study, 1998). One of thatassessment’s components was designed to obtain theviews of users, as well as potential users, of Kenyanextension services, and some of the findings arerelevant to this study. The second assessment was aparticipatory poverty assessment (PPA) conducted inKenya in 1994, and some of the results are alsopertinent for this study (Narayan and Nyamwaya1995).2 The findings of both assessments are comple-mented by those from the OED survey.

Welfare and ProductivityThe ultimate goal of NEP I and NEP II was to improvefarmers’ welfare, primarily by increasing agriculturalproductivity. At the start of the OED survey, farmerswere asked to assess their welfare (defined as a self-assessed standard of living) and farm productivityrelative to their situation 10 to 15 years earlier. Themajority, or 66 percent, thought that their welfare waslower than it had been before; only 25 percent thoughtthat it was better. As for agricultural productivity, more

than 72 percent thought it was lower than before, while25 percent thought it had improved. Similar sentimentswere expressed in the 1994 PPA: most of the respon-dents thought that life had been better eight to ten yearsearlier. In a follow-up PPA conducted in 1996 by theCentral Bureau of Statistics in Kenya, about 70 percentof the participants thought that poverty was worse thanit had been five years earlier. The Actionaid Kenyaassessment presented a similar picture.

Access to ServicesThe OED survey asked farmers about the change intheir access to, and the quality of, 19 social andinfrastructure services, including extension, over thepast 10–15 years. The results are presented in table3.1. A large majority—75percent—thought that theiraccess to extension had notchanged. Very few thoughtquality had improved(11.4 percent), but the larg-est number (39.4 percent)thought it had deteriorated. Compared with otherservices, extension appears to be generally worse-off in

Most respondentsthought that life hadbeen better eight toten years earlier.

14


perceived quality. It is noteworthy that for a number ofservices, private providers have compensated for thedecline in public services. For example, veterinaryservices have been privatized as a matter of policy, andthis is reflected in better access and quality in privateextension services. Other public services that havedeteriorated significantly are roads and public health.

Interactions Between Farmers and AgentsThe PPA and the beneficiary assessment also revealthat access to information is lacking, particularlyamong the poor. Both assessments found extension

services to be spo-radic or irregular,and generally tar-geted toward the richor large landowners,bypassing the poor.Interactions with ex-tension agents werenoted as being few,and service was gen-erally reported to beinconsistent, inad-equate, and infre-

quent.3 Most farmers mentioned the chief’s baraza,the radio, or friends and neighbors as their most

consistent source of information. The few demonstra-tions that are held are on the wealthier or moreprogressive farmers’ fields, which are the ones thatextension agents generally visit. The beneficiaryassessment also found that residents of zones withlower potential have little access to extension andthat NGOs are servicing them. Extension andNGOs, however, have no effective interaction.

The Relevance of Agents’ AdviceThe differing perceptions of the needs of the poor arean important issue in extension service effectiveness.The extension staff target the wealthy farmers, hopingfor faster adoption of new technology, since it is morelikely that the wealthy will be able to afford it. At thesame time, however, the intent and the hope are thatthe demonstrations will encourage the other, rela-tively poorer, farmers to adopt the same high-input,high-cost technology. This contradiction limits theeffectiveness of the current approach to extension. Thecurrent nonusers, or those who do not have access,would like to get information on crops that the largerfarmers do not grow—crops other than maize andcoffee. They also seek advice on less costly technolo-gies, marketing, and diversification. This is not thesort of information they get on demonstration plots,which are usually maize demonstrations. At the same

Extension staff targetwealthy farmers,

hoping that thedemonstrations will

encourage the other,relatively poorer, farmers

to adopt the samehigh-input, high-cost

technology.

TABLE 3.1. CHANGES IN FARMERS’ ACCESS TO SERVICES AND SERVICE QUALITYAccess Quality

No Don’t No Don’tService Better` Worse change know Better Worse change know

Seed dealer 34.6 1.0 63.9 0.5 51.7 6.2 37.5 4.7Fertilizer-chemicals store 31.5 1.0 66.4 1.0 48.7 3.7 40.2 7.4Village shopping center 14.4 0.2 85.2 0.2 69.8 6.9 23.1 0.2Output market 14.5 1.3 84.0 0.2 53.3 10.6 34.3 1.8Banking services 18.7 1.3 77.5 2.5 27.9 3.2 43.2 25.8Other credit sources 10.6 1.4 78.5 9.5 22.5 8.9 40.6 28.0Private health centers 62.0 0.6 34.8 2.6 54.2 4.9 26.7 14.2Public health dispensary 23.7 1.9 74.3 0.2 30.8 48.2 20.2 0.9Primary school 14.5 0.3 85.2 0.0 50.6 14.1 33.7 1.6Secondary school 20.5 1.2 78.3 0.0 51.2 8.7 34.7 5.5Electricity 51.7 1.3 46.3 0.7 20.8 1.1 58.6 19.4Telephone facility 48.6 2.2 48.0 0.3 30.4 8.9 46.4 14.4Piped water 33.7 18.0 59.4 3.8 11.9 20.4 46.7 21.0Tarmac road 14.7 1.9 83.1 0.3 26.1 30.2 40.7 2.8Dry season road access 6.5 3.3 90.0 0.2 22.9 50.3 26.5 0.4Public transport 7.6 1.2 91.2 0.0 41.3 17.9 40.8 0.0Private veterinary services 51.7 0.7 32.5 15.1 42.1 5.1 30.3 22.6Public veterinary services 17.2 2.4 77.3 3.1 16.0 27.3 42.8 13.9Extension services 16.6 3.8 75.0 4.6 11.4 39.4 32.9 16.4

Source: OED survey.

15

B e n e f i c i a r y A s s e s s m e n t

time, the users—that is, the few who do have access—consider the technical advice they are given to berelevant and high in quality.

Farmers’ PrioritiesAmong the services that farmers would like to seeimproved, if they were to pay for the improvements,about 5 percent selected extension as their top priority.In overall ranking, however, extension was sixthamong the 19 services.4 The services ranked higher arealso the ones that the farmers reported as havingdeteriorated in quality.5 They generally gave lowerrankings to services that had improved. This is consis-tent with the perceived deterioration of extensionservices.

ConclusionsThe beneficiary assessment and the survey are consis-tent in finding that welfare has declined over time.The assessment’s findings also show that most farm-ers, especially the poor, have little access to extensionadvice. But this result should be put in perspective.Both the PPA and the beneficiary assessment notedthat those who have access, or the current users ofextension services, recognize the quality of the advicerendered. Extension is also not expected to reach all

farmers, and hence the need for selectivity andreliance on farmer-to-farmer information dissemina-tion. The observation that extension is weak in low-potential zones is also consistent with the focus ofextension, especiallyin NEP I, on zoneswith relatively highpotential.

The key insightsfrom the beneficiaryassessment and thePPA point to the rea-sons for the limitedeffectiveness of thecurrent extension sys-tem. The poor qual-ity of interaction with the vast majority of poor andsmallholder farmers and the irrelevance of advice totheir needs suggest poor targeting and poor respon-siveness. If extension has to be selective, it shouldselect a more representative set of farmers so that theadvice delivered is relevant to a broader range offarmers. It is also clear that the needs of the farmers,particularly the small farmers, are diverse and gowell beyond the production of traditional crops suchas maize and coffee.

The poor quality ofinteraction with the vastmajority of poor andsmallholder farmers andthe irrelevance of adviceto their needs suggestpoor targeting and poorresponsiveness.

17

44EfficacyThe Quantity and Quality of Contact

Efficacy refers to the extent to which the project design successfully delivered

extension advice. For NEP I and NEP II, efficacy can be viewed as the overall

effectiveness of both the extension system’s approach and its outreach. In either

case, efficacy is assessed through the output indicators for the projects: the extent and

nature of contact between farmers and extension. The detailed findings presented

in Working Paper 1 to this study (see Bibliography) are summarized here. The study

analysis uses data from the 1997 OED survey on thefarmers’ access to information on agricultural enter-prises, including specific questions related to the natureand extent of contact with the public extension service.

Program DesignIt is generally believed that extension services before NEP Iwere ineffective and inefficient. The extension systemfavored progressive farmers, especially the larger, moreeducated, and male farmers. NEP I and NEP II sought torectify these biases. Both followed standard T&V principlesand used contact farmers and, later, contact groups as thepoint of interaction with the farming community. (Thedesign of NEP I and NEP II is summarized in Annex B.)

The design called for providing advice biweekly toabout 10 to 15 percent of the farmers, with informationabout up-to-date practices best suited to their specificconditions. Extension agents were to work mainly withthe contact farmers (henceforth, both individual andgroup contact farmers are referred to as contact farmers),but would involve as many other farmers as possible inthe demonstrations and discussions. Extension workerswould visit farmers regularly and receive systematictraining and technical support from research staff.

The initial focus of NEP I was to be on simplemessages, concentrating on a few important cropsand the most important crop production activities.The focus was also on low-cost improvements thatthe majority of farmers could afford. The implicitassumption was that once productivity and revenueshad increased, a farmer could graduate to morecostly technologicalcomponents with theadditional incomegenerated.

Access toInformationPublic extension hashistorically been animportant source ofinformation in rural Kenya. Farmers’ customary sourceof information on agricultural activities is the govern-ment extension service; there are few alternatives. Butalmost half the farmers, including contact farmers,think that information is less available now than it was10 to 15 years ago (see figure 4.1), and less than 30percent think it is more available now. While these

Farmers’ customarysource of information onagricultural activities isthe governmentextension service; thereare few alternatives.

18


FIGURE 4.1. THE AVAILABILITY OF INFORMATION:NOW, RELATIVE TO 10–15 YEARS AGO

FIGURE 4.2. NORMAL PLACE AND FREQUENCY OFEXTENSION AGENT–CONTACT FARMER MEETINGS

0

10

20

30

40

50

60

More Less Same Not sure

Contact farmers

Noncontact farmers

Percent

Source: OED survey.

0

5

10

15

20

25

30

35

Own field Other'sfield

Baraza Co-opSociety

Other

12+ per year

6–11 per year

1–5 per yearOther/irregular

Source: OED survey.

perceptions do not necessarily provide an accuratecomparison with conditions before the projects wereimplemented, they do indicate that the availability ofinformation declined over the life of the two projects.

The Contact Farmer ApproachContact farmers made up about 9 percent of thesample. While this proportion is a bit lower than thepreferred 10 to 15 percent, it is reasonable. However,the data also show that the more educated farmers hada significantly higher probability of being selected as

contact farmers. Being awoman farmer loweredthe probability, althoughthe result is only weaklysignificant (at the 10 per-cent level). This is con-sistent with the findingsof the beneficiary assess-ment and PPA cited inChapter 3, and it showsthat at least some of thebiases of the previous

system continued to affect service.The poor quality of contact is reflected by the finding

that only 22 percent of even the designated contactfarmers meet as often as once a month, much lessbiweekly, as prescribed. Less than a third of the contactfarmers normally meet extension agents in their own ortheir neighbors’ fields (see figure 4.2). The most strikingfinding is that, even on a monthly basis, only about 7

percent of the contact farmers meet extension agents asplanned—that is, regularly, in either their own or aneighbor’s fields, and at least once a month. Following theprojects’ design, using monthly meetings as the norm,about 20 to 30 percent of the population should be inregular contact with extension services. In the entiresample, however, only about 2 percent of the farmersregularly meet with extension agents. Considering thatextension concentrates on a few chosen farmers, this lowlevel of contact is highly unsatisfactory.

The low frequency of farmer-extension agent con-tact is not a statistical artifact of this survey or ofcurrent times. The 1990 ATD survey obtained almostidentical results (Bindlish and Evenson 1993). Even in1990, before the start of NEP II, only about 2 percent ofall farmers were meeting extension agents as planned(only 3 percent of all farmers were meeting extensionagents monthly in any setting in 1990). These resultsindicate the poor efficacy of both NEP I and NEP II indelivering extension advice in Kenya.

The efficacy of the contact farmer approach alsodepends on indirect dissemination through demonstrationand spread effects. The T&V plots established in thecontact farmers’ fields were expected to have strongdemonstration effects on other farmers. But these effectswere likely rather limited, because less than a third—31percent—of noncontact farmers even know of a contactfarmer in their neighborhood.1 The alternative, thespreading of information verbally from contact tononcontact farmers, is also likely to have been limited by

Only about 7 percentof the contact farmers

meet extensionagents as planned—

that is, regularly, ineither their own or a

neighbor’s fields, andat least once a month.

19

E f f i c a c y : T h e Q u a n t i t y a n d Q u a l i t y o f C o n t a c t

FIGURE 4.3. EXTENSION STAFF ALLOCATION AND POVERTY BY DISTRICT

apparently poor communication between the two groups.Among the few noncontact farmers who know of acontact farmer, 58 percent report having received advicefrom the contact farmer on any occasion, and only 22percent (that is, 8 percent of all noncontact farmers)receive advice regularly. However, a significantly higherproportion of the same farmers—86 percent of those whohave ever received information and 93 percent of thosewho receive it regularly—also indicate that they discussgeneral agricultural information with other farmers. Theflow of information from contact to noncontact farmers isclearly not working as well as might have been expected.2

Contrary to the projects’ objectives, the methods thatcurrently appear to be most widely used are the samemethods that were popular before the projects began:most farmers, including contact farmers, rely on barazasand other forums to meet their extension agents. Thisoutcome is significantly at odds with the intent of T&V,which seeks to bring about a qualitative change in thenature of extension contact with farmers by moving awayfrom barazas and other public meetings, as they providelittle opportunity for interaction. Public meetings relylargely on exhortation and are useful for broadcastingsimple messages, but they are not conducive to effectivelearning or substantive exchanges on technical problems.

System PerformanceIt is important to consider the change in the overallefficacy and efficiency of extension under NEP I andNEP II, irrespective of the method used. Outreach is

typically measured by such indicators as whether or notfarmers have met an extension agent, which includesany type of contact, and the time of first contact.Although these measuresare poor indicators ofeither the quality or theeffectiveness of extensionservices, they are usedhere because they allowsome comparison of sys-temic outreach beforeand after NEP I.

Since the start ofNEP I, the proportion offarmers who have metextension workers has increased, and the increase issignificantly higher for newer than for veteran farm-ers.3 However, the increase in outreach is almostdirectly related to the increase in staff numbers,indicating little or no gain in staff productivity orsystemic efficiency.

BiasesThe pre-NEP I extension system was known to bebiased in favor of the more progressive farmers andareas of higher productivity. Statistical tests show thatsome of the earlier biases against women, smallfarmers, and farmers living far from access roads havebeen rectified, largely as a result of the expansion ofservices to previously neglected areas.4 But new biases

Contrary to projectobjectives, themethods that currentlyappear to be mostwidely used are thesame methods thatwere popular beforethe projects began.

20


FIGURE 4.4. IMPACT OF THE SUSPENSION OFNEP II DISBURSEMENTS

have appeared, favoring more educated farmers,higher-potential zones, areas closer to markets, andareas closer to Nairobi. These findings are consistentwith the headquarters-centric development of Kenya’sextension system. The bias against relatively poorerareas, however, has been maintained (see figure 4.3).

Farmer Assessment of Extension AdviceWhile less extension advice is delivered than the projectsexpected, the farmers’ approval rating of meetings andmessages they receive, measured in terms of “usefulness”and “applicability,” is a very high 86 percent.5 Even so,few farmers, about 40 percent, have actually applied theextension agents’ recommendations. More important, themajority of even the contact farmers—51 percent—havenot applied the recommendations. This disparity betweenthe farmers’ positive assessment of the recommendationsand their reluctance to apply them is discouraging.6 The

findings indicate that whilefarmers think that the mes-sages are probably good,they are just not meant forthem, which calls intoquestion the relevance ofthe advice. There is also amismatch between the rec-ommendations that thefarmers find most appli-cable, interpreted as theeffective supply of infor-mation—generally simpleor unsophisticated agro-nomic practices—and their

demand for information about more sophisticated input-application or intensification issues.7

The Suspension of NEP II’s DisbursementsFinally, an incident with potential consequences for thisevaluation was the suspension of NEP II’s disburse-ments in 1996. Although the survey was conducted ninemonths after the suspension was lifted, providingsufficient time for the system to get back on course, thefarmers’ survey responses could reflect the adverseeffects of the suspension on extension activities. Thedisruption of services also provides the rare opportu-nity for a counterfactual to evaluate the currenteffectiveness of the system.

Since contact farmers are the most likely to beaffected by the disruption of services under the T&V

system, their responses are particularly significant. Themajority of the contact farmers, or 60 percent, did notnotice any change in the delivery of extension services in1996, despite the disruption in the flow of funds for almostthe entire year (see figure 4.4.).8 As may be expected, theproportion of noncontact farmers who reported no changewas considerably higher, at 88 percent. Whatever meth-odology the extension system uses, the finding that themajority of the farmers did not notice any change raisestwo possibilities. One is that services were continued asusual by substituting nonproject funds, calling intoquestion the additionality of NEP II funds. The other,more likely, possibility is that the frequency of interactionamong the contact farmers, as for most farmers, wasgenerally low, so that few noticed a disruption in servicedelivery. This more likely scenario indicates the lack ofeffectiveness of the current extension system.

ConclusionsThe key finding is that the contact farmer/groupapproach, central to the design of NEP I and NEP II, isnot working as anticipated. There is no apparentimprovement in either the qualitative or the quantita-tive aspect of the interaction between extension agentsand farmers compared with the assumed pre-projectsituation. More specifically, there has been no improve-ment since 1990, when NEP II started, and even thenthe approach was performing very poorly.

Farmers think that information is less availablenow than it was 10 to 15 years ago. The nature of

0

10

20

30

40

50

Contact farmers Noncontact farmers

Noticed any change

Reported reduced visits

Percent

Source: OED survey.

The majority of thecontact farmers did

not notice anychange in the

delivery of extensionservices 1996,

despite thedisruption in the flow

of funds for almostthe entire year.

21

meetings between extension agents and farmers is notof the quality that was expected and appears to be nobetter today than it was before the projects began. Theextension agents have started using alternative meth-ods, and many appear to have reverted to the oldmethods, particularly barazas, to reach the farmers.For relatively simple messages, this approach may bemore cost-effective. For advice on more complexpractices and for solving problems specific to indi-vidual farmers, however, this method is unlikely to bevery effective. While systemic outreach has increased

and some of the biases of the previous system have beenrectified, this has largely been the result of an increasein staff numbers, rather than improvements in staffproductivity. The relevance of the advice that agentsdeliver is apparently limited, judging from the failureof the majority of farmers to apply the recommenda-tions, and it is not responsive to their needs.

That a service disruption of almost a year wentlargely unnoticed by contact and noncontact farm-ers alike suggests the lack of effectiveness of thecurrent system.

E f f i c a c y : T h e Q u a n t i t y a n d Q u a l i t y o f C o n t a c t

23

55

I

OutcomesFarmer Awareness and Adoption

Extension’s role becomesimportant when thenormal process ofdiffusion is too slow.

n the causal chain from investment in extension to the desired impact, farmers’ aware-

ness and adoption of technological components are important indicators of extension

services’ proximate impact, and they provide a backdrop for assessing their potential

economic impact.1 Impact will surely be limited if extension is unable to appreciably increase

the level of farmer awareness. Further, despite awareness, the potential benefits in the form of

increased productivity will be limited if the farmers do not adopt the recommendations.2

The degree of non-adoption reflects, in part, thequantity and relevance of extension advice, especiallygiven the technical, marketing, and resource con-straints that farmers face.3

Information is typically diffused through a number ofchannels, extension services being only one. Extension’srole becomes important when the normal process ofdiffusion is too slow. It becomes particularly important inconveying more complex information, such as the typeand quantity of fertilizer to use with a particular crop or anew crop variety, and in solving problems specific toindividual farmers or local areas, such as pest control orsoil micronutrient deficiency. In the T&V approach,selectively “infecting” contact farmers with new informa-tion is expected to speed up the usual rate of diffusion.4

Working Paper 2 in support of this study, which dealswith awareness and adoption of extension messages,presents details of the data and analytical methods used,and the results. The main findings are summarized here.The analysis in this chapter deals with recommendationsfor cropping activities.5

AwarenessAll farmers have heard maize-related messages, most areaware of the messages on cash and minor food crops, andabout two-thirds have heard of the crops currently beingpromoted by exten-sion. The proportionof those who areaware is higher in themore productive dis-tricts and where thecrops have a relativelylong history. Thus, at a very basic level, messages appearto be reaching the farmers, with some room for improve-ment for new crops.

Awareness of simple agronomic recommendationsis high, but falls significantly as practices increase incomplexity. Surprisingly, the levels and pattern ofawareness among contact farmers are similar to thoseamong other farmers. The lack of awareness ofcomplex messages among contact farmers is disap-pointing, considering that intensive, face-to-face inter-

24


FIGURE 5.1. FARMERS’ AWARENESS OFSIMPLE-TO-COMPLEX MESSAGES, 1982–97

action has an advantage over other methods of exten-sion in delivering such advice.6

During NEP I and NEP II, farmers’ awarenessappears to have increased only modestly (see figure5.1). More important, the pattern by activity has

remained virtually identi-cal. The simpler mes-sages, for which the cur-rent level of awareness ishigh, were already distin-guished by high aware-ness before NEP I.7 For therelatively complex prac-tices, low levels of aware-

ness have persisted. For most activities, change wasrelatively greater during NEP I than during NEP II.

Awareness is relatively high for maize-relatedmessages, but it is significantly lower for other crops,especially cash and nontraditional crops.8 Even formaize, however, the difference between simpler andmore complex messages persists. Given the high levelsof awareness of simpler messages, the marginal returnsto additional efforts at extending them are likely to below. Data from the current survey, as well as findings

from the 1990 survey, suggest that efforts to extendawareness of simple messages during NEP II are alsolikely to have had limited payoffs.

For farmers who are aware, government extensionis a sizable, but not the largest, direct source ofinformation. This is to be expected, since extension canreach only a small proportion of its client population.But even among contact farmers, who meet extensionagents the most regularly, fewer than half cite exten-sion as their source information. This is consistent withthe earlier findings that even most contact farmers donot meet extension agents regularly.

An intertemporal analysis of information sourcesreveals that even before NEP I, public extension was themain source of information for spacing and the morecomplex practices. As a source of other, simpler messages,extension’s role was relatively small, while friends andfamily were a more important source. During NEP I andNEP II, extension’s share as the main source of simplemessages increased significantly, but its share for complexmessages fell (particularly during NEP II). Equallysignificant is the increase not only in the private sector’sdelivery of messages on complex practices, but also thatof other sources such as specialized services, cooperativesocieties, and youth clubs for simpler practices. Thesetrends reflect the continued focus on simpler messagesduring NEP I and II. They also reflect the dynamism of aninformation system that is undergoing a transition, withnonpublic sources becoming increasingly important pro-viders of information.9

AdoptionThe patterns of adoption follow those of awareness. Ingeneral, the levels at which farmers adopt agents’recommendations, except for some simpler practicesinvolving planting time and weeding, are very low.Less than a quarter of the sample has ever tried any ofthe recommendations for the complex practices. Abreakdown by current and past adopters shows that theproportion of farmers currently applying recommenda-tions is almost negligible for more complex practices(see figure 5.2). Not surprisingly, current adoptionrates are relatively higher for maize than for othercrops. However, a comparison with ATD’s surveyresults, which is feasible only for maize practices,indicates that the levels of adoption have remainedalmost the same since 1990.

As expected, among the reasons that recommenda-tions are not adopted, or are discontinued, farmers cite

0

10

20

30

40

50

60

70

80

90

1 2 3 4 5 6 7 8 9 10 11 12 13

Activity complexity

199719901982

Note:

Source:

The simplest of the 13 types of messages is at the far left of thehorizontal axis, the most complex at the far right. The activities arecrop variety (1), planting time (2), spacing (3), seed rate (4), weedingtime (5), weeding number (6), basal fertilizer type (7), basal fertilizerquantity (8), top-dress fertilizer type (9), fertilizer quantity(10), fertilizer time (11), chemical plant protection (12),and cultural plant protection (13).

OED survey.

top-dresstop-dress

Percent aware

For the relativelycomplex practices,

low levels ofawareness

have persisted.

25

FIGURE 5.2. FARMERS’ ADOPTION OF EXTENSIONRECOMMENDATIONS, 1982–97

O u t c o m e s : F a r m e r A w a r e n e s s a n d A d o p t i o n

lack of funds most often. A sizable proportion of thesample, 40 percent, report their reasons for not adoptingrecommendations as resource constraints, including landand labor. However, an almost equally substantialproportion of the sample, 34 percent, cites reasons thatcould be addressed through proper extension advice.

The most significant finding is that a very largeproportion of those who are aware have adopted thepractices—more than 80 percent for even the morecomplex recommendations. Thus, while credit andresource constraints may be important factors, theprimary constraint on the adoption of recommendedpractices is lack of information.10

Statistical Tests of Extension’s ImpactThe statistical analysis was designed to establish thefactors that influence the probability of awareness oradoption of each recommended practice.11 (Approachesand results are detailed in Working Paper 2.)

The main finding is that the current supply ofextension cannot be associated with a greater probabil-ity of either awareness or adoption of individualextension messages. However, the supply of extensionin 1982 continues to have a strong positive impact oncurrent awareness and adoption. These results suggestthat information diffusion has proceeded at its ownpace; the impact of NEP I and NEP II is not apparentfrom the current data.

The test of the efficacy of the contact farmer

approach also yields disappointing results. Being acontact farmer increases the probability of awarenessonly for spacing and cultural plant protection practices(with the contact variables treated as exogenouslygiven). Allowing for the endogeneity of the measuredcontact variable (that is, with instrumental variables),contact farmers have a higher probability of awarenessfor only 5 of the 13 messages tested.12

Among other factors, social capital increases theprobability of awareness of simpler messages, whilefarm size has the same effect for complex messages.Education has positive short- and long-term impacts—primary education for simple practices, and highereducation for complex practices. Better infrastructure (inthe form of roads) increases the probability of awarenessfor most practices, while the distance to Nairobi and off-farm work generally have a negative effect.

The adoption analysis tried a new specification todirectly estimate delayed impacts, but it fails to show apositive influence of extension activities since 1982.13

The initial stock of knowledge—that is, at the start ofNEP I—has a strong positive influence for everypractice, but the subsequent supply of extension doesnot show any significant or systematic impact. Theonly significant impact discernible is the negative effecton the adoption of complex practices for the early yearsof NEP I.

ConclusionsBoth the descriptive andstatistical analyses showthat the focus of theKenyan crop extension ser-vice has remained on pro-viding advice on simpleagronomic practices. Thepre-1982 bias in favor ofthe simpler messages andmaize has continued. Thedata do not reveal anysignificant correction of this bias. The evolution in thelevels of awareness and adoption also suggests that theunderlying dynamics of the diffusion process have beenlittle influenced by extension activities. Thus, the lesssophisticated messages, which are amenable to quickdiffusion through informal communication channels orcasual contact, and messages that have been known fora long time have continued to spread. The less wellknown and the more sophisticated have lagged behind,

A very largeproportion of thosewho are aware haveadopted thepractices—morethan 80 percent foreven the complexrecommendations.

26


with low levels of awareness and adoption.While the focus on simpler messages and the

primary subsistence crop, maize, may have been justi-fied in the early years, the benefits of such efforts areclearly bounded. For continued increases in productivity,a graduation to more sophisticated practices would bethe natural path of evolution in an effort to intensifyagricultural production. Such a transition, however, is

not evident from thecurrent data, either inawareness or in adop-tion of the more com-plex practices. Thiscould be a result ofeither a poor focus ofextension efforts or alack of expertise indelivering the morecomplex messages, orperhaps both.

The continuedemphasis on simplepractices and maize,

despite seemingly high levels of awareness of thesepractices before the project period, and the lack ofassociation of post-1982 extension with adoption ratessuggest that the potential impact of NEP I and NEP II

on agricultural productivity may be limited. This isapparent, at least for NEP II, from a comparison withthe results of the ATD’s survey of the same population,which shows that adoption rates have remained virtu-ally unchanged since 1990.

This raises concerns about the cost-effectiveness ofthe T&V approach as applied in Kenya, particularlygiven its considerable cost. Because the major share ofthe system’s cost stems from its focus on face-to-faceextension, which is best suited to delivering moresophisticated and context-specific advice, the preoccu-pation under both NEP I and NEP II with simplemessages clearly indicates that the design features ofthe projects were not fully exploited in Kenya. This isparticularly noteworthy for NEP II. At its start, it wasclear (from the ATD study) that most farmers wereaware of, and a large proportion had adopted, thesimpler messages for maize.

The economic justification for the investment inextension under NEP I and NEP II—whether the returnsto extension justify the costs incurred—requires anestimation of the actual impact of extension on agricul-tural productivity. That exercise is the subject of the nextchapter. The findings in this chapter indicate that it isunlikely that the potential impact of the T&V system asimplemented in Kenya since 1982 was significantlygreater than that of the extension system it replaced.

Because face-to-faceextension is best suited

to delivering moresophisticated and

context-specific advice,the preoccupation with

simple messages clearlyindicates that the design

features of the projectswere not fully exploited.

27

66Results IFarmer Efficiency andProductivity Change

T he impact of extension is most directly measured by relating it to farm productivity.

Changes in productivity can result from improved efficiency or technical change.

Productive efficiency is a measure of the farmers’ level of skill and knowledge, often

termed managerial skills, in producing the most with a given set of production inputs, or in

producing a given level of output with minimal inputs. Technical change is a result of

technological advance—for example, improvements in the quality of inputs. Extension has

an important role in both production efficiency andtechnological change: in the first by imparting knowl-edge and advice on best practices suitable to the localcircumstances to improve farmers’ skills, and in thesecond by disseminating information on the latesttechnological advances.

This chapter presents an investigation of farm-ers’ levels of efficiency in crop production at twopoints in time using the data envelopment analysis(DEA). DEA is a nonparametric method that allows aranking of households by their level of efficiency.Also measured is the change in farm-level productiv-ity between 1982 and 1997, using the Malmquistindex. (A detailed description of the technical aspectsof the analysis, data considerations, the assumptionsmaintained, and the results produced is presented inWorking Paper 4 in support of this study.) Theanalysis uses the 1982 RHBS data to describe farmcircumstances before the projects began, and 1997OED survey data to elucidate the current situation.The two surveys have 285 households in common;

these are used to calculate the Malmquist indexes ofproductivity, efficiency, and technical change.

The data for 1982are limited, but didyield some input-basedtechnical efficiencymeasures. The data for1997 are more com-plete and allow theestimation of overallcost-efficiency and itscomponents—the tech-nical, scale, andallocative measures—to help identify thesource of inefficiency.1

What is measured isrelative efficiency—that is, each observa-tion is measured against the best practice, orproduction frontier, generated from within the sample.

Extension has animportant role to play,both in improvingproduction efficiencyand in promotingtechnological change. Itcan impart knowledgeregarding best practicesfor a given circumstanceand disseminateinformation on newtechnology.

28


The low level of cost-efficiency implies that

the allocative efficiencyof households is quite

low, which indicates thatthe farmers are not using

economically optimallevels of inputs.

All observations are ranked against the most efficientfarmer in the sample.

Relative EfficiencyThe average farm was operating at a very low level oftechnical efficiency in 1982 (31 percent), and althoughthere was some improvement, relative efficiency wasstill quite low in 1997 (45 percent). The scale offarmers’ operations has improved, but a majority stilloperate at a suboptimal scale; statistical tests fail toreject the hypothesis of nondecreasing returns to scale.

A comparison of costor allocative efficien-cies between 1982 and1997 is precluded by alack of price data for1982. For 1997, theresults show that theaverage level of eco-nomic (cost) efficiencywas very low (15 per-cent), much lowerthan technical effi-

ciency. Thus, even with the current level of technology,it appears that a simple change in the input mix, to onethat is more economical given the current marketconditions, would offer farmers the potential for signifi-cant savings.

Field consultations.

The measures above, calculated over all regions,are useful in putting the overall picture in perspective.Since regions vary in their productive potential andagroecological endowments, district-specific measureswere also calculated for 1997. As expected, the averagelevel of technical efficiency increased (with an averageefficiency level of 69 percent), but the economicefficiency was still very low (30 percent). The low levelof cost-efficiency implies that the allocative efficiencyof households is quite low, which further indicates thatthe farmers are not using economically optimal levelsof inputs.

A statistical analysis using the 1997 district-specificefficiency measures fails to reveal a statistically signifi-cant relationship between any of the efficiency measures(cost, technical, or allocative) and the supply of exten-sion services. In these tests, the cluster average technicalefficiency for 1982 is used to control for the regionaleffects for each location, and extension supply ismeasured as a weighted average of lagged extensionstaff–farm ratios. Qualitatively, extension has a smallpositive coefficient in the cost and technical efficiencyrelationships, but a negative coefficient for allocativeefficiency. The 1982 efficiency level is positive andsignificant for cost and technical measures, and positivebut weakly significant for allocative efficiency.

Although there is no clear-cut rationale for includ-ing district-specific effects, since efficiency measuresare calculated by district, the consequence is that theeffect of the supply of extension on technical efficiencyis still low and positive (0.056), but is now significantat the 10 percent level. Overall, the results do notchange much, but they do reinforce hints of mildextension effects on technical efficiency. Cost andallocative efficiency results do not change.

To test for the effectiveness of alternative extensionmethodologies, indicators for the normal place andfrequency of meetings were tested in a separate set ofregressions. The frequency of meetings has no influenceon efficiency. Those who interact with extension agentsat cooperative society meetings have a large butweakly significant effect on both cost and technicalefficiencies.2 It might also be expected that contactfarmers would be more efficient, but the current datado not show this for any measure of efficiency.

Among other variables, farm size has a strongnegative effect on cost and technical efficiency; that is,smaller farmers are more efficient. Distance to marketshas a significant negative effect on allocative effi-

29

R e s u l t s I : F a r m e r E f f i c i e n c y a n d P r o d u c t i v i t y C h a n g e

FIGURE 6.1. PRODUCTIVITY CHANGE, 1982–97: TRENDS IN CLUSTER AVERAGES

ciency, no effect on overall cost efficiency, and apositive effect on technical efficiency. This suggeststhat farmers farther from markets may be specializingin specific crops, and hence may be more efficient;those close to markets may be diversifying, and whilethey gain in allocative efficiency, they compromisetechnical efficiency. As may also be expected,agroecological variables variously affect all measuresof efficiency. Larger families and land fragmentationhave a negative influence on cost and technicalefficiency, and farmer age has a weak negative effecton technical and cost efficiency. Social capital has aweakly positive effect on allocative efficiency, whiledistance to dirt roads has weak negative significancefor both cost and allocative efficiency. Householdswhose heads have attained higher levels of educationalso have a lower cost efficiency, but the result is onlyweakly significant.

Productivity ChangeRelative measures for individual years do not indicatehow efficiency or productivity has changed. Thesechanges are measured with the Malmquist index forproductivity change, which is also decomposed intoindexes that measure technical and efficiency changes.The indexes are calculated by district to control forregional effects in production and the economic envi-ronment.3 The results show that, on average, produc-

tivity increased by 28 percent over the period 1982–97. This increase stemmed largely from positive techni-cal change, which raised productivity by about 56percent. The technical improvement, however, wasmoderated by a decline in efficiency of 31 percent, onaverage, relative to 1982levels. By district, themore productive districts(Kericho, Muranga, andTrans Nzoia) experiencednet productivity declines,while the others experi-enced substantial gains(with the largest gains inKisumu and Machakos,followed by Bungomaand Taita Taveta). Mostdistricts show technicalprogress, but declines inefficiency. The exceptions are Kericho, with no techni-cal change but a decline in efficiency, and Trans Nzoia,with significant technical regression but a modest gainin efficiency.4

The measures reveal some unexpected trends. Figure6.1 plots the linear trends in the cluster-level averages ofthe three Malmquist indexes.5 The data are sorted inascending order by the 1982 cluster-level average relativeefficiency. The trends show that clusters that had high

0

1

2

3

4

5

6

0

0.5

1

1.5

2

2.5

3

1982 Staff/farm ratio 1997 Staff/farm ratio

Malmquist index

Efficiency change Technical change

Index Staff/farm ratio (x 1,000)

Note:Source:

Clusters sorted by average 1982 efficiency scores.Results of DEA analysis using OED survey data.

The less productivedistricts haveexperiencedsubstantially greaterproductivity gains.Most districts alsoshow technicalprogress, butdeclines in efficiency.

30


average efficiency levels in 1982, generally in the moreproductive areas, gained less in total productivity, andsome may have regressed.6 The trend in efficiency issimilar, but shows relatively smaller changes. The techni-cal change trend is much flatter, but, again, negativelycorrelated with the 1982 level of efficiency. Without anymajor technological advances, these results show aconvergence across regions toward homogeneity in thelevel of productivity.

How does this relate to extension? Figure 6.1 alsoshows the trends of thestaff-farm ratios in 1982and 1997. Other ratiosare not shown since thesetrends are similar—thatis, positively correlatedwith the 1982 level ofefficiency. The 1990trend is steeper than thatof 1982, indicating thatthe allocation of front-line staff during NEP Igenerally favored the

more productive regions. During NEP II, recruitmentwas frozen. The decline in the slope of the 1997 trendmay thus reflect natural attrition in the frontline workforce. Productivity change is clearly inversely corre-lated with the allocation of extension staff.

ConclusionsThe results of this analysis, especially as summarizedin figure 6.1, are striking. The analysis shows thatthere has been little change in the areas that wererelatively more productive in 1982, while the otherregions have been catching up. This suggests that themore productive areas may have reached an upper

bound, and with little new technology forthcoming tosubstantially raise production, their productivity hasstagnated. However, the efforts of the extension ser-vices have been consistently focused on these areas. Atthe same time, a general lack of improvement inefficiency has meant that even in districts that haveseen technological progress, the overall potential forproductivity gains has been compromised.

Combined with the still very low overall efficiency(as measured by cost-efficiency) and the high degree ofvariation in efficiency among farmers, these resultssuggest that, despite room for improvement, the exten-sion resources have not been used as efficiently as theymight have. While extension may have contributed togrowth in the less-productive areas, its overall effective-ness appears to have been limited. The minor differencesin the cost, technical, and allocative efficiencies, eventhough the estimates are statistically insignificant, sug-gest that extension has generally concentrated on dis-seminating technical messages rather than helping farm-ers optimize their resource use or tailoring its messagesto the prevailing economic environment.

The lesson that emerges is that extension servicescould have been allocated more efficiently. If thepotential for technical gains had been properly assessed(especially in the main areas of maize production thathad already benefited substantially from past researchand extension efforts), a greater deployment of exten-sion staff in less-productive areas might have beenmore cost-effective. Considering the significant re-sources that are needed to sustain the current systemand MALDM’s extremely tight budget constraints, it isapparent that fewer resources could have been used toachieve the same results. And the returns to theinvestment in extension could have been much higher.

If the potential fortechnical gains had

been properlyassessed, a greater

deployment ofextension staff in less-

productive areasmight have been more

cost-effective.

31

77Results IIProduction Effects of Extension

Determining returns to an investment in extension requires an estimate of its impact on

agricultural production. Chapter 6 showed that the allocation of extension has been

biased in favor of the relatively more efficient—and likely more productive—areas.

This bias makes measuring the impact of extension difficult in a cross-sectional framework—that

is, with the use of data from one point in time—and necessitates the use of more reliable methods

using panel data. This problem was demonstrated in the context of ATD’s evaluation of the

impact of extension in Kenya (Bindlish and Evenson1993). Working Paper 3 in support of this studyreconsiders the results of the ATD study, and discussesthe technical details of the difficulty of interpreting itsresults.1 The main finding is that the high estimatedreturns in the ATD study are very sensitive to regionaleffects. At the same time, correcting for inadvertentdata errors makes the results less robust. The sensitivityof the results precludes any judgment that the returnsare positive.

To overcome the methodological limitations of across-sectional framework, the current study used amore robust method. It combined the 1982 RHBSdata and the 1990 ATD data with a fresh survey ofthe same households to develop a panel data set.(The technical details of the analysis and results arepresented in Working Paper 5.) The objective of theanalysis was to identify the impact of extension oncrop production by appropriately controlling for asmany unobserved factors as possible. Of these, theprimary concern was with the unobserved naturalproductivity effects and other inherent regionalsocioeconomic or agroecological effects.

A Fixed-Effects ApproachThe 1997 data confirm the problem with using datafrom a single cross-section. Statistical tests show thatwhile the marginal effects of variable production inputson farm production are stable with respect to alternativeregional and agroecological indicators, the coefficienton the extension variable is sensitive. To overcome thisproblem, the panel nature of the data is exploited.2

A difference modelcan control for the unob-servable regional andagroecological factors.Using this specification,a separate model is esti-mated for each of thethree two-year panels(1982 and 1990, 1982and 1997, and 1990 and 1997). In addition to thevariable production inputs, household characteristicsare included in differenced form, since these alsochanged over time. Varying weather conditions arecontrolled for by including farmer-reported crop perfor-mance indicators (normal or poor, relative to good);

Allocation ofextension has beenbiased in favor ofmore productiveareas.

32


these indicators, however, are available for 1990 and1997, but not 1982.

The remaining complication is the extension sup-ply variable. The data on staff-farm ratios, used tomeasure the supply of extension services in eachlocation, go back only to 1982. Thus, for 1997 and

1990, it is possible tomodel lagged impact ofextension, using weightedlags going back sevenyears each.3 For 1982,only the single-year mea-sure can be used. While itwould be desirable tohave data on the previousyears’ supply of exten-

sion, such data are not available. One way around theproblem is to assume that past extension efforts areembodied in the 1982 level of production. The post-1982 changes in the supply of extension thus helpidentify the impact of the new system. In addition, the1982 staff-farm ratio provides additional control forthe base level of extension supply. What the differencemodel measures, then, is the change in productivitythat can be attributed to changes in extension after1982.4 Extension supply can thus be modeled either asthe difference between the cumulative extension supplyfor one of the later years (1990 or 1997) and the 1982supply, or by allowing the coefficient to vary over timeby including both variables independently.

FindingsThe results detailed in Chapter 6 point to the impor-tance of distinguishing between the program effect—or,more precisely, efficiency in the allocation of extensionresources—and the direct extension effect. Given thatthe allocation of extension staff has been, and continuesto be, biased in favor of the more productive areas, andthat growth in agricultural productivity has beenuneven from area to area, it is necessary to control forthe initial conditions in order to properly identifyextension’s impact.

To control for the effect of initial conditions, amore flexible approach is used than a simple differencemodel. The analytical model is extended to yield anempirical model that allows a distinction betweenprogram efficiency, or the “program effect,” and theimpact of extension, or the “extension effect” (see

Working Paper 5). Average cluster-level yields for thebase year are used as a proxy for the initial condi-tions.5 Applying the model to the three panels confirmsthat the resources have been allocated inefficiently. Thebase year yields have a significantly negative coeffi-cient. The extension effect, however, is not significantin either the production function or the reduced-formsupply function specifications for any of the panels. 6

To confirm the hypothesis of the confounding effectof initial conditions on the impact of the supply ofextension services, the simple difference model for allspecifications yields a result similar to that described inChapter 6. That is, in the pure difference specification,productivity change is negatively and significantlycorrelated with the extension variable. A naïve inter-pretation of this result would be misleading, since itsuggests that extension has a negative impact onproduction.

ConclusionsThe main finding of both this analysis and the materialpresented in Chapter 6 is that extension resources inKenya have been inefficiently allocated and poorlytargeted. At the same time, once the initial conditionsare controlled for, a statistically significant impactcannot be established for extension.

It appears that the less productive farmers andareas have been catching up as new technology reachesthem. And while it is likely that extension has played arole in extending these technologies, this cannot befirmly established with the data in hand.

It is likely that more rational allocation ofresources would have achieved the same results morecost-effectively. The lesson that emerges is that Kenyaneeds to build a flexible and responsive system. Withlittle new technology forthcoming, as in the case ofmaize, it is not economical for extension to maintain ahigh-level presence. Instead, reaching new areas orfarmers previously not covered by extension serviceswould have a greater marginal impact on both produc-tion and poverty. What is needed, perhaps, is a “smart,”flexible system that responds to imbalances in thedelivery of information, targeting existing or emerginggaps between average and best practices. At other times,it would be more efficient to keep a leaner presence tomaintain a local equilibrium, rather than blanketing allregions with intensive coverage at all times.

The less productivefarmers and areas

have beencatching up as

new technologyreaches them.

33

88

T

Client FocusFarmer Valuation of Extension Benefits

he benefit of public service to the ultimate beneficiaries is a critical issue for policy. If the

beneficiaries were paying clients, the value of the services provided to them would

simply be their market price. For most public services, however, there is no market.

Traditionally, extension advice has been free because of its nature as a public good (low

excludability and rivalry), with substantial positive externalities. But with limited resources, the

issue of the efficiency of allocation across a number of possible public goods remains, and hence

the need to measure the benefits of the services provided.A measure of benefits would also allow consideration ofcost recovery measures. Even if it is only partial, costrecovery has several benefits: it provides appropriateincentives, and therefore accountability and client respon-siveness; it brings budgetary respite; and it promotespluralism by allowing alternative providers, particularlyprivate suppliers, to enter the market.

With respect to the relationship between extension andpoor farmers, some pertinent issues are their demand foradvice, their willingness to pay for it, and their ability toafford the payments. Theoretically, the upper limit of whatan individual would be willing to pay for a service wouldbe the maximum private net benefit derived from it. Thisbenefit can be estimated either directly or indirectly. Oneindirect method is to estimate benefit from the impact of theservice on a farmer’s productivity, as discussed in Chapter6. This method, however, assumes that the service isdelivered, and delivered in a manner that is efficient andeffective. More important, it does not reveal whether thefarmer is willing to pay for it. A direct method is thecontingent valuation method (CVM), which elicits fromfarmers their willingness to pay for the service, giving someidea of what they perceive to be its benefits.

The approach this evaluation uses is briefly dis-cussed in Annex F. Working Paper 6, on farmers’willingness to pay, presents a more detailed discussionof the survey design,an important part ofCVM; the tests andcontrols to check theconsistency and reli-ability of the farmers’responses; and the de-tailed descriptive andanalytical results.1

The key findings aresummarized here.

Desired Frequencyof VisitsSome farmers (4 per-cent) indicated thatthey do not want anyextension advice, and some (another 4 percent) do notwant the current service to continue. For the remainder,the median number of desired visits is three each year,with a modal value of two.2 More than two-thirds of

Cost recovery, even ifonly partial, providesappropriate incentives,and hence accountabilityand client responsive-ness; it brings budgetaryrespite; and it promotespluralism by allowingalternative providers,particularly privatesuppliers, to enter themarket.

34


the farmers want fewer than one visit every threemonths. Even among contact farmers, almost half wantto meet the extension agent no more than once everythree months. These responses suggest that the norm ofbiweekly or even monthly visits under NEP I and NEPII exceeds what farmers want.

Willingness to PayOverall, a small proportion (9 percent) of the farmers

who would like to receive advice (including 12 percent ofthe contact farmers) are not willing to pay for it. Morethan half of the farmers are willing to pay individually,

while the rest preferto pay in a group.3

The mean willing-ness to pay (WTP)for individual con-tributors is Ksh 67for each visit, andfor group contribu-tors, it is Ksh 51 foreach visit; the meanWTP is Ksh 60. At

the time of the survey, the daily wage rate for agriculturallabor was Ksh 60. The farmers who are unwilling to pay(at all, or as individuals) most frequently cited lack offunds as the reason.

The average total annual WTP is Ksh 346, with amedian of Ksh 160. The range is quite wide, from 0 to8,640, but the mean WTP is significantly different from0 (standard error of 32). The interquartile range,however, is tighter, between 60 and 360.

Factors That Influence Willingness to PaySystematic variation in WTP by socioeconomic oragroecological characteristics, or with the alternativeextension methodologies currently in use, is importantfor policymakers to more effectively and efficientlytarget future services. It is also important in determin-ing the perceived benefit from the services that areavailable to different farmers.

The influence of existing extension services onWTP is viewed from three perspectives. One is theeffect of its current supply of extension services. Thishas a small positive effect, but does not attain statisti-cal significance in any specification tried in theanalysis. Next, to test the influence of alternativemethodologies, variables indicating the households’normal meeting place with extension agents were

tested. Households that meet extension agents either intheir own fields or in a baraza do not have asignificantly higher WTP than those who do notnormally meet extension agents or who meet them veryinfrequently. Farmers who normally meet at a coopera-tive society have a negative and significant effect.Finally, a variable indicating whether a farmer is acontact farmer was tested, but it failed to attainsignificance.

Listening to radio programs has a strong positiveand highly significant influence on WTP. This could bebecause hearing information on the radio encouragesfarmers to seek additional or more detailed informa-tion—enough to convince them that it would beworthwhile to pay for it. It may also reflect theircurrent lack of access to such additional information,and their consequent inability to follow up on whatthey hear on the radio.

The results also show that WTP is significantlyhigher among those who are willing to pay in a group.Both gender and education make a difference: house-holds headed by women and households whose headshave a primary or, especially, higher-level educationhave a significantly higher WTP. Other variables donot appear to influence WTP. Social capital variablesshow weak but mixed effects. The membership ofhouseholds in groups (either of farmers or theirspouses) increases WTP, but a higher incidence ofgroups within a location has a generally negativeeffect. Perhaps the latter reflects the benefit of havinggreater access to information, while the former re-flects the benefit of collective action. Amonginfrastructural variables, only access to dirt roadsaffects WTP (households that live farther from a dirtroad are willing to pay more), although the effect isweak. Households living in lower-potential zoneshave a lower WTP, as do those living on hills andundulating terrain.

ConclusionsA significant proportion of farmers would like toreceive extension services and are willing to pay forthem. The perceived benefit, however, as reflected inthe total amount that farmers are willing to pay, is wellbelow what the government currently spends per farmon extension services. Also, the frequency of visits thefarmers desire is much lower than was presumed inNEP I and NEP II designs. An econometric analysissuggests that WTP is not related systematically to the

A vast majority of farmersare willing to pay for

extension services. Ofthese, over half are

willing to payindividually, the rest to

pay in a group.

35

C l i e n t F o c u s : F a r m e r Va l u a t i o n o f E x t e n s i o n B e n e f i t s

level or methods of extension currently in use. Oneinterpretation of these findings is that the farmers’ WTPreflects an unmet demand for services, which is alsoindicated by the close, statistically indistinguishablewillingness to pay of contact farmers and those whohave never before received advice. Another indicator isthe desired frequency of visits, which is approximatelythe same across all categories of farmers. Finally, thestrong influence on WTP of farmers’ listening to theradio probably reflects their inability to follow up onthe information they get through that medium.

The study results have important implications forthe design of future extension services. The mostimportant is the implication for cost recovery and thepossibility of incorporating an endogenous qualitycontrol mechanism in the delivery system. Consideringthat even those who do not regularly receive extensionservices, or who meet agents only infrequently and inpublic gatherings, are willing to pay clearly indicates

Farmers valueagricultural adviceand are willing to shareits cost.

that farmers value agricultural advice and are willingto share its cost. It also reflects the current lack of analternative source of information.

At the same time,the uniformity of thelevel of demand (thatis, the frequency of vis-its) and WTP suggeststhat it would be moreefficient to cover alarger number of farmers, but with lower intensity and ahigher quality of contact. The large positive influence ofradio programs on WTP suggests that complementaryextension approaches should be exploited for potentiallysignificant synergy. It is also possible that radio programscould be used to whet the farmers’ search for information,which could then help promote the provision of morespecialized private extension, or extension for a fee.

37

99Conclusions andLessons

he rationale for providing extension services in Kenya is still relevant, but the evidence

suggests that the extension approach used by NEP I and II was not efficacious.

The overall record of the T&V extension system implemented in Kenya has been disap-Tpointing. The extension approach adopted by NEP I and NEP II has not proven to be effective,

and the current system is not sustainable. Although the system’s geographical coverage, research-

extension linkages, and staff skills have been improved by the projects, the outreach of the system

is low, and the interaction between the extensionagents and the farmers is qualitatively well belowwhat was anticipated.

The evaluation reveals that there is an unmetdemand for extension services, and the farmers valueaccess to advice enough to be willing to pay for it.Despite the substantial scope for improvement, how-ever, the data do not provide evidence of any signifi-cant impact of the current extension system on farmerefficiency or crop productivity. On the contrary, allapproaches indicate that the current institutionalarrangements have been ineffective in delivering themuch-needed services to the vast majority of Kenyanfarmers. It is likely that NEP I had some beneficialimpact early in its implementation period. The ben-efits, however, appear to have been short-lived. Theavailable evidence does not indicate any significantimpact, even by 1990. The results do show thatextension resources have been allocated inefficiently.

The various estimates obtained in the evaluation’sanalysis show that a positive rate of return to theexpenditures on extension cannot be established. Fur-ther, the worth of the perceived benefits from thecurrent services, as indicated by the amount that

farmers are willing to pay for them, is well below whatthe government is currently spending per farm todeliver the services. The findings suggest that a morerational allocation of extension resources would havebeen more cost-effective.

Lessons and RecommendationsThe main lessons and recommendations to emerge

from this evaluation include the following.Targeting. The first lesson is the need for more

efficient targeting ofextension services tofocus on areas andgroups where the mar-ginal impact is likely tobe the greatest. Thiscalls for a more flex-ible, “smart” systemthat can identify thegaps between existingbest practice and average practice and allocatescarce resources more rationally. Further, the farm-ers selected for interaction should be more represen-tative of the local socioeconomic environment so

There is an unmetdemand for extensionservices, and thefarmers value accessto advice enough to bewilling to pay for it.

38


that more relevant advice can be delivered todifferent categories of farmers.

Information systems. Targeting calls for appropri-ate flows of timely and reliable information, and hencefor monitoring and evaluation (M&E). An importantlesson emerging from the Kenyan experience is the needto identify farmer demands and tailor the service to suitlocal technological and economic conditions and cir-cumstances. M&E is also critical in identifying thegaps and guiding the “smart” system for more efficienttargeting of services.

Intensity. Reflecting their current experience, farm-ers do not want to see the extension agent too often; andthere are not enough new technological recommenda-tions to sustain a high intensity of visits. It would bemore cost-effective to establish a leaner and less-intensive presence, but with wider coverage. It may bethat with improved quality of service, the demand willincrease. To be ready for such a change, it is impera-tive that a responsive and dynamic delivery system bein place (as in targeting, above).

Pluralism. A blanket approach, using a single oruniform methodology to deliver standard messages, islikely to limit the effectiveness and efficiency ofextension services. Younger, more educated farmers are

taking over from their parents; radio programs arepopular among farmers; and alternative providers arebeginning to emerge in rural Kenya. It would beadvisable to adopt a more cost-effective strategy thatexploits the synergistic effects of low-cost moderncommunications, demonstrations, printed media, andpartnerships with civil society and the private sector.This would leverage the resources to increase outreach,and is likely to have a greater impact with the same orlessened demand on government resources.

Client focus. The central focus of the institutionaldesign should be on empowering the farmer. Aneffective way to incorporate client focus is to consideralternative options that give a voice to the farmer,such as cost-sharing, farmer organizations, decen-tralization, and the like, as an integral part of thedelivery mechanism. Cost recovery (even if onlypartial), in particular, would be advantageous: itprovides appropriate incentives, addressing the issuesof accountability and quality control; it renders theservice more demand-driven and responsive; it pro-vides some budgetary respite; and it encouragesalternative providers. Such institutional arrangementsremain unexplored in Kenya.

ANNEXES

39

ANNEX A. BACKGROUND

BOX A.1. THE DIFFICULTY OF DRAWING INFERENCES FROM FIELD VISITS

M ost of thefield visitsfor this

evaluation werearranged through theextension service. Eachvisit typically entailedan entourage of missionmembers; resident mis-sion staff; ministryrepresentatives; provin-cial or district staff, orboth, often includingthe officer in charge;several subject matterspecialists; divisionalstaff; and the localfrontline staff. Thegroup usually arrivedin a motorcade of threeto five vehicles to visitwith farmers or groupsthat normally workedwith extension agents.Almost always the

more successful farmers,or those who had ben-efited from extension,were visited. The farmersvisited apparentlyreceived many missions,since most of them keptan impressive diary thatthe visiting “dignitaries”were obliged to sign.Such experiences areunlikely to be insightful,however.

The moral hazardconfronting the fieldextension staff wasrevealed during a visitorganized by a bilateraldonor to view an alterna-tive extension approach.Most previous visits hadbeen to districts fundedonly by NEP, and mostdiscussions of the issuesand problems with the

extension services werepositive, pointing to thebenefits of the system.The visit organized bythe bilateral agency,however, was to a districtthat received both NEPand bilateral aid funds.At the start of the visit tothe district extensionoffice, the extension staffseemed uncomfortable.Their problem was thatthe “mission” includedrepresentatives of bothfunding sources, the localprogram head of thebilateral agency, and aBank staff member. Thedistrict officials confidedin the accompanyinglocal staff (some ofwhom had previouslyworked for the govern-ment service but were

now with the bilateralagency) about theirdilemma. The diffi-culty was in decidingwhich program todiscuss and, moreimportant, which pro-gram to praise, sincetheir approaches werevery different. Theproject staff resolvedthe issue by noting thatthe Bank staff wereactually from OED,and that the extensionstaff should feel freeexpress their feelingsabout NEP. Of course,the rest of the discus-sion revealed that theprogram funded by thebilateral agency wasthe program of choice.

Among the several goals often cited for agriculturalextension services, the most common is agriculturaldevelopment (Feder, Willett, and Zijp 1999). Theobjective of extension services is to sustainably increaseagricultural productivity by, among other things,expanding the knowledge farmers have about newcrops, crop varieties, inputs, and better husbandry andmanagement practices. The importance of science-basedtechnological advances in raising farm productivitymakes agricultural extension key to development, andhas brought about consistent Bank support for suchactivities in many of its borrowing countries. Over thepast two decades, the Bank has invested about US$4billion worldwide in extension projects. A large numberof these projects have used the T&V system of manage-ment (along the principles laid out by Benor, Harrison,and Baxter 1984).

In Africa, agricultural extension has been central tothe Bank’s development strategy (Cleaver 1993). Thestrategy for the new millennium, designed to “focus ona few selected national and thereby systemic programsof high impact,” also lists extension as a key area forBank support in Africa (World Bank 1997). In the past,this strategy largely relied on the T&V system, withnational programs in more than 22 countries designedto follow its guidelines (Venkatesan and Kampen 1998).

The effectiveness of the T&V system of extension,particularly its cost-effectiveness, has been subject tomuch debate. The central issue has been the institu-tional design and efficacy of T&V relative to alterna-tive mechanisms for delivering extension advice(Picciotto and Anderson 1997). Within the Bank, thedebate has been passionate, and often emotional. Thefocus has largely been on conceptual issues, but little

40


supporting evidence has been offered. The limitedevidence that is put forth usually comes from assess-ments made during field visits. As expected, support-ers of T&V generally make positive assessments; itscritics generally draw negative conclusions. Whilethere is probably some truth in all the assessments, thereality is largely obscured—most field visits areunlikely to be representative or unbiased, either in themanner in which the data are obtained or in theirinterpretation. The facts are also often colored bymoral hazard on the part of the local extension staffwhen dealing with so-called random donor visits (seebox A.1).

Despite the intensity of the debate, very fewattempts have been made to rigorously measure theimpact of T&V, or the lack of impact. Three notableexceptions attempted to estimate the returns to T&Vinvestments.1 All were conducted by the Bank, andincluded work in India (Feder, Slade, and Lau 1985),Burkina Faso (Bindlish, Evenson, and Gbetibouo 1993),and Kenya (Bindlish and Evenson 1993). Positive butvarying degrees of impact were found. All three studiesused survey data, but were subject to limitationsimposed by the available data. Other studies haveconsidered the effectiveness of the T&V approach inother settings—for example, Hussain, Byerlee, andHeisey (1994) in Pakistan—and the findings have beengenerally mixed. A number of studies of T&V’soperational aspects, most of them critical of theapproach, have failed to assess the full impact of theextension system.

The Kenya study (Bindlish and Evenson 1993), ofparticular interest here, was part of an effort by theAfrica Region to assess the impact of the large amountof development resources going to extension in Africa.The Africa Technical Department (ATD) undertook thestudy to evaluate the impact of the agricultural exten-sion projects it had supported in Kenya and BurkinaFaso in 1990.2 As noted, the studies estimated very highreturns to extension, especially in Kenya. The findingshave been controversial, however, because of theirvarious limitations, some of which the authors them-selves noted (Bindlish and Evenson 1993, p. 29).

At a time when many borrower countries werebecoming concerned with the high costs of the T&Vapproach, and concern was increasing within the Bankabout the development effectiveness of its extensionportfolio, the high estimated returns were greeted withmixed feelings, and even skepticism, in some quarters(World Bank 1994; Purcell and Anderson 1997).Nevertheless, since the evidence was based on house-hold survey data and formal statistical methods wereused, the estimates of high returns lent credibility to theclaims of T&V supporters. 3 The findings vindicatedthe Bank’s stated policy of using extension as a majorplank in the overall rural development strategy forAfrica (Cleaver 1993) and justified speeding up thealready rapid introduction of the T&V system inAfrica. At the end of 1997, 22 countries had a nationalextension program with a T&V system of manage-ment, with active Bank projects supporting a totalinvestment of more than US$700 million.

41

A n n e x e s

ANNEX B. THE DESIGN OF NEP I AND NEP II

The Bank introduced the T&V system of managementas a pilot in two districts in 1982.1 Following the briefpilot, the system was expanded to 30 of Kenya’s 41districts over a 3-year period, covering all high- andmedium-potential areas. NEP I was designed as thefirst phase of a longer-term institutional developmentplan: the T&V system was to be introduced, and thenimproved over time. It was primarily an institution-building project, and it was anticipated that externalassistance would be required for 10 to 15 years.

The project design followed standard T&V prin-ciples. Project activities would provide farmers regular,systematic, up-to-date advice on the farming practicesbest suited to their specific conditions. The programwas initially restricted to the crop extension service.Frontline extension workers (FEWs) would visit farmersregularly and receive systematic training and technicalsupport from research staff. Each FEW was to dividethe farm families in their jurisdiction (then anticipatedto be between 400 and 800 farmers) into 8 groups. Eachgroup was to be visited every fortnight (four one week,and four the next). Since it was impossible to visit everygroup member on any one day, five to ten contactfarmers were to be selected from each group. About 10percent of the farmers would thus be designated contactfarmers, and the extension workers were to workmainly with them, demonstrating practices that wouldbe followed in the next two weeks, and were to involveas many other farmers as possible in the demonstra-tions and discussions. The selection process was toentail an inventory of all farmers, identifying theparticularly poor farmers, whose progress was to bemonitored and evaluated.

The initial focus was to be on simple messages,concentrating on a few important crops and the most

important aspects of crop production activities, as wellas low-cost improvements that the majority of farmerscould afford. The implicit assumption was that onceproductivity and revenues had increased, the farmerscould graduate to more costly technological compo-nents with the additional income generated.

Technical officers and subject matter specialistswere to supervise and back up each FEW. Every twoweeks, the FEWs were to receive a full day of intensivetechnical training on the messages they were to deliverduring the following fortnight. The subject matterspecialists were to upgrade their knowledge and skillsthrough monthly training workshops attended byresearch scientists. By design, at the district level alone,the ratio of non-FEW to FEW staff was 1:3.2 Theproject also provided funding for transport to increasestaff mobility; allowances for field staff; audiovisualequipment; civil works to build office space where noneexisted; incremental operating costs; and the produc-tion, publication, and updating of extension manualsfor all staff by the Agricultural Information Centers(AICs).

NEP II sought to further strengthen extensionservices and support their expansion to uncoveredareas, including the dryer zones; provide funds toimprove staff transportation; foster the use of massmedia and communications; rehabilitate and refurnishFTCs; and promote links between research and exten-sion by funding transportation and equipment, allow-ing greater participation of research staff in extensiontraining sessions, demonstrations, and farm trials. Theproject’s goal was to effectively deliver technicalmessages tailored to the needs of smallholder farmers,especially women, and increase yields of both stapleand export crops.

43

A n n e x e s

ANNEX C. INSTITUTIONAL FEATURES

ManagementAn immediately notable feature of Kenya’s extensionsystem is that not only is monitoring and evaluation(M&E) nonfunctional, but even basic managementinformation is missing. Data are not readily availableon the number of extension staff, their operationalcapacities, or even on extension’s annual expenditures.While some of the underlying factors go beyond theextension department, and indeed beyond MALDM, itis apparent that NEP I and NEP II had no impact onthis important aspect of management. Overall manage-ment of the projects was also weak. Poor financialarrangements compounded the problem of inadequateresources, and poor implementation arrangementshave impeded the functioning of extension services.This proved particularly significant during NEP II,during which the management of extension servicesrested with a working group with insufficient authority;it was unable to coordinate the activities of theagriculture, livestock development, and veterinarydepartments. As a result, the management during mostof NEP II was ineffective. This is particularly signifi-cant since the staff appraisal report for NEP I promi-nently stated that T&V was first and foremost amanagement system.

Another shortcoming of the projects’ work ininstitutional strengthening is the continuing lack of astrategic vision or national policy for agriculturalextension. Several observers have noted that thisreflects a preoccupation of extension’s managementwith the modalities of the delivery mechanisms, with-out regard to policy, planning, or management of theextension services.1 The new institutional paradigmintroduced by NEP I helped increase the level of energythroughout the program in the early years because ofthe large influx of operational and development funds,the availability of new vehicles, payment of allow-ances, and significant training. These changes raisedmorale and motivated field staff, and the detailedimplementation program, with its clear chain of com-mand and well-defined bureaucratic staff assignments,dispensed with the need for policy or planning.

But with a deteriorating financial situation andineffectiveness of the services, efforts were begunrecently toward developing a national policy and aframework for the future development of extension inKenya. Progress so far, however, has been driven by

multifaceted pressures, including involvement of theBank and other donors in the context of the preparationof the Kenya Agricultural Sector Investment Program,as well as support from other Bank projects. Theresulting studies, as MALDM documents, reflect re-newed thinking about extension and deeper and candidassessments of the current state of Kenya’s extensionefforts. Progress toward developing a national policyhas nevertheless been slow, and a final policy paper hasnot yet been produced.

IncentivesAnother consequence of the mechanistic implementa-tion of the projects’ design has been inappropriateincentives, both institutional and individual. The“rules of the game” are key elements of institutionaldevelopment, especially for service delivery, as theydetermine the incentive structure. While the bureaucra-tization of the extension services cannot be attributed tothe projects, the hierarchical structure of the T&Vdesign has not improved the situation over that beforeNEP. The result is that the extension service is both top-heavy and headquarters-centric (supervisory staff isexcessive, with a large concentration in Nairobi).

Extension’s relative success and free flow of fundsin the beginning led to an intolerance of dissent andneglect of emerging problems. The perceived manage-ment benefits of the projects, in the monitorability ofproject outputs and accountability of staff (for example,in terms of number of visits, number of trainingsessions, and whether or not FEWs were strictlyfollowing their assigned routes) and the focus ondelivery of specific and well-defined messages, put inplace adverse incentives. As in any bureaucratic orga-nization, staff accounted to their supervisors, not to theclients, and what was monitored was the number ofvisits, not the quality of the meetings.

This lack of accountability to the farmers isobservable in both the household and the staff surveys.The household survey shows that the proportion offarmers who need advice—that is, the demand forinformation—on more complex messages is twice thatof the farmers who need information on simple agro-nomic practices. In contrast, the proportion that findssimple agronomic messages most applicable, whichreflects the effective supply of information, is twice thatfinding the more complex messages applicable. Simi-

44


larly, the mismatch between what the farmers want andwhat extension services supply is reflected inextension’s methodology. According to the FEWs them-selves, field/home visits constitute the least popularmethod among the farmers (only 5 percent preferthem), and demonstrations and field days are the mostpopular. Yet the majority of the FEWs prefer to use fieldvisits.

SustainabilityThe available estimates of total expenditure on exten-sion do not present a complete picture.2 How expendi-tures have evolved over time is not known.3 The mostreliable are the “printed” estimates, which suggest thatfor the year 1996–97, expenditures were approximatelyKP 156–177 million (or US$54–61 million)4 out of atotal ministry budget of KP 340 billion.5 That is, about46 percent of the ministry’s budget goes to extensionactivities. A review of public expenditures for agricul-ture also estimated that about 60 percent of theagricultural budget is devoted to extension activities, ofwhich 70 percent is donor-funded.

With Kenya’s total of about 3.44 million farmfamilies, according to the Welfare Monitoring Surveyand the Staffing Norms Study, these estimates suggestthat Kenya spent an average of about US$15.11 per farmfamily in 1996–97 for extension services, or Ksh876.38.6 Comparable estimates for 1982 are not avail-able, but estimates for the districts in the 1990 ATDstudy were US$3.92 for 1982 and US$4.67 for 1990 (inconstant 1991 dollars). These figures compare with thecurrent estimate of US$13.29 (in 1991 constant shillingsat the 1991 average exchange rate of Ksh 27.5 per US$1)or US$15.11 (in 1997 dollars, Ksh 58 per US$1).

While the optimal extension expenditure level isdebatable, the problem facing MALDM is that thecurrent system is too expensive and not financiallysustainable. Even toward the end of NEP II, projectfunds were financing 90 percent of the system’snonsalary operating costs. The government budget isinsufficient to keep the staff mobile and effective. Avast majority of both FEWs and SMSs confirm thatfunds for transportation and allowances are a “seriousor very serious” constraint on the effective delivery ofextension services. Similarly, training sessions andmonthly workshops have been reduced significantlybecause of a shortage of funds.

A major reason for the inadequacy of funds is thelarge number of extension staff. Estimates indicate that

at the end of 1996, more than 48 percent of all ministrystaff were engaged in extension. At the field level, theagricultural FEWs numbered approximately 6,841,more than double the 3,328 of 1982. Veterinary andlivestock production FEWs numbered 357 and 547,respectively, in 1996 (comparable estimates for 1982are not available). This yields an average of about 500farms per agricultural FEW. The number of technical(non-FEW) agricultural staff is estimated at about1,577—that is, there are about 4 FEWs for eachsupporting technical staff member.

The unsustainable growth in staff through the1980s, which currently takes up about 80 percent ofoperating costs, led to a recruitment freeze in 1990.While the increase in staff numbers was the result ofgovernment policy and presumably unrelated to NEPI (during which most of the increase occurred), it isnoteworthy that the ratios of farm families to FEWsand FEWs to technical staff are very close to theoriginal NEP I prescription of 500:1 and 3:1, respec-tively. It appears that the number of farm families wasunderestimated at the start of NEP I, when the farm-to-FEW ratio was much higher than the reported500:1.

These estimates of staff strength, however, areinconsistent with the data provided by the districts. Theratio of farms to FEWs is, on average, about 1,100:1.The discrepancy is significant and not easy to explain.It may stem from the poor identification and classifica-tion of staff by their current assigned duties (which donot necessarily correspond to the assigned job codes),and probably also reflects a concentration of staff atheadquarters, provincial, and district offices, wherestaff are involved in essentially nonfield activities. Thisis reflected in the staff survey: a majority of the SMSsindicated that they had responsibilities in subject areasother than those of their expertise. But regardless of theparticulars, it is apparent that the current system isoverstaffed and expensive.

PluralismPluralism was clearly not a characteristic of T&V asimplemented in Kenya. NEP I introduced the contactfarmer, and later the contact group, approach, with allthe extension staff time devoted to delivering advice tothe contact farmers. This was achieved by eliminatingall other extension activities, which may not have beeneffective in all cases, but were nevertheless efficientalternatives for delivering certain types of general

45

A n n e x e s

information. The NEP I and NEP II approaches,however, were not very successful in reachingnoncontact farmers (the briefing in Annex D, preparedby extension staff in one of the districts for a focusgroup meeting, gives some insight into the reasons, andthe effectiveness of the NEP I and NEP II approaches),but institutional energy has remained focused on theT&V approach.

A vast number of FEWs and supervisors acknowl-edge working on projects of other donors and workingwith NGOs and the private sector, mostly for theadditional incentives such as allowances, mobility, andtraining. Most such activities are also differentlyorganized, and staff generally consider them to bemore effective. Whether or not they truly are, or if thisbelief is a reflection of better working conditions, isunknown. However, there has been no effort to institu-tionalize the significant level of ad hoc activity to makethe system more effective or rational.

A negative impact of NEP I on an alternativeinstitutional arrangement was the discontinuation ofthe soil conservation program (Tiffen and others1996). The program had been well established andfunctioning reasonably well since 1974, but thecollective action required for soil conservation exten-sion was not amenable to the contact farmer or even asmall group, approach. It cannot be demonstrated onsmall plots, nor can it be reduced to simple messages.The program was therefore discontinued. It wasreintroduced in 1988 as a separate branch office,supervised and supported by the Swedish InternationalDevelopment Authority. The new program introducedthe catchment area approach.

A more recent unintended impact has been onlivestock extension. NEP I had concentrated on agricul-ture. Even though the Ministry of Livestock wasmerged with the Ministry of Agriculture briefly in theearly 1980s, the livestock extension department oper-ated separately until 1991, using its own approach,largely funded by other donors. Early attempts atincluding livestock extension in NEP did not succeed,because the livestock department resisted the T&Vapproach. The reamalgamation of the ministries andthe more recent unified approach to extension pro-moted by NEP II are creating significant tensions. Ingeneral, livestock extension is not amenable to deliver-ing messages at predetermined times, and it does notrequire frequent visits. Livestock advice is based onsolving specific problems and is not seasonal. It

requires a significant amount of training to convertspecialists into generalists and vice versa, which thealready trained and experienced livestock staff resent.

TrainingThe benefit of NEP I and NEP II that is most widelyagreed on is the upgrading of staff skills throughtraining.7 NEP I provided substantial training to olderstaff and newer, untrained staff. The regular trainingschedules were effective and had a positive impact onstaff quality. After the initial years, however, fundingconstraints, strained research-extension links, and thelack of new technology reduced the effectiveness oftraining sessions. The quality of extension staff is alsoattested to by farmers’ positive assessments of theircompetence (as noted in Chapters 3 and 4) and byassessments from NGOs and other donor projects thatoften use extension staff (although with additionaltraining).

Despite these significant efforts, however, a major-ity of the SMSs feel that FEWs are not qualified tocarry out their responsibilities. A majority also feel thatthere are too many subjects to handle effectively; alarge number of FEWs also voiced this sentiment.Finally, limited funds have restricted training sessions,although the primary reason the vast majority of SMSsgive for reduced frequency was that there was “nothingnew to say.” This reduced frequency of training,however, is reflected in the FEWs’ demand for moretraining sessions, since they see their effectivenessdeclining.

Research-Extension LinksA critical element in the high-intensity T&V approachis a regular flow of messages from research. In theearly years, the link between research and extensionwas weak, but sufficient to ensure a supply of simplemessages. Over time, as funding became tight, prob-lems started to emerge. Eventually, when KARI wasseparated from MALDM, the link was totally severed.In 1993, renewed efforts under NEP II led to amemorandum of understanding between KARI andMALDM to reestablish the link. The staff survey,however, reveals that the link is still very weak. Themajority of both FEWs and SMSs have noted thatmeetings with researchers are inadequate and infre-quent and that participation in field trials is limited.One constraint is the lack of adequate adaptiveresearch to generate new messages. Limited feedback

46


from farmers through extension has led severalregional research centers to conduct their own partici-patory rural appraisals to identify the farmers’ prob-lems and to target adaptive research. More recently,under the Farming Systems Approach to Research,Extension, and Training Initiative, with KARI’s leader-ship, the link appears to be improving, and some newtechnologies are emerging.

47

A n n e x e s

ANNEX D. FOCUS GROUP BRIEFING ON NEP’S IMPACT

A field extension staff member prepared the followingbriefing points for a focus group meeting that discussedthe impact of NEP I and NEP II in one of the districts.

NEP I• Farmer selection was not handled well. In most cases,

the farmers were handpicked by extension staff.• The farmers were unwilling to go to the same home

every time, so that a program of fortnightly visitswas not well received.

• In most cases, poor follow-up led to the failure offollower farmers to take up agents’ messages andreplicate them in their homes.

• Repetitive messages and the lack of clear technologypackages led to monotony in the information beingpassed to the farmers.

• Feedback mechanisms—from the farmers, throughextension, to research, and vice versa—were noteffective.

• Research programs rarely addressed the farmers’needs in the field, which led to poor linkages.

• Individual farmers who were left out felt that theextension service was aimed only at a few well-to-dofarmers. This is the reason for the administration’spermanent attack on extension—that its agents werenever seen.

• Division and district staff supervision was difficultbecause transportation was lacking and route mapswere unrealistic.

• Integration with other programs such as soil conser-vation and home economics was minimal.

• While funding was adequate for program activities,more than 90 percent of support went to staffactivities, and less than 10 percent went directly tothe farmers. This was the reason for low adoptionrates. Extension packages were judged impracticalfor simple farmers.

NEP II• Coverage was fairer than the contact farmer approach,

but most areas did not have worthwhile groups. Groupsmade up mostly of women had a “merry-go-round”agenda with very little agricultural activity.

• Frontline extension workers meet with four groups aday, primarily women’s groups. They meet once amonth, and therefore the FEW would either meet theowner of the home (a chairlady) alone or very fewmembers of the group.

• The lack of serious technical packages resulted,again, in repetitive and boring messages.

• Most of the demonstration sites were based at thechairladies’ homes, with little benefit to the members.

• No clear packages came from research except for thenormal agronomic messages that the farmers hadalready practiced for a long time.

• The farmers’ problems have still not been solved bythe existing extension approach. For example, (a)correct seed varieties for beans, maize, sunflowers,and the like; (b) crop pests and diseases are still beingresearched, and there is little assurance that successwill come soon; and (c) the lack of credit supply hasled to poor or low adoption rates.

• Mobility has continued to be the greatest problem atthe frontline, leading to very poor coverage at thefield level. The districts have good vehicles butinsufficient funds to maintain them. Field staff arepoorly remunerated.

• The administration of funds, particularly problemswith district treasuries’ liquidity, always led to thefailure of demonstration plots compared with those ofthe farmers.

• Funding for research programs depends on theinterest of the donor agencies, which generally do notaddress farmers’ needs. For example, only one clusterwas selected for an entire district of 14 divisions thathad very varied farmer needs.

• The top-down approach resulted in farmers’ expect-ing free things, and the projects’ approach as it wasimplemented seemed to be imposed on the farmers.Farmers’ views on how extension should be con-ducted should have been taken into account.

49

A n n e x e s

ANNEX E. AWARENESS AND ADOPTION OF EXTENSION MESSAGES

This annex briefly summarizes the data used for theawareness and adoption analysis; results are discussedin Chapter 5. Complete details on OED’s survey, themethodology of the analysis, and the results are givenin Working Paper 2.

Stock of MessagesThe first step of this strategy was to establish aninventory of extension messages and technologies thatare available from the research system. This effortyielded limited results. Few new technologies wererecommended during NEP I and NEP II, and those fewgenerally took the form of updated varieties, withoutmajor changes in practices. The recommendations formost practices have remained essentially the same forthe past 15 years. Obtaining specific extension mes-sages proved to be difficult, because district farmmanagement guidelines have not been updated. Thatthe technology stock and associated messages haveremained fairly constant is noteworthy. The evalua-tion, however, was able to establish that recommenda-tions for the 13 main activities for the crops grownmost commonly in the study districts do exist.

Survey DesignOED’s survey covered four crop categories: the maincrop, which was maize, cash crops, minor food crops,and new or promotional crops.1 The questionnaire,which covered messages on 13 cropping activities, wasdesigned to gain an appreciation of the sophisticationof the farmers’ knowledge.2 The coverage of thedifferent crop categories was motivated by the status ofthe main crop, which was maize for all study districts.Maize has been the target of extension activities sincethe mid-1960s, and therefore may not be very useful indetermining the impact of extension services at the timeof the survey. Cash and minor (or non-maize) foodcrops were covered to assemble a more completepicture, and the new and promotional crop categorywas included as a test of the effectiveness of theinformation dissemination system.

Some limited comparative results for 1990 areavailable from the ATD survey. It should be noted,however, that the ATD data have information on onlyone crop (mainly maize) for the vast majority of thesample of 420.3 Results from the ATD and OEDsurveys must thus be compared with caution.

51

A n n e x e s

ANNEX F. THE CONTINGENT VALUATION METHOD AND ITS APPLICATION

The contingent valuation method (CVM) is a tool foreliciting individuals’ use and nonuse values for avariety of public and private goods and services. Themethod relies on describing a hypothetical situation toa sample of individuals and asking them to state theirwillingness to pay to avoid a particular change in thatsituation, or their willingness to accept compensationfor a proposed change. The name of the techniquederives from the fact that what the individuals report iscontingent on the scenario that is described to them.

In its simplest form, CVM asks individuals directlyabout their willingness to pay to maintain the statusquo; that is, to avoid a change in the provision orquality of goods or services. The method has been usedwidely in industrial countries to estimate nonusevalues, typically for environmental and public goods.Its applications in developing countries are growing. Ithas been used to value, among other things, improvedsanitation services, household water services, surfacewater quality improvements, tsetse control, forestprotection, and wildlife viewing. This is the first knownapplication of CVM to elicit the willingness to pay foragricultural extension services. This is also the firstknown application for an impact evaluation of aproject.

In the OED survey, the farmers were first askedwhether they wanted to continue receiving extensionadvice, or would like to start receiving advice. Thosewho answered in the affirmative were then askedhow many extension visits they would like to receiveeach year. The following statement was then read tothe farmers:

The cost of providing extension advice (includingtransport costs, salaries, etc.) has been mostly financedby the government. The lack of funds is a majorobstacle in providing extension services. This couldlead to irregular visits by the extension workers, and adeterioration in the quality of the service. There is alsothe possibility that the extension program could beeliminated altogether.

This was followed by a question on whether thefarmers wanted the extension program to continue. Thefarmers who did want it to continue were reminded ofthe number of annual visits they wished to have andasked whether they would be willing to pay individu-ally for extension services. Those who said no werethen asked whether they would be willing pay as agroup member, if such a group were to be organized.The farmers were then asked how much they would bewilling to pay per visit.

The format of the CV questions was alteredbetween the first and second rounds of the survey inorder to test for various biases. In the first round, abouthalf of the farmers were asked double-bounded referen-dum, or closed-ended, questions, and the other halfwere asked open-ended questions. In the second round,farmers asked closed-ended questions in the first roundwere asked the open-ended questions, while the restwere asked a costless choice question—that is, whetherthey would prefer to receive extension services or amarket good (such as sugar or kerosene) worth approxi-mately 100 Ksh. The results reported in this evaluationare from the pooled responses to the open-endedquestions. The detailed description of the survey designand results are given in Working Paper 6.

53

A n n e x e s

ANNEX G. COMMENTS FROM THE GOVERNMENT OF KENYA

Gregory K. IngramManagerSector and ThematicEvaluations GroupOperations Evaluation DepartmentThe World BankWashington, DC 20433USA

Fax 202-522 3123

Dear

RE: COMMENTS ON IMPACT EVALUATION OF AGRICULTURAL EXTENSION PROJECTS

Refer to your letter dated 10th June 1999 on the above subject.

Please enclosed herewith final our comments on the document for your consideration and incorporationto the final document.

Yours

AMB. JOSHUA K. TERERPERMANENT SECRETARY

Encl.

THE PERMANENT SECRETARYKILMO HOUSE

CATHEDRAL ROADP.O. BOX 30028

NAIROBI

28 June, 1999

Telephone 718870, Fax 720568When replying please quoteRef. No. OFTA/9

and date

54


Ministry of Agriculture’s comments on the Impact Evaluation of Agricultural Extension Projects in Kenya

[Numbers refer to original paragraph numbers in draftsubmitted.]

1.0 Introduction1.1 Generally the document has captured most of the

areas in evaluating the Agricultural ExtensionProjects in Kenya. However, we have the follow-ing comments to make.

2.0 Specific Comments2.1 In 1994, the animal health extension was incor-

porated in NEP II activities after the Mid-TermReview of August, 1994, however, the contribu-tion of T&V towards livestock activities has notbeen captured in the document.

2.2 After four joint IDA and GOK monitoring andevaluation Missions. It was concluded that man-agement of NEP II was the main constrainttowards unification and co-ordination of exten-sion services. Firstly, Management of NEP IIwas through a Project-Working Group and thismay have had a far reaching effect on the projectimplementation. This was a very importantconclusion that the evaluators of efficacy ofT&V should have considered. Secondly, theimplementation process followed the procedureslaid down by the Ministry of Agriculture. Pro-curement was an important management tool inNEP I and NEP II. Motor vehicles, motor cycles,bicycles and office equipment and furniture wereprocured. The extent to, which it was realizedshould evaluated.

2.3 Extension Services provision in NEP II were mainlyconfined to monthly workshops, monthly trainingsessions, staff seminars and courses, staff/farmereducational tours, on-farm demonstrations farmersfield days, agricultural shows, farmers courses andseminars, group visits and meetings. These wereextension methods used by extension agents tocreate awareness and encourage adoption of tech-nologies by farmers. However, one page 6 of thedocument in table 1, it has been observed that T&Vhad its strengths and weaknesses that the evalua-tors should have considered to be able to arrive at abalanced evaluation report about the efficacy ofT&V Management of Extension Services.

2.4 In the document enumerators have used a struc-tured questionnaire. Unfortunately no sample of

questionnaire is appended hence it is difficult toestablish the quality of data. There is also noindication of use of informal survey methods toverify the results from the formal survey.

2.5 Comparison has been made using data andinformation of 10-15 years ago. There is noindication of any survey done before the start ofproject implementation, hence the rationalementioned here does not hold. It is also stated inthe document that there was no baseline studydone to ascertain pre-project situation.

2.6 The document has concentrated basically oninformation access/dissemination within thehousehold and no other stakeholders are in-volved. In addition, the indication that only 47%of respondent say that information is less avail-able now than it was before relates to askingdirect questions or through a questionnairewhich may not be conclusive. Other forms ofverification should have been explored.

2.7 The document also indicates that key findings ofthe survey are that there has not been apparentimprovement in the quality of delivering of exten-sion services. There are no conclusive indication ofthe same. The kind of data the evaluators haveobtained from the questionnaire can not be conclu-sive. Other verifiable indicators are required.

2.8 Extension Services covers a wide range of agro-ecological zones. In all places it is not possible toapply contact farmer and follow-up farmersprinciples of T&V. So the extension agent has toapply other methods to deliver information to thefarmers. In ASAL areas they can use extensionbarazas but it should be noted that these are notadministrative barazas. Also in administrativebarazas, extension agents are invited to addressthe public but this should not be construed tomean it is an extension barazas.

2.9 One year of suspension is too short to notice anysignificant change in delivery of extension ser-vices as reflected in the document. There wereother parallel extension related projects, whichmay have supported extension services in 1996,when NEP II funds were suspended. The Govern-ment may have redirected the recurrent fundsfrom other sources towards extension servicesduring that period of suspension.

55

2.10 Regarding the outreach, first contact with farm-ers has been used to analyze the effectiveness ofan Extension Services. This measure does not saymuch about the efficacy of T&V. It leaves a lotof room for criticism because the baseline dataon pre- T&V situation were not available. Thisis may therefore be treated as a proxy outputindicator.

2.11 The document has largely covered crops at theexpense of livestock which is often a majorenterprise in some farms, therefore it may notrepresent a holistic picture of the ExtensionServices.

2.12 In T&V extension approach, it is assumed thatfarmers learn about innovation/technology bybecoming aware of it, become interested in it andadopt it. Learning occurs by adopting ideas thatcome from outside. The evaluation has notconsidered the possibility that farmers are activeproblem solvers on their own and that they arenot passive consumers of technology, but part ofits development born out of long experiences infarming. The term “message” here could acquirea more pluralistic meaning if the term “informa-tion” was used instead of message.

2.13 The use of statistical analysis to be able toexplain process such as awareness and adoptionis not plausible. It is because there are other

confounding variable like attitudes and pastexperiences of farmers that will very muchinfluence farmers probability of becoming awareof technology or information and adopting it.This section is an academic exercise and doesnot explain actual reality.

2.14 The Ministry of Agriculture through the DivisionExtension Service also recognized the constraintsof non-adoption due to irrelevant technology. Alinkage was established between extension andKARI with the mandate to address the problem ofinappropriate technology. Indeed some NEP Ifunds were allocated to KARI Regional ResearchCenters (RRC) for linkage activities. The resultsmay not have been captured during the survey.

4.15 The evaluators conducted T&V systems as statedin theoretical books and did not consider themodification that this approach had undergone,the complex farming system and socioeconomiccircumstances of smallholders. Therefore theevaluation is academic.

4.16 Despite the increased farmer/extension contactand significant increase in level of technologyawareness as observed in the documents, theevaluators have not considered that adoptionrate for these new technologies was also lim-ited to a greater extent by low returns fromfarm produce.

A n n e x e s

57

A n n e x e s

ANNEX H. OED’S RESPONSE TO BORROWER COMMENTS

This report summarizes the main findings of the OEDevaluation, detailed descriptions of the analysis areincluded in the six supporting working papers. Theevaluation is based on evidence obtained through abroad-ranging household survey and on the informationand data available from two earlier surveys, in 1990 and1992. The evaluation also surveyed frontline extensionstaff, subject matter specialists, and supervisors in thestudy districts. As noted in the report, these data werecombined with secondary sources of information ob-tained from various reports and documents, includingseveral reports of the Ministry of Agriculture (MOA).The historical evidence was collated from studies reach-ing back to the early 1970s (see references in the reportand the Working Papers). In addition, the evaluationused the government of Kenya’s participatory povertyassessments (as noted in Chapter 3) and an independentbeneficiary assessment conducted by an NGO in fullcooperation with the MOA. Responses to specific com-ments from the government of Kenya follow. (Allparagraph references refer to the original draft.)

Para. 2.1. The household and staff surveys and thebeneficiary assessments covered all extension activities,including livestock production, animal health, soilconservation, and the like. Thus, the analysis in Chap-ters 2–4 is not restricted to crop production. In otherchapters, dealing more specifically with physical ortechnical impact, the analysis is restricted to cropactivities. Accordingly, the text in Chapters 5–7 has beenmodified and a footnote added to clarify this. Theprimary reason for focusing on crop-related activitieswas that livestock production was included in NEP IIonly after 1994. For most of the period under study,livestock extension activities were financed and man-aged under a different system. Also, a preliminaryanalysis of the awareness and adoption of livestockproduction practices showed that almost all farmers hadheard of and adopted those practices by 1994. A deeperanalysis of livestock production activities will be under-taken and communicated separately, although the asso-ciation with NEP II is likely to be tenuous. Further, asdiscussed in Annex C, the limited experience of thelivestock extension staff with T&V has not been positive.

Para. 2.2. OED considers procurement to be afunction of project management, not a managementtool. The text in Chapter 2 and Annex C has beensuitably modified.

Para. 2.3. OED assesses efficacy by comparing aproject’s outcomes with its goals. The findings inChapters 3 and 4 show that project design was notefficacious, for reasons outlined in table 2.1 and notedin the conclusions section of Chapter 2.

Para. 2.4. The questionnaire was deliberatelyomitted from the final report for brevity, but isavailable on request. A copy was sent to the Ministry ofAgriculture at the time the survey was conducted.

Para. 2.5. The comparison with the situation 10 to15 years earlier was a subjective assessment by the usersof the extension services, the farmers, of the change intheir access to extension services and their quality. OEDtook these beneficiaries’ perceptions at face value, buttheir views were not the only basis for the analysis. In theabsence of appropriate baseline data on various aspectsof the extension services, OED used recall data wherefeasible. Formal statistical tests in Working Paper 1compare recall estimates with estimates from threeindependent surveys for different time periods between1982 and 1993 on contacts made with the extensionsystem. The tests demonstrate that recall bias is unlikelyto be significant. Further comparisons in Working Paper2 on the awareness and adoption of maize practices alsosuggest the same. More important, comparable findingsfrom a number of studies noted in Working Paper 1,some from as early as 1972, provide a significant degreeof confidence in the broad results emerging from theOED survey. As for the technical impact on farmers’efficiency and productivity, the 1982 RHBS provided anappropriate baseline for NEP I, the 1990 ATD survey forNEP II.

Para. 2.6. As noted above, the evaluation includeda survey of extension staff. Most staff reported adecline in the system’s effectiveness after 1982 and1990, as noted in Chapter 2. Table 3.1 summarizes analternative assessment by the beneficiaries that queried19 services in a contextually independent manner.There is no compelling reason to believe that theresponses for extension services are biased. In thiscontext, it should be noted that the evaluation strategyincluded beneficiary assessments. It is unfortunate thatthe beneficiary assessments were not carried out. It wasagreed that beneficiary assessments would be con-ducted by the MOA with the full support and coopera-tion of OED (including financial contribution) and theBank’s Kenya Country Department. However, after

58


initial preparations, the ministry unilaterally decidednot to implement the beneficiary assessment.

Para. 2.7. The qualitative aspect of the delivery ofservices refers to the projects’ design, and the text hasbeen modified to clarify this point. The projects’ designanticipated high-intensity and qualitatively differentmeetings between extension agents and the contactgroups. As discussed in Chapters 3 and 4, the outcomefor this aspect is unsatisfactory.

Para. 2.8. OED agrees with the assessment, andthe report is consistent in noting that a uniform methodshould not be used in all locations, as advocated in theT&V design. The report has been modified to reflectthe correct interpretation of baraza.

Para. 2.9. The impact of the suspension of funds isdiscussed in further detail in Working Paper 1. Consider-ing the prescribed schedule of meetings between exten-sion agents and contact farmers and contact groupsenvisaged by T&V, the suspension would be noticeable ifthe system were working as anticipated. While it ispossible that other funds could have been used, this didnot happen. None of the officials met during theevaluation indicated that alternative funds were madeavailable. In any event, if other funds had been used, itwould have indicated that the Bank’s funds were notneeded, raising the issue of the additionality of NEP IIfunds.

Para. 2.10. OED agrees with the comment, Work-ing Paper 1 makes the same point, and the report hasbeen modified to further clarify this point. The analy-sis, however, does go beyond first contact and coversmore meaningful aspects of meetings between exten-sion agents and farmers.

Para. 2.11. The response is the same as that forparagraph 2.1.

Para. 2.12 OED fully agrees with the premise ofthe comment, which is the basis for judging theincremental impact attributable to NEP I and NEP II. Itshould be reiterated that this is an evaluation of theimpact of NEP I and NEP II; both relied on the time-bound delivery of messages.

Para. 2.13. The analysis of awareness and adoption inWorking Paper 2 includes a number of social andagroecological factors in addition to extension. The statisti-cal results are consistent with the descriptive findings andthe focus of NEP I and NEP II activities, as noted by severalsenior extension staff and other observers.

Para. 2.14. The report takes note in Chapter 2 thatstaff training and improved links between research andextension are benefits of the projects. As inputs into theextension system, the results of these activities wouldnaturally affect the flow and content of the extensionactivities in the field. These would thus be fullyreflected in the impact at the farm level in greaterawareness and adoption of practices and improvedproductivity.

Para. 2.15. Farmer characteristics are included inall aspects of the evaluation. The analysis is, for themost part, general and encompasses alternative meth-ods, as long as they were carried out by governmentextension agents. However, as an evaluation of NEP Iand NEP II, the focus had to be on the projects’objectives and design as contained in their StaffAppraisal Reports and as presented in the variousMOA documents noted earlier. Therefore, the evalua-tion paid particular attention to the specific aspects ofthe extension methodology, but the farmers’ quantita-tive and qualitative assessments were generally inde-pendent of any particular method.

Para. 2.16. The results show increases in aware-ness and adoption for only a few simple agronomicmessages. As noted in the relevant sections of Chapter5, these are also known to have been relatively higheven at the start of NEP I, and especially during NEP II.The levels for the more complex messages are still verylow. It is true that low returns, along with otherconstraints such as the availability of credit or inputs,are important factors. However, a negligible propor-tion of farmers indicated low returns as the reason fornot adopting agents’ recommendations (see WorkingPaper 2). A number of farmers did note financialconstraints, and some noted input market problems.Also, almost 80 percent of those who were aware of therecommendations (including the more complex ones)adopted them. This clearly shows that while otherconcerns may be valid, lack of information is a seriousconstraint. Finally, according to the 1998 EconomicSurvey published by the Government of Kenya, theagricultural output price index for 1997 was 598.9(1982=100), the total input price index was 520, andthe fertilizer price index was 314.3. Thus, it is unclearwhether adverse market conditions were behind thefailure of farmers to adopt the recommended practices.

59

ANNEX I. HOUSEHOLD SURVEY DATA

OED conducted a comprehensive household surveyin 1997 to collect data on a wide range of topics.1 Thesurvey was designed to generate panel data by revisit-ing as many households as could be relocated from a1990 household survey conducted by the Africa Techni-cal Department (ATD), which drew from a subsampleof the 1982 Rural Household Budget Survey (RHBS).2

As may be expected, the number of households com-mon to all three surveys is considerably smaller thanthe total observations in each year. The interveningperiods are relatively long and it is not surprising that anumber of households could not be traced (because ofdeaths, migration, and the like).

The 1982 survey was a stratified random samplerepresentative of about 95 percent of the nationalpopulation, and drawn from the existing nationalsampling frame (NASSEP I).3 The sample represented2.8 percent of the rural population, drawn from 640clusters from all but the North-Eastern province(covering about 54 percent of the land area). The 1990survey randomly picked clusters in seven of thedistricts, representing a broad cross-section of Kenyanagriculture, but excluded districts in which NEP I hadnot been implemented. From the list of householdssurveyed in 1982 in these districts, about 700 house-holds were randomly picked for the ATD survey (allhouseholds with household identifiers ending in 7).The 1997 survey targeted the same subsample as the1990 survey, using the original list of respondentsfrom the 1982 survey.

For this analysis, only the clusters included in thesubsequent surveys are retained from the RHBS data.This yields about 611 observations for which agricul-tural input and output information is available. The1990 survey targeted about 700 households, butcomplete data for production analysis are availablefor about 450 households. The 1997 survey targetedhouseholds belonging to the same clusters retained forthe 1990 ATD survey, but clusters with fewer than fourtarget households were dropped. Another 11 clusterscould not be traced, following the many changes inadministrative boundaries that have occurred sincethe original sample selection. Of these, two werediscarded altogether; for the other nine, replacementhouseholds were randomly selected in the samelocation as the original clusters. Overall, 293 of theoriginal respondent households could be contacted.

Another 62 of the households belonged to descendantsof the original respondents, and these were retained inthe sample. In addition, 241 new households wererandomly picked from the same clusters as the missinghouseholds, bringing the total sample size for 1997 to596 households.4 The purpose of selecting replace-ment households was to develop a sample of reason-able size, but at the same time to maintain thelocational and socioeconomic characteristics of thesample. In all, the OED survey collected data from 73clusters spread over 12 (current) districts.

The OED survey was structured to begin withcontextually general questions—that is, questions thatwere not specific to any particular agency or organiza-tion. The enumerators were specifically instructed torecord unprompted responses whenever feasible.5

Farmers were prompted only when it was necessary totest for specific issues or to pursue certain preconceivednotions based on past experiences within and outsideKenya. Special care was taken to phrase and sequencethe questions to ensure that there were no leadingquestions. Accordingly, it was important to maintainthe identity of the surveyors as far removed from thegovernment or the extension service.

To dissociate the survey from the extension ser-vice, the government, or the World Bank, the OEDsurvey was conducted by the Tegemeo Institute ofEgerton University. The enumerators were collegegraduates, selected for their familiarity with agricul-tural issues and fluency in the local dialects of thesurvey sites. The enumerators were trained in thesurvey methodology, particularly for questions relat-ing to the extension and contingent valuation mod-ules. The survey instrument was pre-tested to fine-tunethe nuances of the questions and language. The surveywas implemented in two rounds, one at the start of the1997 long rains season (August/September 1997) andthe second after harvest (November/December 1997).In the second round, 34 of the respondents could notbe contacted again.

The type of data available from the three surveysvaries. But although the surveys are not fully compa-rable, they are sufficiently so to allow some meaningfulanalysis. The 1997 data is the most detailed, by design.In the planning stages, the objective of the OED surveywas to be able to allow a comparison with the 1990data. However, the questionnaire was expanded to

A n n e x e s

60


collect additional information on various aspects of theproduction process. The details of the differencesamong the surveys are discussed in the Working Papersas needed for analysis or comparisons. The OEDsurvey questionnaire is available from the author onrequest.

6 1

ENDNOTES

Chapter 11. The ATD Kenyan study is one of the three notable

attempts to rigorously estimate the returns to T&V investments,as described in Annex A. But as discussed in Chapter 7, theestimates of high returns are not robust.

2. A public expenditure review for agriculture revealed thatin 1996–97, extension claimed 61 percent of the developmentresources allocated for MALDM’s core services. Extension alsoaccounts for about 45 percent of MALDM’s total expenditureand more than half of its staff.

3. Aggregate crop statistics are notoriously poor in Kenya.Maize data are likely to be the most accurate because of theimportance of maize in Kenyan agriculture and the attention itreceives from all quarters. Nevertheless, the accuracy of eventhese data cannot be affirmed.

4. Average growth between 1970 and 1996 is estimated at2.7 percent, reflecting a deterioration of yields in the 1990s. Therainfall-controlled growth rate cannot be calculated from 1990onward because rainfall data are lacking. The rainfall data usedhere are average annual millimeters of rain from 14 stations fromthe south, southeast, central, and western parts of Kenya.

5. Most opinions are based on anecdotal evidence from fieldvisits. See box A.1 in Annex A on the difficulty of using suchevidence to draw inferences.

6. These weaknesses, of course, were not specific to Kenya(Feder, Slade, and Lau 1985).

7. As the perception of the roles and functions of extensionhas evolved, so have the models used to capture the interactionsof research, alternative extension providers and methodologies,and farmers. The simplistic model here is meant to reflect theinteractions assumed for NEP I and NEP II.

Chapter 21. The key reference reports are MALDM 1997b, c, and

Kandie 1997.

2. It was found later that a miscalculation of the number offarm families had led to a farm-to-staff ratio much higher thanthe 500:1 that the staff appraisal report judged to be adequate.

3. Outreach is defined here as any type of contact betweenfarmers and the extension system. As discussed in Chapter 4,however, this can be a poor measure of effectiveness. It is alsoinconsistent with the role of “contact farmer” or “contact group”that T&V advocates. Nevertheless, for reasons discussed below,field staff are using alternative methods, particularly barazas, toincrease their outreach.

4. About 26 percent of those who were in service in 1990thought that the system was more effective at the time of thesurvey; 49 percent thought that it was less effective, and 25percent thought that it had remained the same.

5. The general principles are professionalism, a single line ofcommand, concentration of effort, time-bound work, client orien-tation, and regular training.

6. Most institutional analyses focus on the characteristics of

goods or services from the suppliers’ perspective. Thus, thesubtractability and rivalry of benefits from goods and servicesprovide guidelines for the optimal delivery mechanism. But evenfor a public good, viewing benefits from the demand side helpsconceptualize efficiency gains from cost recovery, even if costrecovery is only partial.

Chapter 31. In addition to complementing the quantitative, survey-

based findings of this study, the beneficiary assessment wasintended was to allow a comparison of the quantitative resultswith those from a more qualitative and participatory approach.

2. Even though these were “poverty” assessments, a substan-tial number of people belonged to land classes that are compa-rable to the OED survey households.

3. On average, once a year in low-potential zones, twice ayear in medium-potential zones, and often in higher-potentialzones.

4. The ranking used a simple tabulation of the reported firstchoices. Considering that basic preferences are likely to be givenhigher priority, alternative rankings were tried that used the topthree and five choices and the assigned ranks as weights foraggregating across observations. These results were consistentwith extension retaining the fifth or sixth ranking.

5. The only exception was electricity. In declining order, therankings were piped water, public health dispensaries, dry seasonroad access, electricity, and tarmac roads.

Chapter 41. It is possible that some noncontact farmers do not know

that a neighbor is a designated contact farmer, and consequentlyreport their source of information as “friends and neighbors.”However, it is unlikely that in a small community, especiallywhere group activities are reportedly common, that farmerswould not observe the regular and frequent visits of an extensionagent or other farmers to one particular farm. In either case, thelack of publicity about extension activities in given locations islikely to have reduced their potential for impact.

2. These results are consistent with recent findings frombeneficiary assessments in several African countries, which showthat contact farmers are likely to be less well connected with therest of the community than hoped (Salmen 1999).

3. This analysis is based on farmers’ recall of their firstmeeting with extension agents. Comparisons with three indepen-dent data sets for three different time periods since 1992 showthat the recall bias is unlikely to be significant.

4. As noted earlier, however, the bias against women in theselection of contact farmers appears to have persisted.

5. That is, among the farmers who reported receiving adviceat least once a year, or about 41 percent of the sample.

6. In part, the problem is that “useful,” a term used often todetermine the effectiveness of extension services, is ill-defined andvague. Farmers are also reluctant to volunteer criticism.

62


7. Supply is defined as the effective supply of information asrevealed by the farmers’ responses about the recommendationsthat they find most applicable. It is assumed that these recommen-dations are either those they receive advice on or those that arerelevant to their circumstances.

8. Of the contact farmers who reported noticing a change,about 23 percent actually reported an increase in extension visits,37 percent reported fewer visits, and 32 percent reported no visitsat all. The remainder gave unspecified or other responses. Surpris-ingly, the pattern of the contact farmers’ responses is almost thesame as that of the noncontact farmers.

Chapter 51. Assessing the actual impact of extension requires measur-

ing the associated increase in farmers’ agricultural productivity.This is dealt with in Chapter 6.

2. Increased productivity can take the form of diversificationinto higher-valued crops, an increase in the efficiency of input use,a change in the use of productive inputs, or a combination of thethree.

3. A farmer’s decision to adopt a particular technology isinfluenced by a number of economic and technological factors,including extension advice (Feder, Just, and Zilberman 1985).

4. This process is akin to the spread of an epidemic or aninfectious disease (Feder and Umali 1993).

5. Livestock extension services were included in NEP II onlyafter 1994. A separate, preliminary analysis shows that the vastmajority of the households that were aware of livestock-relatedrecommendations were already aware of them by 1994. Attributingthe awareness or adoption of these recommendations to NEP II isthus likely to be tenuous. The analysis of the awareness and adoptionof livestock recommendations will be conducted separately.

6. Less than a third of the sample was aware of therecommendations on fertilizer and other chemical inputs.

7. Several early surveys showed that activities in many of thesimpler and maize-related recommendations were already beingperformed widely at the start of NEP I (see Gerhart 1975 andOngaro 1990).

8. Since maize has been the focus of government extensionsince the mid-1960s, the cumulative impact of its efforts onrelative levels of awareness for maize practices is to be expected.

9. The results could also be interpreted as reflecting theefficient working of the spread effect of the contact farmerapproach. But if this were the case, there should be a corre-sponding increase in the share of friends, neighbors, and familyas a source of information. The results show a contrary trend:the share of friends, neighbors, and family has steadily declinedfor all activities.

10. It is reasonable to assume that, with more than 80percent of those who are aware adopting recommendations, if theremaining two-thirds who are currently unaware of the recom-mended practices were provided with appropriate advice, almost50 percent of the sample could be potential adopters.

11. These include the supply of extension, measured as theratio of extension worker to farm families in each location. The

cumulative effect of extension advice over years is captured byusing a weighted, lagged structure of extension supply over anumber of years.

12. For spacing, seed rates, and chemical plant protectionmeasures, the predicted contact farmer variable is significant atthe 5 percent level; for crop variety and cultural pest control, it issignificant at 10 percent.

13. This new specification, a multiplicative parametric speci-fication of the supply of extension over discrete time intervalsbetween 1982 and 1997, was also tried for the awarenessestimation. The results for awareness with respect to extensionwere similar to the adoption results.

Chapter 61. Technical efficiency measures physical productivity—that

is, it relates physical inputs to output. Scale efficiency measuresthe deviation of each farm from the optimal size of operation.Allocative efficiency measures the deviation from the optimal ofthe input mix given the current market conditions as reflected bythe current prices. For 1982, input price data are not available,and hence only the technical and scale measures are calculated.

2. Weak significance refers to significance at the 10 percent level.

3. Since only 285 observations were used, some tests were doneto check for potential selection bias for the retained observations (aresult of inability to contact all the households in the 1997 surveylocations). A simple t-test of the 1982 level of relative efficiencyshows that the average efficiency of retained observations is slightlylower than that of the rest of the 1982 sample, and significant at the5 percent level. But when agroecological factors are controlled for, ina Probit regression, the 1982 difference in efficiency level is no longersignificant at the 5 percent level.

4. A drawback of DEA is its sensitivity to measurementerrors. To minimize these, the analysis used only observationswith positive outputs, and for all variables, observations in thetop and bottom 1 percent of distribution of intensities (output orinput per unit of area) were eliminated; the exception to this wasobservations with zero nonlabor cash inputs.

5. Malmquist indexes are calculated so that scores below 1represent a positive change or gain, while scores above 1 representregress. To make the presentation more transparent, the graphsdepict the inverse—that is, scores greater than 1 represent gains inproductivity or efficiency.

6. These trends are consistent with farmer complaints in thehigh-potential districts, such as Trans Nzoia, that they areobtaining lower yields with the same or more inputs.

Chapter 71. The working paper has been recently published as World

Bank Policy Research Working Paper 2098.

2. Because of attrition and incomplete data for some vari-ables, the number of observations common to the 1982 and 1990data are 306; to the 1990 and 1997 data, 216; to the 1982 and1990 data, 258.

3. Alternative lag structures and lengths were tried, but thequalitative results did not change significantly.

63

E n d n o t e s

4. To the extent that there are any carryover effects from theprevious system, the impact of NEP I’s early years is likely to beoverestimated, but this is not considered to be a major limitation.

5. For each observation, the average was taken over all otherobservations in the cluster to avoid spurious statistical associa-tion, since household production level enters the dependentvariable calculation.

6. Note that district dummy variables are included, inconjunction with distances to market and roads, as proxy for localprices, which are not available for the 1982 and 1990 data.

Chapter 81. The survey followed the professionally accepted guide-

lines for CVM questioning: interviews were conducted in person;the more conservative willingness to pay (rather than willingnessto accept) was elicited; sample and item nonresponse were quitelow; open-ended as well as double-bounded referendum questionswere asked; the program was well understood by the farmers; andthe farmers were given opportunities to explain their responses(see Arrow and others 1993).

2. Eleven respondents (less than 2 percent of the sample)mentioned “on-demand” in response to the desired frequency ofvisits. These observations have been dropped for the rest of theanalysis.

3. Farmers were also asked the mode of payment theypreferred, and the majority chose cash.

Annex A1. A larger number of studies have attempted to measure the

effectiveness of extension in general, including several studies ofT&V systems. However, most of the studies suffer from concep-tual or methodological limitations, as Birkhaeuser, Feder, andEvenson (1991) and Feder and Umali (1993) have noted.

2. Kenya was the first country in Africa to reform itsnational extension service along T&V lines.

3. The Central Bureau of Statistics, an independent agency,collected the data, not MALDM.

Annex B1. The pilot focused only on maize and was conducted with

weekly visits rather than the fortnightly visits prescribed by T&Vstandards.

2. Non-FEW staff included the DAO, SMS/AO, DEO, TO,senior account clerk, clerks and enumerators, and drivers.

Annex C1. Such observations were made during individual and group

meetings of current and past extension and nonextension staff ofthe ministry, academics, researchers, and donors.

2. Expenditure estimates for wholly extension projects canbe identified, but expenditures on extension as a component of

other projects (a number of which are donor-funded) are not yetaccounted for.

3. The data MALDM provided to OED for NEP I areinaccurate. A cross-check with the appropriation accounts for1994–95 revealed that those data were for expenditures (develop-ment and recurrent) for the whole ministry, and not just forextension.

4. The conversion rate is US$1 to KP 2.9 (Ksh 58) for 1997.

5. The exact total depends on how much of the KP 21.7million that was spent on information management was attribut-able to extension activities.

6. An alternative estimate can be derived from the StaffingNorms Study. Assuming that the short-term staffing norms pro-posed for 1998–99 reflected the current staffing levels, personnelcosts for the study districts for district-level staff alone wereabout Ksh 374 per farm family. Adjusting for 15 percent inflation,assuming that these costs represented 80 percent of the recurrentbudget, and doubling it to account for the development budget,yields an estimate of Ksh 794 per farm family for 1998–99. This isclose to the estimate from the 1996–97 budget, considering that itis an underestimate because it omits staff expenses for all staffabove the district level.

7. Part of the improvement in staff quality can also beattributed to the rapid increase in staff numbers through the1980s, which brought in better trained graduates from theagricultural colleges.

Annex E1. The questions on awareness and adoption of practices

were administered in the second round of the OED survey, when34 households from the original sample of 596 could not becontacted again. The following analysis is based on responsesfrom the remaining 562 households.

2. Targeted recommendations ranged from simple messages(on crop varieties, planting time, spacing, seed rate, weedingtime, number of weedings, and the like), to practices of intermedi-ate complexity (such as applying types of basal and top-dressfertilizers), to more complex messages (such as knowing thequantity of basal and top-dress fertilizers, time of top-dressing,and chemical and cultural pest and disease control measures).

3. The results reported in Bindlish and Evenson (1993) aremostly for maize, with beans as the only second crop for about 25percent of the 1990 sample. Coverage of crops other than maizeintercrops was negligible.

Annex I1. The survey included modules on the nature and extent of

interaction with agricultural extension services and other extensionactivities; detailed input and output data for crop production;animal health and livestock production data; household demograph-ics; farm equipment and agroecological characteristics; awarenessand adoption of extension messages; a contingent valuation module

64


to elicit willingness to pay for extension services; infrastructure andother services; credit; and group activities.

2. These three surveys generate a panel data set for approxi-mately 300 households. The surveys cover household demograph-ics, farm characteristics, and input-output data on agriculturalproduction. The 1990 and 1997 surveys also collect informationon contact with extension services, including awareness andadoption of extension messages.

3. NASSEP stands for the National Sample Survey andEvaluation Program.

4. The selection criteria for the replacement households wasto select the fourth household to the left (facing out at front gate)of the missing household’s farm.

5. With subsequent post-coding of the responses.

BIBLIOGRAPHY

6 5

Working PapersAll Working Papers are available on request.

1. Gautam, Madhur. 1999. “The Efficacy of the T&VSystem of Agricultural Extension in Kenya: Resultsfrom a Household Survey.” OED, World Bank,Washington, D.C. Photocopy.

2. Gautam, Madhur. 1999. “Awareness and Adoptionof Extension Messages.” OED, World Bank, Wash-ington, D.C. Photocopy.

3. Gautam, Madhur, and Jock Anderson. 1999. “Re-considering the Evidence on Returns to T&V Exten-sion in Kenya.” OED, World Bank, Washington,D.C. Photocopy. (Reprinted as World Bank PolicyResearch Working Paper 2098.)

4. Gautam, Madhur, and Jonathan Alevy. 1999.“Farmers’ Efficiency and Productivity Change inKenya: An Application of the Data EnvelopmentAnalysis.” OED, World Bank, Washington, D.C.Photocopy.

5. Gautam, Madhur. 1999. “The Impact of Extensionon Production in Kenya.” OED, World Bank, Wash-ington, D.C. Photocopy.

6. Gautam, Madhur, and Amy Gautam. 1999. “TheWillingness to Pay for Agricultural ExtensionServices in Kenya: An Application of the Contin-gent Valuation Method.” OED, Washington, D.C.Photocopy.

ReferencesActionaid Kenya. 1997. Kwa Maoni Yetu (From Our

Perspective). Washington, D.C.: World Bank, Opera-tions Evaluation Department.

Arrow, K., R. Solow, P. R. Portney, E. E. Leamer, R.Radner, and H. Shuman. 1993. “Report of theNOAA Panel on Contingent Valuation.” FederalRegister 58: 4601–14.

Benor, D., and M. Baxter. 1984. Training and VisitExtension. Washington, D.C.: World Bank.

Benor, D., J. Q. Harrison, and M. Baxter. 1984.Agricultural Extension: The Training and Visit Sys-tem. Washington, D.C.: World Bank.

Birkhaeuser, D., R. E. Evenson, and G. Feder. 1991.“The Economic Impact of Agricultural Extension: AReview.” Economic Development and CulturalChange: 607–50.

Bindlish, Vishva, and R. E. Evenson. 1997. “TheImpact of T&V Extension in Africa: The Experience

of Kenya and Burkina Faso.” The World BankResearch Observer 12(2): 183–202.

———. 1993. Evaluation of the Performance of T&VExtension in Kenya. Technical Paper 208, WorldBank Africa Technical Department. Washington,D.C.

Bindlish, Vishva, R. E. Evenson, and MathurinGbetibouo. 1993. Evaluation of T&V Extension inBurkina Faso. Technical Paper 226, World BankAfrica Technical Department. Washington, D.C.

Cleaver, Kevin. 1993. A Strategy to Develop Agricul-ture in Sub-Saharan Africa and a Focus for the WorldBank. Technical Paper 203, World Bank AfricaTechnical Department. Washington, D.C.

Feder, Gershon, and R. H. Slade. 1986. “The Impact ofAgricultural Extension: The Training and Visit Sys-tem in India.” World Bank Research Observer 1(2):139–61.

Feder, Gershon, and Dina Umali. 1993. “The Adoptionof Agricultural Innovations: A Review.” Technologi-cal Forecasting and Social Change 43: 215–39.

Feder, Gershon, R. Just, and D. Zilberman. 1985.“Adoption of Agricultural Innovations in DevelopingCountries: A Survey.” Economic Development andCultural Change 33: 255–98.

Feder, Gershon, L. J. Lau, and R. H. Slade. 1987.“Does Agricultural Extension Pay? The Training andVisit System in Northwest India.” American Journalof Agricultural Economics 69(3): 677–86.

Feder, Gershon, R. H. Slade, and L. J. Lau. 1985. TheImpact of Agricultural Extension: The Training andVisit System in Haryana. World Bank Staff WorkingPaper 756. Washington, D.C.

Feder, Gershon, A. Willett, and W. Zijp. 1999. “Agri-cultural Extension: Generic Challenges and the Ingre-dients for Solutions.” World Bank Policy ResearchWorking Paper 2129, Rural Development Depart-ment. Washington, D.C.

Gautam, Madhur, and Jack R. Anderson. 1999. “Re-considering the Evidence on Returns to T&V Exten-sion in Kenya.” World Bank Policy Research Work-ing Paper 2098. Washington, D.C.

Gerhart, J. 1975. The Diffusion of Hybrid Maize inWestern Kenya. Abridged by CIMMYT. MexicoCity: Centro Internacional de Mejoramiento de Maizy Trigo.

Hussain, Syed Sajidin, Derek Byerlee, and Paul W.Heisey. 1994. “Impacts of the Training and Visit

66


Extension System on Farmers’ Knowledge and Adop-tion of Technology: Evidence from Pakistan.” Agri-cultural Economics 10: 39–47.

Johnson, Charles W., Keith M. Byergo, Patrick Fleuret,Emmy Simmons, and Gary Wasserman. 1980. KitaleMaize: The Limits of Success. Project Impact Evalua-tion No. 2, U.S. Agency for International Develop-ment. Washington, D.C.

Kandie, E. K. 1997. “Proposals to Improve Effective-ness of the Agricultural Extension Services inKenya.” Agricultural Extension Policy Project, Min-istry of Agriculture, Livestock Development, andMarketing, Nairobi, Kenya, and German Agency forTechnical Cooperation. Nairobi.

MALDM (Ministry of Agriculture, Livestock Develop-ment, and Marketing). 1997a. Alternative Ap-proaches to Agricultural Extension: ASIP Prepara-tion Study. Nairobi.

———. 1997b. Proposal on the National Agriculturaland Livestock Extension Programme. Nairobi.

———. 1997c. “Staffing Norms Analysis,” Vols. 1–3.Agriculture Sector Management Project II. Nairobi.

Mukui, J. T. 1994. “Kenya: Poverty Profiles.” Office ofthe Vice-President and Ministry of Planning andNational Development. Nairobi.

Narayan, D., and D. Nyamwaya. 1995. A Participa-tory Poverty Assessment Study—Kenya. Report pre-pared for the World Bank, sponsored by BritishOverseas Development Association, United King-dom, and UNICEF. (Revised, 1996. Learning fromthe Poor: A Participatory Poverty Assessment in

Kenya. World Bank Environment Department Par-ticipation Series Paper 0341996. Washington, D.C.)

Ongaro, W. 1990. “Modern Maize Technology, YieldVariations and Efficiency Differentials: A Case ofSmall Farms in Western Kenya.” Eastern AfricaEconomic Review 6(1): 11–29.

Picciotto, Robert, and J. R. Anderson. 1997. “Recon-sidering Agricultural Extension.” World Bank Re-search Observer 12(2): 249–59.

Purcell, D. L., and J. R. Anderson. 1997. AgriculturalExtension and Research: Achievements and Prob-lems in National Systems. Operations EvaluationStudy. Washington, D.C.: World Bank.

Salmen, L. 1999. “The Voice of the Farmer in Agricul-tural Extension: A Review of Beneficiary Assessmentsof Agricultural Extension and an Inquiry into TheirPotential as a Management Tool.” World Bank AKISThematic Team. Washington, D.C. Photocopy.

Tiffen, M., and others. 1996. National Soil and WaterConservation Programme, Kenya. SIDA Evaluation96/25, Department for Natural Resources and theEnvironment. Stockholm.

Venkatesan, V., and J. Kampen. 1998. Evolution ofAgricultural Services in Sub-Saharan Africa. Discus-sion Paper 390, World Bank Africa Region Series.Washington, D.C.

World Bank. 1997. Rural Development: From Vision toAction. Washington, D.C.

———. 1994. Agricultural Extension: Lessons fromCompleted Projects. Operations Evaluation Depart-ment. Washington, D.C.

67

OPERATIONS EVALUATION DEPARTMENT PUBLICATIONS

Ordering World Bank PublicationsCustomers in the United States and in territories notserved by any of the Bank’s publication distributorsmay send publication orders to:

The World BankP.O. Box 960Herndon, VA 20172-0960Fax: (703) 661-1501Telephone: (703) 661-1580The address for the World Bank publication databaseon the Internet is: http://www.worldbank.org (selectpublications/project info).E-mail: [email protected] number: (202) 522-1500Telephone number: (202) 458-5454

The Operations Evaluation Department (OED), anindependent evaluation unit reporting to the WorldBank’s Executive Directors, rates the developmentimpact and performance of all the Bank’s completedlending operations. Results and recommendations arereported to the Executive Directors and fed back intothe design and implementation of new policies andprojects. In addition to the individual operations andcountry assistance programs, OED evaluates theBank’s policies and processes.

Summaries of studies and the full text of the Précisand Lessons & Practices can be read on the Internet athttp://www.worldbank.org/html/oed/index.htm

How To Order OED PublicationsOperations evaluation studies, World Bank discussionpapers, and all other documents are available from theWorld Bank Infoshop.

Documents listed with a stock number and pricecode may be obtained through the World Bank’s mailorder service or from its InfoShop in downtownWashington, D.C. For information on all other docu-ments, contact the World Bank InfoShop.

For more information about this study or OED’sother evaluation work, please contact ElizabethCampbell-Pagé or the OED Help Desk.

Operations Evaluation DepartmentPartnerships & Knowledge Programs (OEDPK)E-mail: [email protected]: OED Help [email protected]: (202) 473-4497Facsimile: (202) 522-3200

The World Bank InfoShop serves walk-in customersonly. The InfoShop is located at:

701 18th Street, NWWashington, DC 20433, USA

All other customers must place their orders throughtheir local distributors.

Ordering by e-mailIf you have an established account with the WorldBank, you may transmit your order by electronic mailon the Internet to: [email protected]. Please in-clude your account number, billing and shippingaddresses, the title and order number, quantity, andunit price for each item.