23002 r&cis~~~~~~~~~~ W O R L D B A N K O P E R A T I O N SEVALUATION DEPARTMENT W I N T E R2 0 0 0 NUMBER 1 9 8 Agricultural Extension_ The Kenya Experience Ttt HE TRAINING AND VISIT (T&V) SYSTEM OF Iagricultural extension service management was introduced in Kenya by the World Bank in 1982. Two Bank projects, the First and Second National Extension Projects (NEP I and II), supported the program until 1998, and had two main goals: institutional development of extension services and sustained increases in agricultural productiv- ity. The effectiveness of the approach has been the subject of much debate. A recent study by the Operations Evaluation Department (OED) set out to determine the impact of the projects. In interpreting the study findings, it is Back gr o u n d important to distinguish between the ratio- Rural and agricultural development is nale for extension services per se and the integral to any strategy to alleviate poverty outcomes that resulted from the design and promote broad-based growth in Kenya. and implementation of the two projects To this end, the government adopted the in Kenya. While there continues to be T&V system of agricultural extension in demand for extension services, the evalua- 1982 with support from the World Bank. tion found limited progress in institutional The performance of the extension system development and an extension approach has been challenged ever since. that was neither efficacious nor finan- The debate on the effectiveness of cially sustainable. Kenya's extension system is part of a The Kenyan system lacks a focus on broader discussion of the cost-effectiveness farmer empowerment. It is based on a tradi- of the T&V approach. The disagreement tional top-down, supply-driven approach centers on the returns to the heavy invest- that provides little or no voice to the farmer. ment of borrowing countries in the T&V While NEP I made some positive contribu- system. But despite the large investment tions in its early years, there is no evidence and the important role of agricultural of a significant or sustained impact on agri- extension in the Bank's development strat- cultural productivity. A positive return to egy for Africa, few attempts were made to the expenditures on the extension service rigorously measure the impact of its could not be established, projects. 2 World Bank Operations Evaluation Department one-third of farmers, and most farmers (81 percent) were small- An eeatholders, extension efforts focused on men and large landowners. A key objective of NEP I and II The debate in Kenya was prompted by the lack of Thewas to develop institutional arrange- ments to deliver agricultural services to smallbolders efficiently and effectively. The goal was to develop l &V, urter ueld th cotroers. Boh dnor an men an opratonaa cadre of well-informed, village- level extension workers who would sharply critical of uren etesin pacics,an sek staf eredtoreucvisit farmers frequently and regularly to provi'de relevant technical messages, and bring farmers' prob- lems to the attention of researchers. The extension staff was to receive regular training, with much- improved research-extension link- ages. Begun as a pi'lot in two districts in 1982, the projpect was rapidly An extension agent meeting with Kenyan farmers. Photo courtesy of World Bank Photo Librarv expanded to cover about 90 percent of Kenya's arable land. The debate in Kenya was prompted by the lack of The projects succeeded in putting an integrated visible results in agricultural performance. The high esti- national system in place, improving staff quality mated returns to T&V extension in Kenya, put forward through training, and establishing better research- by one of the few studies on the economic impact of extension linkages. The influx of large sums of develop- T&V, further fueled the controversy. Both donors and ment and operational funds energized the service and nongovernmental organizations (NGOs) have been raised staff morale in the early years. The increase in sharply critical of current extension practices, and seek staff served to reduce some of the earlier biases against reforms to make the system more demand-driven and women, young farmers, and remote areas. But new accommodating of alternative ways of reaching farmers. biases were introduced in favor of more educated The OED evaluation adopted an empirical approach farmers and more productive and better-off areas. to assess the impact of NEP I and wh. A household survey Overall, project outcomes suffered because of poor was undertaken to revisit households surveyed in 1982 and project implementation arrangements and weak manage- 1990 to create a panel data set. These data were comple- ment. Most critically, the new system proved to be finan- mented by an extension staff survey, existing secondary cially unsustainable. Poor financial arrangements com- statistics, beneficiary assessments, and reports and other pounded the problem of inadequate budgetary resources, documents. The evaluation was based on a conceptual and many of the problems of the previous extension sys- impact model (see figure 1) that underpinned the design of tem have persisted. The current system is significantly NEP I and II. Following the results-based management more expensive, but hardly more efficient. Over one-half framework, the key indicators for the expected outcomes of the staff who were in service in 1982 believe that the and results were measured, and related to the projects' system is less effective now than it was before; only one- input and output indicators, third consider it to be more effective. Key institutional features of NEP I and ac were Institutional Development poorly developed. Inappropriate incentives and the fall- Agricultural extension in Kenya dates back to the early ure to incorporate mechanisms to give farmers a voice 1900s, but its only notable success was in the dissemina- have led to a lack of accountability and responsiveness tion of hybrid maize technology in the late 1960s and to farmers' needs. This is evident in the mismatch early 1970s. In 1982, when T&V was introduced, the between what farmers want (advice on complex prac- extension system suffered from a number of deficiencies. rices) and what they actually get (simple agronomic It was a mix of ad hoc project components, lacking a messages), as well as in the methodologies preferred by consistent national strategy. These arrangements were the farmers (demonstrations) and the extension agents expensive and ineffective. Despite a well-established line (home visits).- of command down to the frontline extension worker, and Table 1 summarizes Kenyan policymakers' views staff numbers presumed to be adequate at the time, the about the projects' strong and weak points. In response service was judged to he performing well below its to operational difficulties and a lack of new technical potential. In addition, although women made up almost messages, some field staff have gone back to the old sys- Precis 198 * Figure 1: Impact Model |Food security Land quality Friends, neighbors, Fo secrit + innovative farmers Cr t9 l Adaptive Public extension service Spillovers research Weather andpests i Farmer organizations, NGOs - I I ~ Private sect or: input suppli Household HR l / i ".+_ I processors, consultants objectives OutPut welfare < t +;~~~00 Media: audio, video, print Pt/ i Basic Indigenous' i I research systems ii + Field days Demonstrations/field trials L Prices Inastru Education I ~~~~I I I1 Technology Generation Knowledge Delivery Farm Decisionmaking Impact Inputs Activities Output Outcomes Results Note: Broken line: feedback from farmers to decisiorimvakers. Solid line: delivers of time-bound "messages" by extension wirkers from rescarchers to farmers. HH: household. tem of disseminating messages through public meetings, access to extension has remained unchanged, although or barazas, and working with alternative providers, such many perceive the quality of extension service to have as NGOs, the private sector, and other projects. These deteriorated. The poor, in particular, have little access to actions represent a rational response to prevailing information. circumstances, but they are far from the intended project Extension cannot be expected to reach every farmer- outcomes. They are also ad hoc arrangements and do hence, the need for selectivity and reliance on farmer-to- not represent any systematic or conscious effort to farmer dissemination. The results, however, indicate poor exploit alternative institutional options to make the targeting and a lack of responsiveness. Instead of working system more effective. with representative groups of farmers, the extension work- ers generally target the better-off and those who can afford Beneficiary Assessment the new technologies. Not surprisingly, more of the better- The study drew on a participatory assessment of extension off farmers consider extension advice to be relevant to their services by Actionaid Kenya and the findings of a 1994 needs. Poorer farmers are less satisfied. They want advice participatory poverty assessment to complement the quali- on less costly technologies, marketing, and diversification, tative findings from the household survey done for this along with information about crops that the wealthier evaluation. The findings from the structured questionnaire farmers do not grow. survey and the participatory assessments are consistent. Comparing their current situation to that of 10 to 15 Efficacy: Quantity and Quality of Contact years ago, most farmers noted a decline in their qua]ity NEP I and II sought to improve the effectiveness of of life and in agricultural productivity. At the same time, extension services by using contact farmers, and later 4 World Bank Operationas Evaluation Department Table 1: Local Perspectives on the Strengths and Weaknesses of NEP I and 11 Weakness Strength Extremely broad objective Wide coverage Top-doxvn planning still strong All types of farmers included No specific target Strong staff training Weak farmer participation Professionalism developed at district-officer level Low staff motivation Strong frontline extension workers Weak monitoring and evaluation Procurement of transportation and office accommodation Excessively supply-driven messages Donor dependent Low flexibilitv Low accountability Source: Ministry of Agriculture, Livestock Development and Marketing. contact groups, as the points of regular and systematic 1996, when field work was seriously disrupted by a lack interaction with the farming community. However, most of funds, a result of the suspension of NEP 11 disburse- farmers surveyed think that information is less available ments, for almost the entire year. As may be expected, now than it was 10-15 years ago. the proportion of noncontact farmers failing to note a The level of outreach is well below that antici- change was higher (88 percent). pated. Few, even among the select group of contact farmers (including contact group members), interact Outcomes with extension agents on a regular basis and in a set- Farmer awareness of new crops and techniques is an ting prescribed by the project design. In the study essential first step toward their adoption. The farmers sample, only 7 percent of contact farmers, and 2 per- surveyed, however, demonstrate a very uneven knowl- cent of all farmers, regularly meet with extension edge of extension messages (see figure 2). Awareness is agents at least once a month in their own or a high for simple agronomic recommendations, particu- neighbor's fields. Similar results on low outreach are larly for maize, but falls significantly-among both con- also obtained from the 1990 survey of the same popu- tact farmers and noncontact farmers-for non-maize lation, indicating limited efficacy not only at the end crops and for more complex practices, reflecting the of NEP II, but also at the end of NEP I. The service continuation of the pre-1982 bias in favor of maize and was expected to make regular field visits to 10-15 simpler messages. In addition, data do not reveal any percent of the farmers on a biweekly basis, or 20-30 significant impact of extension activities on the underly- percent on a monthly schedule. ing dynamics of the diffusion process. Less sophisticated In addition to direct contact, the T&V methodology messages that are amenable to quick diffusion through relies on indirect dissemination through contact between informal channels and messages that have been known farmers. The expected effects of the spread of informa- for some time have continued to spread. The less well tion from contact to noncontact farmers appear to have known and the more sophisticated messages still register been limited because of poor communication between low levels of awareness and adoption. A finding of con- the two groups. Farmers get most of their information siderable importance is that a high percentage of those from public fora, primarily barazas, a practice that NEP who are aware of a recommendation have adopted it I and II were supposed to change. Barazas are useful for (over 80 percent for even the most complex practice). broadcasting simple messages, but are not conducive to Thus, while resource constraints may be important, the substantive interaction on technical problems. The data primary impediment to adoption seems to be lack of on farmer-extension contact suggest little improvement information. in staff productivity or systemic efficiency. The early focus on simpler messages and on the Despite the low level of contact, farmers who primary subsistence crop, maize, may be justified. Over receive extension messages rate them as useful and time, however, one would expect extension advice to applicable. A majority, however, even among contact graduate to more sophisticated practices. This has not farmers, have not actually applied the extension recom- occurred, either for lack of focus in extension efforts or mendations, which raises questions about the relevance lack of staff skills, or perhaps both. of the advice provided. These findings also highlight the The projects' lack of impact on awareness and adop- problems in interpreting frequently used, but often tion of recommended practices indicates limited poten- poorly defined, indicators such as the term useful. tial for impact on agricultural productivity. This, in The consequences of an incident in 1996 help illumi- turn, implies low cost-effectiveness for the T&V nate the limited effectiveness of the extension service. A approach as applied in Kenya. The major share of the majority of the contact farmers (60 percent) failed to cost stems from the projects' focus on high-frequency, notice any change in the delivery of extension services in face-to-face contact, which is particularly suited to the Precis 198 Figure 2: Timing of Awareness Figure 3: Adoption of Recommendations Percent aware Percent 90 80 SO U 98 All fairmers 70 - 3 4 60 871 Contact farmners 40 30 20 ~~~~~~~~~~~~~~20 10 0 I II I I I I I0 1 2 3 4 5 6 7 8 9 10 11 J2) 13 1 2 3 4 5 6 7 8 9 10 11 12 1 3 Activity complexity Activity complexity Source: OED survey. Souirce: OED survevy delivery of sophisticated and context-specific advice. and responsive system. For instance, with little new The continued preoccupation with simple messages iDdi- technology forthcoming for the higher-productivity cates that the projects' design features were not fully zones, it was no longer economical to maintain a high exploited. intensity of extension-farmer contact. Reaching out to new areas or previously uncovered farmers could have Results had a greater marginal impact on both productivity and The most direct way to measure impact is to relate the poverty. supply of extension services to farm productivity. Changes in productivity can occur through improved Willingness to Pay efficiency or technical change. With the data avail- Governments have traditionally provided extension able, the evaluation could not establish a significant service at no cost to beneficiaries because of it's per- positive impact of the supply of extension services, ceived nature as a public good. With limited resources either on farmer efficiency or on farm productivity. It to allocate amnong a numliber of alternative public also found that areas that were relatively more pro- services, those who make public policy need an assess- ductive in 1982 have shown little change, while the ment of the benefits realized from the services provided. less productive regions have performed better. It is A reliable measure of benefits would also help in consid- likely that the more productive regions may have ering the potential for cost-recovery measures. reached an upper limit, and with little new technology In the context of extension and poor farmers, perti- forthcoming, their productivity has stagnated. The nent issues include their willingness and ability to pay, continued focus of extension efforts in these areas is which one would assume would be proportional to the thus likely to have yielded low returns. At the same benefits they derive. A significant proportion of farmers time, the lcss productive farmers and areas have been surveyed indicated that they would like to continue to catching up as new technology has reached them. It is receive extension services, and, more important, that possible that the extension service may have played they would be willing to pay for them. The amount they an early positive role in disseminating simple techno- would pay, however, is well below current government logical messages to previously neglected areas, but expenditures. An important finding is that, based on this could not be established conclusively with the their current experience with the service, farmers want available data. significantly fewer visits by the extension agent than These results, combined with the low level of overall were stipulated by the projects. Overall, a more cost- economic efficiency and a significant variation in techni- efficient approach would be to cover a larger number of cal efficiency among farmers, suggest that the extension farmers with less intensity, but a higher quality of con- resources could have been used in a more efficient man- tact, complemented by other dissemination approaches. net. Extension seems to have focused on disseminating such as radio broadcasts and printed media. These find- simple technical messages rather than on helping farm- ings also suggest that cost-recovery, even if only partial ers make the best use of their resources. or nominal, is a possibility that remains unexplored. A more rational deployment of extension staff could Besides bringing some budgetary respite, cost-recovery have been more cost-effective-it is possible that the offers other important benefits: it provides appropriate same results could have been achieved with fewer incentives, and it promotes pluralism by allowing alter- resources, yielding potentially higher returns to the native providers, particularly private operators, to enter investment in extension. What was needed was a flexible the market. 6 World Banik Operations Evaloation Department Conclusions and L essons for a more flexible system that can identify the gaps The rationale for providing extension services in Kenya between best and average practice and allocate scarce is still relevant, but the evidence suggests that the exten- resources more rationally. In addition, the farmers sion approach used by NEP I and II was not efficacious. selected for interaction should represent local socio- The performance of the T&V system as applied in economic conditions. Kenya has been disappointing. The system as imple- u Information systems. Targeting calls for appropriate mented has been ineffective, inefficient, and unsustain- flows of timely and reliable information, and hence able. While the projects helped improve the system's for monitoring and evaluation. Farmers' demands coverage, research-extension linkages, and the skills of should be identified, and the extension service tailored extension staff, the overall outreach and the quality of to suit local technological and economic circum- interaction between extension agents and farmers have stances. been well below expectations. The evidence does not u Intensity. Given farmers' desire for less frequent visits indicate a significant impact either on farmer efficiency and the lack of sufficient technological recommenda- or on crop productivity. While it is likely that the first tions to sustain a high level of visits, a leaner and less project had some early beneficial impact, it intensive system with wider coverage would be more appears to have been short-lived. A positive rate of cost-effective. With improved quality, demand could return on investment expenditures on extension could not well increase, which reinforces the need for a respon- be established. The evaluation did reveal that there is sive and dynamic system and effective targeting for unmet demand for extension services, and that farmers maximum results. are willing to pay for them. The worth of the perceived * Pluralism. The use of a uniform methodology to benefits from the current services, as reflected in the deliver standard messages limits the system's effective- amount that the farmers are willing to pay, however, is ness and efficiencv. A strategy that exploits low-cost significantly lower than current government spending communication methods such as radio, demonstrations, per farm to deliver them. Overall, the findings indicate printed media, and partnerships with civil society and that a more rational allocation of extension resources the private sector might be more effective. could have been more cost-effective. * Client focus. The system's central focus should be to The main lessons that have emerged from the evalu- empower farmers by giving them a voice in the exten- ation are as follows: sion delivery system. This can be achieved in a number of ways, such as cost sharing, farmers' organizations, - Targeting. Extension services need to be efficiently and decentralization. Such alternatives should be an targeted to focus on the areas and groups where the integral part of the delivery mechanism. marginal impact is likely to be the greatest. This calls 0 10 This PrtPcis is based on \ Pa!Rnk Vgi Vt,a xte'nsi,nz !Yirs i~i;rzv by Madhuor (autani_ Report No. 19523, June 30, 1999. Available to Bank Executive Directors and staff from the Internal Documents Unit and from regional information service centers, and to the public from the World Bank InfoShop. \\ \v\v.\v\ (lk)rIJ ll .o.l-rghltlln/h!smI OED Publications Recent OED Precis 197 Toward a Comprehensive Development Strategy 196 Evaluation and the Development Challenge 195 Poverty Assessment: Maximizing Impact 194 Involuntary Resettlement: The Large Dam Experience 193 Partnership for Education in Jordan 192 Reforming Bolivia's Power Sector 191 Bangladesh: Progress Through Partnership 190 Transport in Indonesia 189 Health Care in Brazil: Addressing Complexity 188 Health Care in Mali: Building on Community Involvement 187 Health Care in India: Learning from Experience 186 Global Health: Meeting the Challenge 185 Aid Coordination and Post-Conflict Reconstruction: The West Bank and Gaza Experience 184 Land Administration and Rural Development: Two Cases from Thailand 183 Monitoring and Evaluation Capacity Development in Africa 182 Sub-Saharan Africa: Lessons from Four Sectors 181 Grassroots Pastoral Organizations in Mauritania 180 Capacity Building in the Agricultural Sector in Africa 179 Yemen: Forging a New Consensus (available in Arabic) 178 Municipal Development Projects: Building Institutions and Financing Local Development 177 Nongovernmental Organizations in World Bank-Supported Projects 176 Meeting the Health Care Challenge in Zimbabwe 175 Development Effectiveness, 1998: Opportunities in a Volatile Environment 174 Disaster Assistance Precis a2ussi disponible en francais lb hnp://sw x. orllbana wg/ tll lAl' Precis en espahiol tambien disponible @b hsrp://svsvsv.sA orldhan k.orlg/h tnillocs!d OED Study Series 1999 Annual Review of Development Effectiveness Developing Towns and Cities: Lessons from Brazil and the Philippines Evaluation and Development: The Institutional Dimension (Transaction Publishers) Financial Sector Reform: A Review of World Bank Assistance India: The Dairy Revolution Investing in Health: Development Effectiveness in the Health, Nutrition, and Population Sector Nongovernmental Organizations in World Bank-Supported Projects The World Bank's Experience with Post-Conflict Reconstruction OED Multilingual Series Assessing Development Effectiveness: Evaluation in the World Bank and the International Finance Corporation Appreciation de l'efficacite do developpement: L'evaluation a la Banqupe mondiale et a la Societe financiere internationale Determzinar la eftcacia de las actividades de desarrollo: La evaluacion en el Banco Mundial y la Corporacion Finanzciera Internacional Cote d'lvoire Revue de ('aide de la Baoqoie mondiale au pays Philippines: From Crisis to Opportunity Filipinas: Crisis y oportunidades Rebuilding the Mozambique Economy: Assessment of a Development Partnership Reconstruiir a Economia de Mocainbique S heries OED Carolin McE en it Assistan: u ~Pr~cis are ava fille at Cnc rg QEDHep esk t:1 -202/5-47 rfxtefr belwxt: 1-7032611522012.Yucnase-ilyr requn o elt ine@ wo rldbank.org FroimtEhe , World Bankhoepgsect blicati le-mat il: Editor-sinOCief rg XffB;ankOpraiosvlaio DpatmnXtParitnersh,X;0ips 0 Qantitywldg Title PrEPKOtrahcn issemi n laseo addt mye namew to thePcis ma ilingr lioseft.e Op PrtonstEaluCode iounstry dtr ndsiudntb tri la o s p c c the World Bank, i it orgaizatons 'noSo ¢r4+ate0;; 0 tel: 1-02/458-5454 fax: 1-025-100, or; 00 -O thro ~ugh email ctbooks@wrdan.r hs198tp: Agricultwwod ralExtelnosd 00: TEom orde Four books plmepasef sd co tact:;0 0000 Th Seigi:IIs($oibnog World0; Bank0 Fulillen Center000:0 i Herndon, VA 20172-096Ttluaio0, USpArtVt a t;;00;00 'Int:erniet:l htutp:/w.o rldbVank. orgtn:hluo i>eV ;f t: Fro eth:e World Bnkhomepg, slet publat0ications.ft erf0ations Evleuation staffors an editors an should0f not00f:000 be;;t ; Pr6cis 198 2 : 0;Agricultural Extension: The 0Kenyaf Epeienc.e0 ; 00