80766 Latin America and the Caribbean Region LCSSD Occasional Paper Series on Food Prices Determinants of  Agricultural Extension Services: The Case of  Haiti Diego Arias Juan José Leguía Abdoulaye Sy May 24, 2013 Cover photos courtesy of Mrs Barbara Coello. 3 The workEl been Agropecuario Desarrollo has y Rural partly financed by en Paraguay. the Diagnóstico Trust Fund y opciones de política for Environmentally and Socially Sustainable Development (TFESSD) LATIN AMERICA AND THE CARIBBEAN REGION LCSSD Food Papers Series Determinants of Agricultural Extension Services: The Case of HaitI Diego Arias Juan José Leguía Abdoulaye Sy World Bank, LCSAR May 24, 2013 Determinants of Agricultural Extension Services: The Case of HaitI 1 EXECUTIVE SUMMARY This paper extracts relevant lessons from historical data 3. There are no statistical differences between men and on factors influencing the receipt of extension services women in terms of receipt of extension services; how- in Haiti, taking stock of the use of agricultural extension ever, the impact of agricultural training and farm size services prior to the 2010 earthquake. The goal is to influ- change when the head of household is a woman. ence future policies and development projects involving the provision of extension services as well as the type 4. Education level has a positive, yet small, effect on re- of extension services offered. ceiving extension services. This paper uses data from the 2010 Agricultural Cen- 5. Prior agricultural training is a major determinant of the sus and examines the characteristics of farmers in Haiti recipients of extension services. receiving extension services by gender, education, agricultural training, farm size, and type of crop. Through 6. Rehabilitation of the Ecoles Moyennes Agricoles in-depth study of each variable and a review of trends (EMAs) for vocational and farmer field education in the receipt of agricultural extension services, the study on a nationwide scale would increase the demand for analyzes the equilibrium between the demand for and extension services, especially among small farmers. supply of extension services to particular farmer groups. 7. Farmers with larger farms receive more agricultural Using a fixed effects probit model to isolate the marginal extension services. effect of each characteristic on the likelihood of receiv- ing extension services, and controlling for various factors, 8. Coffee producers make more use of extension services the study draws the following nine key conclusions: than other farmers. 1. The proportion of households receiving agricultural 9. Promoting a hybrid system of extension may be more extension services in Haiti is non-negligible. efficient than supporting only public or NGO-provided extension services. 2. Location is an important determinant of the recipients of agricultural extension services. 2 Determinants of Agricultural Extension Services: The Case of HaitI Table of Contents I. Overview of Agricultural Extension Services in Haiti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Institutional Structure of Agricultural Extension Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 II. Data and Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 III. Analysis of Potential Determinants of Agricultural Extension . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Agricultural Training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Farm Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Type of Crop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 IV. Conclusions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Annexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Determinants of Agricultural Extension Services: The Case of HaitI 3 I. Overview of Agricultural Extension Services in Haiti Background market access.� According to Christoplos et al. (2012), agricultural extension services can be classified primarily The Haitian population is among the poorest in the world, into three areas: with over 78 percent living on less than US$2 a day and over 50 percent living on less than US$1 a day. In rural • Technology and information sharing areas, 88 percent of individuals live below the poverty line • Advice related to farm, organizational, and business and basic services are practically nonexistent. The devas- management tating January 12, 2010 earthquake was a major setback • Facilitation and brokerage in rural development to the economy and aggravated an already precari- value chains. ous social situation. Relaunching agricultural production is among the Haitian Government’s top priorities of the The most recent Agricultural Census in Haiti, conducted country’s reconstruction program. The transfer of knowl- by the Ministry of Agriculture, Natural Resources, and edge, technologies, and practices through agricultural Rural Development (MARNDR) during the 2008–2010 pe- extension services is a critical building block to raising ag- riod, classified extension services in the following nine ricultural productivity and production in an environment categories: (i) advisory services related to seed/crop se- dominated by very small farmers. This paper takes stock lection, (ii) arboriculture techniques, (iii) soil preparation of the different uses of extension services in Haiti during and conditioning, (iv) livestock, (v) aviculture, (vi) api- the 2008−2010 period and aims to provide some historical culture, (vii) aquaculture, (viii) post-harvest techniques, lessons as a tool for investing most effectively in agricul- and (ix) commercialization. Using the aforementioned tural extension services in a post-earthquake era. classification, categories (i) to (viii) transferred informa- tion and knowledge to farmers and provided them with The concept of extension services has changed over guidance on farm management skills, while category time. While technological transfer is still important, more (ix) may have given farmers business management skills emphasis is being placed on expanding the skills and and facilitated their linkage to value chains and mar- knowledge of farmers (i.e., human capital development), kets. enhancing rural livelihoods, achieving food security, and creating more efficient farmer-based organizations Institutional Structure of Agricultural (Swanson, 2008). Christoplos et al. (2010) defines extension Extension Services as “all the different activities that provide the information and advisory services that are needed and demanded The MARNDR is responsible for the provision of exten- by farmers and other actors in agrifood systems and rural sion services (through the organic law of Septem- development.� It also includes, for instance, “facilitation, ber 30, 1987), and is divided into several decentralized brokering and coaching of different actors to improve structures: 10 Departmental Agriculture Directorates 4 Determinants of Agricultural Extension Services: The Case of HaitI (Direction Départementale d’Agriculture, DDA ), four husbandry, and natural resource management. While the sub-Departmental Directorates, and several Agriculture MARNDR and its sub-branches fund the provision of vari- Bureaus (Bureaux Agricoles ) located in 30 municipalities ous services for plant production, animal husbandry, and (among 135 in the country). In addition, about 15 re- natural resource management and steer and control the search and training centers are located throughout the regulation of the agricultural sector, the provision of ser- country and are directly linked to central services (mainly vices and the implementation of investments are gener- R&D) in the MARNDR. These institutions contribute to the ally handled by NGOs, producer organizations, or private provision of various services for plant production, animal entities. Determinants of Agricultural Extension Services: The Case of HaitI 5 II. Data and Summary Statistics Extension service coverage in Latin America and the Center, and South. In these departments, 13.9 percent Caribbean varies widely across countries. The OECD of household heads reported having received at least (2011) points out that in Mexico, 3 percent to 10 percent one of the nine aforementioned extension services. of agricultural units are provided with technical as- Graph 1 shows the relative importance of each type sistance, whereas in Chile, the Institute of Agricultural of extension service out of the total delivered in Haiti. Development delivered technical assistance and credit The services most frequently delivered are those related programs to 42 percent of small farmers in 2006. In Nica- to the first stages of the value chain (production), namely ragua, a country with poverty levels comparable to those choice of seeds and varieties and agricultural techniques of Haiti, the Nicaraguan Institute of Agricultural Technol- and practices, which account for over 50 percent of all ogy (INTA) serves about 20 percent of all farm families, services delivered. Extension services for livestock (cattle according to the 2001 Agricultural Census. and poultry) account for another 42 percent of services received while post-harvest services (storage, processing, Owing to the limited availability of data, this study con- and marketing) account for only 6 percent of services siders three out of 10 departments in Haiti: South East, delivered. Graph 1: Type and composition of extension services received by farmers Aquaculture Conditioning, Storage, (0%) and Transformation Apiculture (3%) (1%) Crop Election (14%) Aviculture (20%) Arboriculture Techniques (17%) Livestock Field Techniques (22%) (20%) Source: Agricultural Census 2008–2010. Authors’ calculations. 6 Determinants of Agricultural Extension Services: The Case of HaitI Tables 2.1 and 2.2 display information on both the number they do not need extension services). In fact, everyone of households that received extension services and the who needs extension services in these departments number of households that reported that they needed seems to have access to them. Furthermore, reporting extension services in the South East and Center de- that the services are needed does not ensure a marginal partments. For instance, in the South East department, private benefit of extension (i.e., the demand) since there 13.51 percent of heads of household received extension are transaction costs involved in requesting and partici- services, while only 11.54 percent reported they needed pating in the service. Hence, the demand for extension them. In the Center department, 14.16 percent of house- services may be even lower than that reflected in the holds received extension services, while 11.13 percent census. reported they needed extension. It appears that in these departments, demand for extension services is fully met. By contrast, data collected from the South department and displayed in Table 2.3 tell a different story. In that Therefore, this analysis addresses both how the determi- department, 98.79 percent of household heads reported nants of the receipt of extension services proposed in this that they needed extension services, while only 13.79 per- paper interact not only with the supply (i.e., why these cent received at least one service. While many explana- farmers have less or more access to extension services), tions can be entertained, a mechanical explanation but also with the demand (i.e., why these farmers think should not be discarded. The census in Haiti was carried Table 2.1: Demand for Extension Services – South East Need Not Need Total N % N % N % Received 9,969 11.54 1,705 1.97 11,674 13.51 Not received 0 0.00 74,735 86.49 74,735 86.49 Total 9,969 11.54 76,440 88.46 86,409 100.00 Source: Agricultural Census 2008−2010. Authors’ calculations. Table 2.2: Demand for Extension Services – Center Need Not Need Total N % N % N % Received 13,764 11.06 3,859 3.10 17,623 14.16 Not received 91 0.07 106,757 85.77 106,848 85.84 Total 13,855 11.13 110,616 88.87 124,471 100.00 Source: Agricultural Census 2008−2010. Authors’ calculations. Table 2.3: Demand for Extension Services – South Need Not Need Total N % N % N % Received 12,490 13.71 73 0.08 12,563 13.79 Not received 77,528 85.08 1,031 1.13 78,559 86.21 Total 90,018 98.79 1,104 1.21 91,122 100.00 Source: Agricultural Census 2008−2010. Authors’ calculations. Determinants of Agricultural Extension Services: The Case of HaitI 7 out over a period of three years (2008−2010), which Map 1 shows the percentage of household heads by com- means that some households were surveyed after the mune that have received some extension services. Com- earthquake of 2010. If some places were systematically munes are classified into three distinct groups according surveyed after the earthquake (for example, the South to the terciles of the distribution in which they fall—less department), the tremendous shock caused by the disas- than 5.64 percent, between 5.65 percent and 14.37 per- ter could explain these differences. However, this unex- cent, and 14.38 percent and over. In Map 1 we observe plained difference in demand for extension services in the that there are pockets of low and high receipt of extension South does not alter the econometric results of this paper services. These pockets could be influenced by factors as our dependent variable is receipt of and not demand such as irrigation, geography, past interventions, political for extension services. configurations, or distance to the closest DDA. Map 1: Agricultural extension services at the commune leveL Source: Agricultural Census 2008–2010. Authors’ preparation. 8 Determinants of Agricultural Extension Services: The Case of HaitI III. Analysis of Potential Determinants of Agricultural Extension The previous section highlighted overall trends in agri- analysis such as gender, education (a dummy variable cultural extension services in Haiti, concluding that there for each level), agricultural training (a dummy variable are vast differences across communes. However, there for each level), farm size (a dummy variable for each might also be differences within communes. Indeed, size range), crop type (a dummy variable for each type by exploiting the variations within them, we are able of crop considered), and interactions of each of these to study the relationship between farmer-level character- variables with gender. The reason we include gender istics and the likelihood of receiving extension services. interactions is to assess the effect of each of these vari- We are specifically interested in assessing the correlation ables conditioned on the gender of the household head. between extension services and the following farmer- Finally, δj is the commune-specific fixed effect term, and eij specific variables: gender of head of household, educa- is the idiosyncratic error term. We run the regression using tion level, agricultural training, farm size, and type of crop data pooled from all the departments under study and produced. To better isolate the importance of each also for each department separately (South East, Cen- of these variables in predicting which farmers are more ter, and South). We used clustered standard errors at the likely to receive extension services, we take into account district level (section communale ). the effect of all unobserved commune-specific variables that may be affecting both the variables under study and Table 3.1 shows the results of the regression. The coeffi- the receipt of extension services, particularly the distance cients for the commune dummies are not presented in the to the nearest DDA, geography, irrigation, and political tables; however in all cases, they are jointly significant structures. In order to do this, we introduce “Commune at the 0.05 level. Therefore, as discussed previously, loca- Fixed Effects� into our probit model. The purpose of this ex- tion is quite important in determining the level of exten- ercise is not to find the causal effects, but the conditional sion services, and it is necessary to further investigate correlations between the variables under examination commune-specific variables causing these pockets of low and the likelihood of receiving extension services. We de- reception of extension services. For instance, as already fine the following econometric specification: mentioned, it may be that the distribution of DDAs is un- equal across communes. Even if the majority of extension Pr(Yij = 1) = G (β’Xij + δj + eij ) services are provided by NGOs or private entities, distance to the nearest DDA may still have an effect if NGOs and The equation above describes a fixed effects probit mod- development projects are located near DDAs or Bureaux el, where Yij = 1 if the household head receives at least Agricoles Communales (BACs). This may be the case one type of extension service and is 0 otherwise; G is the for two reasons: (i) When targeting beneficiaries, NGOs normal cumulative density function; β is a row vector with may follow the advice of DDAs, which may tend to favor all the coefficients of the variables under study; Xij is a col- people located nearby, and (ii) DDAs may implement de- umn vector with all the farmer characteristics under velopment projects or co-manage projects with NGOs. Determinants of Agricultural Extension Services: The Case of HaitI 9 Table 3.1: Regression Results Determinants of usage of agricultural extension services Dependent variable: use of extension services = 1 if use at least one extension service, 0 otherwise Regressor All South East Center South Female 0.0135 –0.08 0.123 0.0008 (0.0439) (0.0495) (0.0803) (0.0498) Education None omitted omitted omitted omitted — — — — Literate 0.1927*** 0.0639 0.1433 0.369*** (0.0738) (0.0627) (0.1242) (0.1235) Elementary 0.0183 –0.0992 –0.022 0.1526 (0.0787) (0.1413) (0.1603) (0.1164) High School 0.0752 0.0245 0.0661 0.1525 (0.0624) (0.0972) (0.1381) (0.1024) Professional 0.1115 0.0173 –0.5746*** 0.4518*** (0.1236) (0.2078) (0.1037) (0.1143) University 0.1231* 0.2117* –0.0106 0.2026** (0.0672) (0.1272) (0.2003) (0.0895) Agricultural Training Empirical omitted omitted omitted omitted — — — — Occasional 0.9046*** 0.6633*** 0.9499*** 1.0046*** (0.1032) (0.1804) (0.0921) (0.1673) Technical 0.9257*** 1.3219*** 0.5434*** 0.881*** (0.1599) (0.3451) (0.0709) (0.1784) University 0.0113 0.2258 0.0863 –0.1214 (0.1275) (0.2245) (0.1532) (0.18) Farm Size (hectares) Less than 0.15 omitted omitted omitted omitted — — — — From 0.15 to 0.3 0.1187 0.0012 0.1019 0.2254*** (0.0758) (0.1057) (0.2272) (0.0769) From 0.3 to 0.6 0.2525*** 0.1416 0.4005 0.3354*** (0.0844) (0.1134) (0.2504) (0.1006) From 0.6 to 1.2 0.2359** 0.0372 0.5151 0.3005*** (0.1012) (0.1501) (0.2777) (0.1018) From 1.2 to 2.4 0.1519 0.0166 0.3559 0.2694*** (0.1149) (0.1699) (0.3343) (0.0986) More than 2.4 0.1372 –0.0823 0.3664 0.2688* (0.1284) (0.1715) (0.3075) (0.1561) Crops Maize 0.1653 –0.0075 –0.1178* 0.2729 (0.1085) (0.1415) (0.0656) (0.1829) (continued on next page) 10 Determinants of Agricultural Extension Services: The Case of HaitI Table 3.1: Regression Results (continued) Determinants of usage of agricultural extension services Dependent variable: use of extension services = 1 if use at least one extension service, 0 otherwise Regressor All South East Center South Beans –0.0573 0.0282 –0.1616 0.025 (0.1108) (0.117) (0.2752) (0.1774) Bananas 0.02 –0.0453 0.2337* –0.2122** (0.0814) (0.1373) (0.1223) (0.084) Coffee 0.2728** 0.445*** 0.0166 –0.0379 (0.1204) (0.1633) (0.118) (0.1404) Mangoes –0.0285 –0.184 0.0719 0.0802 (0.0797) (0.1198) (0.1411) (0.1283) Intercept –1.6973*** –1.7464*** –2.3228*** –1.9243*** (0.3777) (0.2301) (0.4424) (0.4018) Observations 300100 85763 123814 90523 Pseudo R2 0.1817 0.0757 0.2997 0.1527 Source: Authors. * p-value < 0.1, ** p-value < 0.05, *** p-value < 0.01. Gender Center department, a larger proportion of female- headed households received extension services com- Women play an important role in Haitian agriculture. pared to male-headed households. Nevertheless, these One fourth of heads of household are women in the results may be hiding other variables correlated to both South and Center departments, and in the South East the gender of the head of household and the likelihood department, the proportion is even larger (34 percent). of receiving extension services, introducing a bias in the Moreover, a recent survey conducted by the Conseil interpretation of the unconditional relationship between National de Sécurité Alimentaire (2011) indicates that gender and receipt of extension services. For instance, the proportion of female-headed households (pooling being a female-headed household can be correlated data from the South East, Center, and South) is 45 per- with farm size. If female-headed households had larger cent. According to Lastarria-Cornhiel (2006), the pro- farms on average, and larger farms tended to receive portion of rural female-headed households for the more extension services, they would likely receive equal late 1990s across 13 countries in Latin America reached or more extension services than men, not because of their nearly 23 percent (Lastarria-Cornhiel, 2006). Hence, it can gender, but because of the size of their farms. be argued that the proportion of female-headed house- holds in Haiti is higher than the regional average. This Nevertheless, on average, female-headed households is in accordance with Saito and Spurling’s (1992) argu- have smaller farms than men. Table 3.2 tells us that for ment that it is increasingly common for women to man- the three departments analyzed in the data, female- age or operate farms on a daily basis in all parts of the headed farms are much smaller than male-headed ones. world, as men leave farms in search of paid employment. For instance, in the Center department, which seems It is important, therefore, to examine if there are any to be the area where farmers have the biggest farms, the systematic differences between men and women in terms size of male-headed farms is, on average, 1.33 hectares, of their receipt of extension services. while the size of female-headed farms is 1.12 hectares. The differences are fairly similar in the South East and the As illustrated by Graph 2, there is no systematic trend South departments and even larger in the Center depart- regarding the degree to which male- or female-headed ment when we look at the median values of farm size. households receive extension services. Moreover, in the In Tables A.1, A.2, and A.3, we examine the proportion Determinants of Agricultural Extension Services: The Case of HaitI 11 Graph 2: Households receiving extension services by gender of household head 16 Male Female 14 12 Percentage (%) 10 8 14.08 14.99 14.1 14.03 6 12.38 12.82 4 2 0 South East Center South Department Source: Agricultural Census 2008−2010. Authors’ calculations. Table 3.2: Average (median) farm size of its head. The maps are fairly similar, yet there are im- in hectares by gender of household head portant differences in relation to Map 1. In almost every commune, the proportion of households who received extension services is lower than the commune average Gender South East Center South if the head of household is female. Interestingly though, Male 0.91 (0.65) 1.33 0.94 (0.65) (0.97) when the average rate of extension reception is high, Female 0.68 (0.48) 1.12 0.68 (0.48) female-headed households receive more extension ser- (0.81) vices than male-headed households. For instance, in the Percentage Difference 34% (35%) 12% 38% (35%) Cerca La Source commune in the Center department, (20%) the average rate of extension reception is 53 percent, yet Source: Agricultural Census 2008–2010. Authors’ calcula- for female-headed households it is 64 percent. It seems tions. Farm size is calculated at the household level (where that when the supply of extension services is scarce, men each can have more than one plot), whereas Tables A1, are favored over women; when supply is fairly high, the A2, and A3 are calculated at the plot level. supply of extension services may be the same for both male-headed and female-headed households, thus the quantity of services allocated is solely demand-driven. of female-headed and male-headed households by de- partment for each bracket of plot size (not farm size). In other words, when extension is widely available, receipt We observe clearly that as plot size increases, the propor- of extension services may depend primarily on the de- tion of female-headed households decreases, except for mand for extension services in both female-headed and the last bracket size in the Center and South, where the male-headed households, which appears to be higher for proportion of female-headed households slightly increas- female-headed households. This observation has important es in comparison to the previous bracket. implications for the interpretation of equilibrium between the supply of and demand for extension services—the We further examine if the underlying features present rather small differences between men and women in terms in each commune that are affecting receipt of extension of their receipt of extension services may be explained services interact differently with male-headed and fe- by a higher demand for extension services by female- male-headed households. Maps 2 and 3 demonstrate the headed households, and perhaps less access. Hence, the level of extension services reception across communes equilibrium would misleadingly appear to be the same for for each type of household according to the gender male-headed and female-headed households. 12 Determinants of Agricultural Extension Services: The Case of HaitI Map 2: Agricultural extension services at the commune level – Male-headed households Source: Authors. Map 3: Agricultural extension services at the commune level – Female-headed households Source: Authors. Determinants of Agricultural Extension Services: The Case of HaitI 13 According to our econometric model, gender itself is not services. Sometimes, in this context, a female-headed important in explaining supply and demand equilibrium household may receive a lower amount, as we previously levels of extension services. The fact that gender is not observed in Maps 2 and 3, in locations where overall ac- significant when controlling for these covariates and cess is low. If we assume that the aforementioned house- location means that the initial rather small differences holds were being offered the same amount of extension in extension reception observed in Graph 2 were not the services, we may conclude that no further interventions result of underlying differences in education, agricultural are necessary to correct the tendency to favor men, training, farm size, type of crop produced, and location when in reality, discrimination may be latent—factors between male-headed and female-headed households. such as the time of the day services are offered, night We also ran two separate regressions (results not shown): travel, and long distances, among others, have been one only for farmers located at Cerca La Source (a loca- documented in Haiti as issues that prevent women from tion with a high level of extension) in the Center depart- accessing services. ment and the other for those farmers located at St. Louis Du Sud (a location with a low level of extension) in the Education South department. In the case of Cerca La Source, the coefficient on the female dummy is positive and signifi- Haiti faces challenges of both supply and demand cant at the 0.1 level. In that commune, a female-headed in the education marketplace. These challenges are household has an 11.48 percent greater chance of receiv- compounded in rural areas by high poverty and difficult ing extension services than a male-headed household access. On the supply side, there are simply not enough controlling for education, agricultural training, farm size, spaces for children to enroll in school. It is estimated and type of crop. In St. Louis Du Sud, the coefficient that 400,000 to 500,000 children aged 6 to12, the major- on the female dummy is not significant. Therefore, the ity of whom live in rural areas, are not attending school. relationship between the gender of the head of house- On the demand side, the average cost of US$70 tuition per hold and the receipt of extension services, if any, may child per year is prohibitive for poor families, especially for favor women. In those places with a high overall availabil- those living in rural areas characterized by poverty rates ity of extension services, women receive systematically of 82 percent (77 percent living in extreme poverty).1 Even more extension services than men. In those places with when schools are accessible, the quality of the educa- an overall low availability of extension services, there are tion offered is uneven, and often very low. This is demon- no significant differences between men and women after strated by the findings of the recent Early Grade Reading controlling for other covariates in the model. Assessment (EGRA), carried out in 2008 and 2009 in Haiti. On average, children in Grade 3 are able to read fewer Recall that we are observing the equilibrium of demand than 23 words per minute.2 For those students studying for and supply of extension services, which means that in Creole, 29 percent were unable to read a single word even when female-headed and male-headed house- by Grade 3. Reading comprehension is even weaker, with holds receive the same level of extension services (pro- children able to answer less than 10 percent and 17 per- vided they have the same education level, agricultural cent of reading comprehension questions correctly, training, farm size, and produce the same type of crop), in French and Creole respectively.3 the interaction between supply and demand by which they receive the same services can be different. For Opportunities to improve small farmers’ competitiveness instance, extension services in a particular commune are reduced as extremely poor levels of education ham- may be provided primarily to male-headed households, per the implementation of new productivity-enhancing yet the demand from female-headed households could agricultural technologies. According to the Agricultural be significantly higher than that from male-headed ones, Census (see Table 3.3), 57.09 percent of heads of house- resulting in the receipt of the same number of extension hold are illiterate. If we further discriminate by gender, the 1  The World Bank. Education for All Project – Phase II (APL). October, 2011. 2  Sixty words per minute is standard for early primary reading fluency. 3  Research Triangle Institute. Haiti Early Grade Reading Assessment (EGRA): Rapport pour le MENFP et la Banque Mondiale. Avril 2010. 14 Determinants of Agricultural Extension Services: The Case of HaitI Table 3.3: Education by gender of head of household Male-Female Ratios Total Male Female Male Female Education N % N % N % % % None 172,011 57.09 118,758 53.86 53,253 65.88 69.04 30.96 Literate 60,865 20.20 47,137 21.38 13,728 16.98 77.45 22.55 Elementary 46,386 15.39 36,490 16.55 9,896 12.24 78.67 21.33 High School 18,866 6.26 15,366 6.97 3,500 4.33 81.45 18.55 Professional 2,305 0.76 1,960 0.89 345 0.43 85.03 14.97 University 875 0.29 765 0.35 110 0.14 87.43 12.57 Total 301,308 100.00 220,476 100.00 80,832 100.00 73.17 26.83 Source: Agricultural Census 2008–2010. Authors’ calculations. level of illiteracy in female heads of household reach- an illiterate to a literate farmer seems to have a positive es 65.88 percent. In Table 3.3, we clearly observe how the effect on receiving extension services for all departments, proportion of male-headed households increases as the presumably, as a result of required reading material. How- level of education increases, indicating that women are ever, even if the ability to read is not necessary to receive less favored than men in terms of education. For example, extension services, literate people are more likely not the proportion of female-headed households in the only to be aware of the benefits of receiving agricultural three departments analyzed is 26.83 percent; however, extension services, but also to understand the procedures among those heads of households with university-level for receiving extension services and how to implement education, the proportion of female-headed households what they learn or what they receive as inputs for their is only 12.57 percent. farms. Berger et al. (1984) points out that “education enhances the ability of farmers to acquire accurate Graph 3 provides useful insights that may clarify the information, evaluate new production processes, and use mechanisms through which receipt of extension ser- new agricultural inputs and practices efficiently. Better vices is influenced by education. Moving from being educated farmers are twice as likely to be in contact with Graph 3: Households receiving extension services by education level 30 South East Center 25 South Percentage (%) 20 15 10 5 0 None Literacy Elementary High School Professional University Education Level Source: Agricultural Census 2008–2010. Authors’ calculations. Determinants of Agricultural Extension Services: The Case of HaitI 15 agricultural extension agents, indicating that farmers with and the “knowledge effect.� On the other hand, there higher levels of education benefit most from extension is one clear education-based force affecting the supply services.� In addition, “educated farmers may push the of extension: the eagerness of extension agents to pro- extension system to deliver what they need and make vide services to more educated farmers. The dynamics sure the knowledge is appropriate to their resources.� of these forces may explain the different levels of exten- sion services received depending on a farmer’s level Nevertheless, elementary schooling appears to have of education. For instance, at first glance, it might seem a negative effect on extension reception relative to mere strange that the positive effect of education fades be- literacy. It is important to note that as people become yond mere literacy and then returns after university-level more educated, they acquire skills that can be better re- education. However, if we assume both that extension warded in non-farm activities. Hence, the more educated agents tend to favor educated farmers and that demand a person is beyond literacy, the lower their demand for for extension services is lower for higher levels of educa- agricultural extension services may be. However, if wage tion, this result can be reasonable. It can also be argued jobs are scarce, or the opportunity costs related to leav- that the demand for extension services can even in- ing their farms are fairly high, then we would presumably crease at high levels of education as farm owners might see an increasing relationship between education and hire farm workers that receive extension services. the receipt of extension services, as seems to be the case in the South East department. These forces can also explain the apparent heterogeneity that we observe across departments. For example, in the When controlling for other factors, the positive effect of South department, education has a more consistently posi- being literate is smaller than that observed in Graph 3. In tive effect overall on receipt of extension services com- particular, literacy increases the likelihood of receiving pared to in other departments. Presumably, in the South extension services by only 3.44 percent. This trend is mainly department, the opportunity costs of leaving agriculture driven by the South department, where literacy increases as a main activity are higher than in other departments. this likelihood by 7.43 percent. In the other departments, Furthermore, as we observe in Table A.4 (see Annexes), the the effect is not even statistically significant. Moreover, in South department has more farmers reporting livestock the Center department, having professional education and fisheries as their main economic activities. These activi- decreases the likelihood of receiving extension services, ties may be more difficult to leave behind, which means whereas, in the South, it increases the likelihood by 10.23 that they may be more profitable than agriculture. percent, and university-level education increases the like- lihood of receiving extension services by 2.21 percent. Agricultural Training On the supply side, there is the possibility that for low As demonstrated by Graph 4, there seems to be an in- levels of education, access to extension services is still verted u-shaped relationship between agricultural training extremely low as extension agents may prefer to provide and the receipt of extension services. Even after controlling extension services to more educated farmers where the for other covariates, the results confirm the concavity and possibility of implementing newly acquired knowledge show that having “occasional agricultural training� (OAT) is higher. At the same time, on the demand side, farmers increases the likelihood of receiving extension services with more education are less prone to demand extension by 23.98 percent compared to having just empirical train- services as they are able to learn and apply new tech- ing. Furthermore, having technical agricultural training nologies or knowledge by themselves, what we could increases the likelihood by 25.12 percent, which means call the “knowledge effect.� Furthermore, the possibility that the positive effect of agricultural training is decreas- of looking for non-farm jobs is higher for those with bet- ing. Apart from the “awareness effect� and the “knowledge ter education. Therefore, on the one hand, there may effect,� which were also discussed in the case of education be three education-based forces affecting the demand (and which may be even more pronounced in this case), for extension: the “awareness effect,� the possibility receiving OAT from specialized agencies, such as a DDA of finding a non-farm wage job that is economically more or NGOs, may create an enabling environment for farmers, convenient than the farmer’s agriculture-related activity, putting forward adequate channel factors for both farmers 16 Determinants of Agricultural Extension Services: The Case of HaitI demanding extension services and extension providers informed and knowledgeable about agricultural topics— supplying the services. Furthermore, extension agents may and may even be more knowledgeable than extension naturally target farmers with high agricultural training since facilitators themselves. adoption of new technologies and knowledge received is more likely and thus their work can be properly measured One public sector supply of OAT is the Ecoles Moyennes and rewarded. It is also important to note that the positive Agricoles (EMAs) for Vocational and Farmer Field Educa- effect of having OAT in terms of the receipt of extension tion on a nationwide scale. Having the proper channels services diminishes significantly when the head of house- through which extension services are delivered not only hold is a woman (see Table A.6 in Annexes) since women increases the supply of extension, but also stimulates the may benefit less from the opportunities brought about demand for these services. The EMAs are well-known by the channel factors mentioned above. Other possible agricultural training institutions supported by the World explanations are that the “knowledge effect� may be more Bank, Canada, USAID/USDA, and other development pronounced in the case of female-headed households, organizations working in Haiti. The MARNDR is seeking or the supply of extension services to female-headed to leverage and strengthen the EMAs as part of the na- households may be low even when they have high-level tional strategic plan (PDVA) to expand extension services agricultural training. in Haiti. The positive impact of agricultural training on the uptake Some might reasonably argue that occasional agricul- of extension services starts to sink in at the technical level. tural training is so statistically significant in explaining ag- It is possible that within agricultural training, the “knowl- ricultural extension services because OAT and extension edge effect� discussed previously is dominating the services are being perceived by the farmers interviewed dynamics of receiving extension services. In other words, as being the same thing. However, if this is true, then the people with technical agricultural training might per- correlation between receiving extension services and ceive the benefits of receiving extension services as mini- having occasional training should be nearly one. In order mal, or even non-existent. For example, the FAO found to assess the possibility that OAT and extension services that 40 percent of extension personnel used in developing might be perceived as being the same, we present countries had only secondary school education (Feder Table 3.4, which shows the relationship between OAT and et al., 1999). Hence, not surprisingly, uptake of extension receipt of extension services, based on data pooled from services is significantly diminished as people get more the three departments. Graph 4: Proportion of households receiving extension services by agricultural training 60 South East Center 50 South Percentage (%) 40 30 20 10 0 Empirical Occasional Technical University Agricultural Training Source: Agricultural Census 2008−2010. Authors’ calculations. Determinants of Agricultural Extension Services: The Case of HaitI 17 Table 3.4: Occasional Agricultural Training (OAT) and extension services Extension Received Not Received Total OAT N % N % N % Received 2,819 0.93 4,562 1.51 7,381 2.44 Not received 39,160 12.97 255,461 84.59 294,621 97.56 Total 41,979 13.9 260,023 86.1 302,002 100 Source: Authors. According to Table 3.4, among those who did not re- for farmers with access to other alternatives for acquiring ceived OAT, the ratio between those who received exten- knowledge (such as fee-based extension). sion services and those who did not is 0.15 (=12.97/84.59), yet within those who did receive OAT, the ratio is 0.62. However, again, we are observing the equilibrium between Therefore, there is a positive correlation between OAT and the supply of and demand for extension services. These extension services, however the correlation is rather low preliminary results may be explained not only by issues (0.11). Moreover, there is a significant proportion of the of the marginal benefit of implementing extension advice, population who did not receive OAT and who did receive but also by issues related to the marginal propensity to offer extension services. Hence, we cannot conclude that OAT extension advice. In other words, it is possible that the sup- and extension services are exactly overlapping events. ply of extension services is more targeted to smaller farms. Nevertheless, the literature suggests that the opposite Farm Size is true. Feder et al. (1999) stresses that there is a tendency of extension agents to favor more responsive clients, who Graph 5 plots the relationship between receiving exten- are typically better endowed and more capable of under- sion services and farm size. Both empirical and theoreti- taking risks. Consequently, this reinforces the possibility that cal studies suggest that farmers with larger farms adopt the concavity of the relationship between farm size and extension services more quickly (Fischer, 1985); thus receiving extension services is better explained by a low de- we would expect a greater use of extension services mand for extension services from farmers with larger farms. in larger farms. As we can see in Graph 5, the relationship between farm size and receipt of extension services is in- The concave relationship described above between deed positive, yet the relationship is concave, meaning receiving extension services and farm size is somewhat that the rate of receipt of extension services decreases supported by the results of the regression; yet if we dis- as farm size increases. Moreover, for the South and South criminate by department, we observe that only in the East departments, these curves correlate very well with South is the relationship significant. Having a farm of be- those of the previous graph. It seems that the marginal tween 0.3 and 0.6 hectares increases the likelihood benefits of implementing extension services might be con- of receiving extension services by 4.57 percent compared stantly reducing as farm size increases, ultimately affect- to having a farm of less than 0.15 hectares; however, hav- ing the demand for extension services. Feder (1999) shows ing a farm of between 0.6 and 1.2 hectares decreases the that the effectiveness of extension investment is highly probability of receiving extension services by 0.44 percent contingent on relaxing wider barriers to the successful in relation to the previous size bracket. In the South, the development of the agricultural sector as a whole, includ- concavity is even more pronounced. Therefore, larger ing such potentially limiting factors as credit, technology farms either received proportionally (to size) fewer exten- stock, input supplies, price incentives, institutions, and hu- sion services or received fewer extension services in abso- man resource constraints. Therefore, it may be reasonable lute terms. Taking into account the tendency of extension to argue that extension services in Haiti are not highly ef- agents to favor more responsive clients, who are typically fective, and so the demand for these services is rather low better endowed and more capable of undertaking risks 18 Determinants of Agricultural Extension Services: The Case of HaitI Graph 5: Households receiving extension services by farm size 18 South East Center 16 South Percentage (%) 14 12 10 8 6 4 2 0 Less than 0.15 to 0.3 to 0.6 to 1.2 to More than 0.15 0.3 0.6 1.2 2.4 2.4 Hectares Source: Agricultural Census 2008−2010. Authors’ calculations. (Feder et al., 1999), this result may be driven by low de- extension has been practiced across the public, parastat- mand rather than by a lack of adequate supply. al, private, and social sectors, including agroprocessing and marketing firms and farmers’ associations. The focus In summary, up to a certain farm size, the receipt of ex- is often on one commercial or export crop (i.e., cash tension services increases as farm size increases, possibly crops) linked to established marketing or processing out- because of a greater supply for larger farmers, but also lets (Feder et al., 1999). However, according to Graph 6, because of economies of scale, making the implemen- a larger proportion of maize producers seem to receive tation of new technologies more feasible, which in turn extension services compared to other producers, which increases the demand for extension services. However, contradicts the aforementioned notion of the prefer- beyond that point, it is likely that demand for extension ence for cash crop farmers. For instance, in Graph 6, the services decreases as farmers with larger farms have more number 4.48 in the horizontal bar with upward diagonals leverage to acquire new knowledge from more efficient in the South East department indicates that the propor- sources (such as fee-based extension). tion of maize producers receiving agricultural extension services is larger than that of all non-maize producers It is also important to note that for large farms, the effect by 4.48 percentage points. In the Center department, the on the likelihood of receiving extension services is not advantage for maize producers is even larger. significant; however, for female-headed households it is significant and positive. Assuming that the supply However, after controlling for the covariates described of extension services is not higher for female-headed in the model (which includes other crops), being a maize households, a feasible explanation for this result may farmer is not statistically significant. On the contrary, in the be that women are generally more risk averse (see for Center department, being a maize farmer decreases the example Eckel and Grossman (2008)) and prefer not to in- likelihood of receiving extension services by 1.73 percent. vest in more expensive—though more efficient—services, These results suggest that being a maize farmer is corre- and so rely on free extension services although they may lated with at least one of the covariates in the economet- have the resources to acquire fee-based extension. ric specification. In the model, being a maize farmer does not mean that the farmer does not grow any other crops, Type of Crop but indicates that the farmer grows maize regardless of any other crops that he or she may work with. In other Finally, we also assess if receipt of extension services is in- words, we acknowledge the practice of multi-cropping fluenced by the type of crop grown. Commodity-specific by considering a dummy for each crop. Naturally, the fact Determinants of Agricultural Extension Services: The Case of HaitI 19 Graph 6: Households receiving extension services by type of crop Maiz –4.34 Mangoes South 4.39 Coffee 1.93 Beans –1.04 Bananas 8.31 10.18 Center –9.26 0.84 5 4.48 0.04 South East –4.59 –2.36 –5.32 Source: Agricultural Census 2008–2010. Authors’ calculations. Numbers can be negative as they reflect the difference between the proportion of households growing that crop and those not growing it. that the farmer is growing other crops (e.g., coffee) can favored in terms of the amount of extension services re- affect both being a maize farmer and the likelihood of re- ceived. Coffee has been a leading cash crop in Haiti for ceiving extension services. However, we are reducing this many years, accounting for a sizeable proportion of crop bias, when controlling for other crops, such as bananas, exports for the country. beans, coffee, and mangoes, which are the most popular crops in terms of the number of farmers growing them. For future research, it will be important to investigate both whether (i) coffee producers are better organized (at In addition, contrary to our observations in Graph 6, being least in the South East department) than other crop farm- a coffee producer increases the likelihood of receiving ers, such as banana farmers, who apparently systemati- extension services by 5.15 percent. This trend is mainly driv- cally received fewer extension services; and (ii) whether en by the high numbers of coffee producers in the South there are explicit commodity-specific extension services East department receiving extension services, where being provided for coffee producers. Additionally, it is im- the likelihood of receiving extension services increases portant to take into account that farmers with certain by 9.33 percent for coffee producers. These results make types of crops may demand more extension services than much more sense when placed in context by the relevant others as their pre-harvest and/or post-harvest processes literature review, which demonstrates that cash crops are are more complex and require more expertise. 20 Determinants of Agricultural Extension Services: The Case of HaitI IV. Conclusions and Recommendations The coefficients obtained from the regression results are whether the variation in the receipt of extension services not causal as there may be other unobserved farmer- is caused mainly by either changes in the demand specific variables affecting both the likelihood of receiv- or in the supply. It would also allow for a refined under- ing extension services and one or more of the covariates standing of the suggested mechanisms through which analyzed in the model, thereby causing omitted variable gender, education, agricultural training, farm size, and bias. Instead, the results of the fixed effects probit model type of crop affect the demand for and supply of exten- allow us to establish conditional correlations between the sion. likelihood of receiving agricultural extension services and each of the covariates. These correlations are only valid Some key conclusions and recommenda- in the case of farmers located in the South East, Center, tions that arise from this analysis in- and South departments. Based on this model, we can clude the following: predict the likelihood of receiving extension services conditioned on arbitrarily chosen values of the covari- 1. The proportion of households receiving agricultural ates under analysis. In other words, we can calculate the extension services in Haiti is non-negligible. Indeed, likelihood of receiving extension services for a household there are places in Haiti where, by regional standards, with a specific profile based on location, gender, educa- a large proportion of households receive agricultural tion, and agricultural training of the head of household, extension services. Although public sector extension farm size, and type of crop being produced. Furthermore, services have virtually disappeared in recent decades, based on a set of profiles (those who received fewer the relatively widespread availability of extension services extension services), a development project can use shows that donor funded projects, the private sector, the results of this paper to better target a specific group and NGOs are providing a significant level of agricultural of marginalized farmers so that the effects of the project services. This highlights the importance of mainstreaming will be maximized. This in turn has the potential to increase and integrating current agricultural extension services into the power of statistical tests performed to evaluate the the national level agriculture system led by the MARNDR, project’s impact, making impact evaluation feasible facilitating coordination and funding, to avoid duplica- or even reducing the costs of evaluation because of a re- tion and allowing for clear priorities and comprehensive duced sample size. engagement. In this study we only observe extension allocation result- 2. Location is an important determinant of the re- ing from the market equilibrium between the demand for cipients of agricultural extension services. There are and supply of extension services. In future studies, it would pockets of both low and high receipt of extension services be useful to assess the demand for and supply of ex- at the commune level. However, the differences in exten- tension services separately to more clearly understand sion reception at the commune level may be reflecting Determinants of Agricultural Extension Services: The Case of HaitI 21 other variables not addressed by the Agricultural Cen- this positive effect fades at higher education levels and sus, such as distance to the nearest DDA, irrigation, then returns with university-level education. On the one geography (topography and communications), and hand, extension agents have incentives that favor edu- political configurations, among others. In particular, the cated farmers, as the positive effects of extension would decentralization process in Haiti remains a major chal- be more pronounced on more educated subjects. On the lenge, although the MARNDR is the Ministry with the other hand, we identify three forces influencing demand: strongest presence in rural areas. As already discussed, the “awareness effect,� the “knowledge effect,� and the the current centralized scheme may favor those com- possibility of getting a non-farm wage job. The influence munes located near the capital or those that are easily of these three factors and the supply of extension ser- reached by a DDA/BAC. This calls for particular attention vices will ultimately determine the allocation of extension to be paid by DDAs and BACs in the provision of exten- services according to a specific education level. Commu- sion services in order to provide and coordinate extension nication campaigns could exploit the “awareness effect� support (public and private) that not only reaches all to provide information about the benefits of extension farmers, but that is also adapted to local conditions and services to farmers with lower levels of education, which demands. would also allow for an increase in demand (and thus supply). Indeed, demand-driven agricultural extension 3. There are no statistical differences between men services can be an effective way of allocating such ser- and women in terms of receipt of extension services; vices if farmers are aware of the benefits beforehand. however, the impact of agricultural training and farm size change when the head of household is a woman. 5. Prior agricultural training is a major determinant Specifically, being a female-headed household dimin- of the recipients of extension services. From the ishes the positive effect of having occasional agricultural perspective of demand, apart from the fact that the training (OAT) and amplifies the positive effect of having “awareness effect� and “knowledge effect� are even larger farms on the likelihood of receiving agricultural more pronounced than for the case of education, extension services. In addition, women seem to receive receiving training in agriculture may create an enabling more extension services than men where the overall sup- environment for farmers who need extension services. ply of extension services is high. This indicates that given For instance, on some occasions, receipt of extension the opportunity to have access to agricultural exten- services may be just a matter of knowing the person sion services, women avail themselves of these services responsible for providing the services or being familiar more than men do, which suggests that the apparent with the administrative processes for receiving exten- equivalence between men and women may be better sion services. In other words, having agricultural training interpreted as men having either equal or less demand increases awareness not only of the benefits of extension, for extension services than women rather than equal but also of the people, procedures, and mechanisms access. Therefore, agricultural extension services need through which extension services are provided, which to ensure that women are not excluded, as it has been in turn increases the demand for these services as farmers proven that if given the opportunity, women will make have a clearer picture of how to acquire them. From the use of such services. The time of the day, the need for supply side, extension agents may be inclined to favor night travel, long distances, and other factors have been those whom they know and are more likely to effectively documented in Haiti as issues that can prevent women implement the knowledge provided. from accessing services. Therefore, details on when, how, and where extension services are provided are 6. Rehabilitation of the Ecoles Moyennes Agricoles key to including (or excluding) women, and thus need (EMAs) for vocational and farmer field education to be carefully thought out in order to offer women a fair on a nationwide scale would increase the demand opportunity to participate. for extension services, especially among small farm- ers. Our results indicate that OAT significantly increases 4. Education level has a positive yet small effect the likelihood of receiving agricultural extension services on receiving extension services. Being literate increases as it opens channel factors through which farmers can the likelihood of receiving extension services; however, develop a better understanding of the basic steps toward 22 Determinants of Agricultural Extension Services: The Case of HaitI receiving extension services and make contact with key Nevertheless, production of other cash crops, such as ba- players (e.g., extension agents). In other words, having the nanas, has a negative impact on the likelihood of receiv- proper channels through which extension services are de- ing extension, which suggests that the criteria for favoring livered not only increases the supply of extension services, one crop over another goes beyond its categorization but also stimulates the demand for these services. EMAs as a cash crop. Presumably, the level of coordination are well-known agricultural training institutions supported among producers, the presence of cooperatives, mana- by the World Bank, Canada, USAID/USDA, and other gerial sophistication, and/or the complexity of relevant development organizations working in Haiti. The MARNDR processes may play a key role in determining both the is seeking to leverage and strengthen the EMAs as part demand and supply of extension services for different of the national strategic plan (PDVA) to expand extension crop producers. services in Haiti. 9. Promoting a hybrid system of extension may 7. Farmers with larger farms receive more agricultural be more efficient than supporting only public extension services. The relationship between farm size or NGO-provided extension services. In recent years, and access to extension is positive and concave, mean- the improvement of agricultural extension services has ing that farmers with larger farms receive more extension, been the focus of attention of recent agriculture policies yet the rate of extension reception diminishes as farm size and programs in Haiti, and the World Bank, IADB, and increases. The supply of extension services may be great- USAID have been increasing investments in this area. er for larger farms, since both economies of scale (more However, as we observed, the demand for extension with less) and the likely reduction of transaction costs mo- services decreases beyond certain thresholds of farm tivate extension agents to favor large-scale farmers. How- size and agricultural training. The mechanisms by which ever, the demand for extension services may be reduced this may occur point to the fact that some farmers (those as farm size increases, given that wealthy farmers can with larger farms and greater knowledge of agricultural afford both more expensive and more efficient alterna- techniques) may have a latent demand for fee-based tives to learning innovative and productive technologies extension services. Therefore, a more efficient alterna- to be applied on their farms (e.g., fee-based extension), tive may be to offer targeted public and NGO-provided provided the marginal benefits of the currently free exten- extension services only to those farmers who cannot sion services are low. access fee-based extension services (e.g., small farm- ers), taking into account the specific services different This conclusion is key for future agricultural extension farmers require. programs in that there should be no discrimination based on farm size, in particular against the smallest plots. Al- In addition, instead of using fiscal resources to provide ex- though over 90 percent of farms in Haiti are under 5 hect- tension services to farmers who may not even need them, ares, agricultural extension should adapt to the demand the government could channel these resources toward from different segments, tailoring support to different farm creating an environment in which private investment for sizes, types of services required, and other logistical and extension services is feasible, fostering the development demographic considerations. of a parallel fee-based extension market. Yet another al- ternative would be to subsidize private extension services, 8. Coffee producers make more use of extension crowding-in private companies to the extension market services than other farmers. Being a coffee producer until the demand is substantial enough to fully priva- increases the likelihood of receiving extension services tize extension. A combination of these measures would by 5.15 percent. However, this effect is largely driven serve to promote the demand, equity, and effectiveness by the South East department, where coffee producers of agricultural extension services, yielding greater ben- are 9.33 percent more likely to receive extension services. efits in terms of increased demand for extension services It appears that a commodity-specific type of extension as well as increased and more equitable farmer partici- mechanism is operating in that region, favoring crops pation—particularly female—regardless of location, crop that are mainly oriented toward export (i.e., cash crops). type, or farm size. Determinants of Agricultural Extension Services: The Case of HaitI 23 References Alex, G. et al. 2000. Decentralizing Agricultural Extension: Lessons and Good Practice. The World Bank, Washington, DC. Berger, M. et al. 1984. Bridging the Gender Gap in Agricultural Extension. International Center for Research on Women. Washington, DC. Christoplos, I. 2010. Mobilizing the potential of rural and agricultural extension. Food and Agriculture Organization of the United Nations. Rome, Italy. Christoplos, I. et al. 2012. Guide to Evaluating Rural Extension. Global Forum for Rural Advisory Services (GFRAS). Switzerland. Feder, G. et al. 1999. Agricultural Extension: Generic Challenges and Some Ingredients for Solutions. The World Bank. Washington, DC. Fischer, A. 1985. On the Provision of Extension Services in Third World Agriculture. The World Bank. Washington, DC. Lastarria-Cornhiel, S. 2006. Feminization of Agriculture: Trends and Driving Forces. Background paper for the WRD 2008. The World Bank. Washington, DC. McMahon, M., and Valdés, A. 2011. Análisis del Extensionismo Agrícola en México. OECD. Paris, Francia. Rivera, W. and Alex, G. 2004. Volume 3. Demand-Driven Approaches to Agriculture Extension: Case Studies of Interna- tional Initiatives. The World Bank. Washington, DC. Rivera, W. and Alex, G. 2004. Volume 5. National Strategy and Reform Process: Case Studies of International Initiatives. The World Bank. Washington, DC. Saito, K. and Spurling, D. 1992. Developing Agricultural Extension for Women Farmers. The World Bank. Washington, DC. Swanson, B. 2008. Global Review of Good Agricultural Extension and Advisory Service Practices. Food and Agriculture Organization of the United Nations. Rome, Italy. Swanson, B. and Rajalahti, R. 2010. Strengthening Agricultural Extension and Advisory Systems: Procedures for Assessing, Transforming, and Evaluating Extension Systems. The World Bank. Washington, DC. Umali, D. and Schwartz, L. 1994. Public and Private Agricultural Extension: Beyond Traditional Frontiers. The World Bank. Washington, DC. 24 Determinants of Agricultural Extension Services: The Case of HaitI Annexes Table A.1: Farm size by gender of household head – South East Male-Female Ratios Male Female Total Male Female Hectares N % N % N % % % Less than 0.15 46,323 25.73 29,937 24.09 16,386 29.38 64.63 35.37 0.15 to 0.3 43,874 24.37 30,026 24.16 13,848 24.83 68.44 31.56 0.3 to 0.6 51,052 28.36 35,721 28.74 15,331 27.49 69.97 30.03 0.6 to 1.2 28,490 15.82 20,652 16.62 7,838 14.05 72.49 27.51 1.2 to 2.4 8,222 4.57 6,287 5.06 1,935 3.47 76.47 23.53 More than 2.4 2,082 1.16 1,651 1.33 431 0.77 79.30 20.70 Total 180,043 100 124,274 100 55,769 100 69.02 30.98 Source: Authors. Table A.2: Farm size by gender of household head – Center Male-Female Ratios Male Female Total Male Female Hectares N % N % N % % % Less than 0.15 3,878 2.30 2,769 2.10 1109 2.98 71.40 28.60 0.15 to 0.3 10,139 6.00 7,457 5.66 2,682 7.21 73.55 26.45 0.3 to 0.6 39,438 23.35 29,878 22.68 9,560 25.70 75.76 24.24 0.6 to 1.2 60,395 35.75 47,822 36.30 12,573 33.80 79.18 20.82 1.2 to 2.4 41,991 24.86 33,696 25.58 8,295 22.30 80.25 19.75 More than 2.4 13,084 7.75 10,104 7.67 2,980 8.01 77.22 22.78 Total 168,925 100 131,726 100 37,199 100 77.98 22.02 Source: Authors. Table A.3: Farm size by gender of household head – South Male-Female Ratios Male Female Total Male Female Hectares N % N % N % % % Less than 0.15 20,017 12.54 15,050 12.04 4,967 14.37 75.19 24.81 0.15 to 0.3 34,252 21.46 26,396 21.11 7,856 22.73 77.06 22.94 0.3 to 0.6 53,961 33.81 41,735 33.38 12,226 35.37 77.34 22.66 0.6 to 1.2 36,214 22.69 29,484 23.58 6,730 19.47 81.42 18.58 1.2 to 2.4 11,235 7.04 9,484 7.58 1,751 5.07 84.41 15.59 More than 2.4 3,925 2.46 2,891 2.31 1,034 2.99 73.66 26.34 Total 159,604 100 125,040 100 34,564 100 78.34 21.66 Source: Authors. Determinants of Agricultural Extension Services: The Case of HaitI 25 Table A.4: Principal production activities by gender of household head South East Center South Activity N % N % N % Agriculture Male 49,252 86.33 90,641 95.06 56,408 82.60 Female 22,737 77.62 25,415 87.80 14,348 63.21 Total 72,021 83.35 116,218 93.37 70,840 77.74 Fisheries Male 1,135 1.99 410 0.43 1,983 2.90 Female 204 0.70 118 0.41 186 0.82 Total 1,340 1.55 529 0.42 2,173 2.38 Livestock Male 823 1.44 1,158 1.21 2,590 3.79 Female 242 0.70 587 2.03 1,017 4.48 Total 1,073 1.24 1,747 1.40 3,610 3.96 Buildings and public works Male 489 0.86 443 0.46 1,117 1.64 Female 39 0.13 22 0.08 55 0.24 Total 528 0.61 465 0.37 1,174 1.29 Administration Male 360 0.63 354 0.37 762 1.12 Female 46 0.16 69 0.24 190 0.84 Total 411 0.48 423 0.34 971 1.07 Mines and quarries Male 35 0.06 155 0.16 303 0.44 Female 2 0.01 3 0.01 121 0.53 Total 37 0.04 158 0.13 426 0.47 Transformation Male 82 0.14 34 0.04 83 0.12 Female 38 0.13 7 0.02 83 0.37 Total 122 0.14 41 0.03 166 0.18 Commerce Male 2,096 3.67 1,406 1.47 2,180 3.19 Female 4,891 16.70 2,612 9.02 5,782 25.47 Total 6,991 8.09 4,024 3.23 7,971 8.75 Services Male 2,019 3.54 624 0.65 2,474 3.62 Female 529 1.81 90 0.31 770 3.39 Total 2,564 2.97 715 0.57 3,251 3.57 Handicraft Male 758 1.33 129 0.14 259 0.38 Female 563 1.92 22 0.08 120 0.53 Total 1,322 1.53 151 0.12 379 0.42 Source: Authors. 26 Determinants of Agricultural Extension Services: The Case of HaitI Table A.5: Area used for principal crops by gender of household head Total Male Female Major crop Ha HHs Avg. Size Ha HHs Avg. Size Ha HHs Avg. Size Maize 124,051.6 201,814 0.61 96,888.52 147,989 0.65 27,163.06 53,601 0.51 Beans 58,644.76 102,095 0.57 44,489.71 72,652 0.61 14,155.06 29,333 0.48 Bananas 23,331.87 72,307 0.32 17,792.01 52,750 0.34 5,539.86 19,467 0.28 Mangoes 304.13 3,161 0.10 262.53 2,309 0.11 41.6 849 0.05 Coconuts 311.16 3,785 0.08 246.02 2,681 0.09 65.13 1,102 0.06 Coffee 7,883.74 23,909 0.33 5,777.54 16,715 0.35 2,106.19 7,181 0.29 Source: Authors. Table A.6: Regression results – with interactions Determinants of usage of agricultural extension services Dependent variable: use of extension services = 1 if use at least one extension service, 0 otherwise Regressor All South East Center South Female –0.081 –0.2249* 0.2635* –0.0205 (0.0704) (0.1269) (0.1385) (0.0705) Education None omitted omitted omitted omitted — — — — Literate 0.204*** 0.0582 0.1764 0.3684*** (0.075) (0.0539) (0.1308) (0.1178) Elementary 0.0119 –0.1046 –0.0084 0.1322 (0.0796) 0.1437 (0.1654) (0.1204) High School 0.0755 0.0303 0.084 0.1357 (0.063) (0.1049) (0.1261) (0.1101) Professional 0.1221 0.0153 –0.5215*** 0.4422*** (0.1183) (0.2311) (–0.1401) (0.116) University 0.1282* 0.2228* –0.0332 0.2007** (0.0706) (0.1326) (0.2707) (0.089) Agricultural Training Empirical omitted omitted omitted omitted — — — — Occasional 0.9234*** 0.662*** 0.9713*** 1.0301*** (0.1091) (0.1882) (0.0957) (0.1785) Technical 0.9276*** 1.2237*** 0.5524*** 0.9493*** (0.1445) (0.3013) (0.0813) (0.1988) University 0.1003 0.4331* 0.0719 –0.016 (0.1565) (0.2598) (0.2146) (0.181) Farm size (hectares) Less than 0.15 omitted omitted omitted omitted — — — — (continued on next page) Determinants of Agricultural Extension Services: The Case of HaitI 27 Table A.6: Regression results – with interactions (continued) Determinants of usage of agricultural extension services Dependent variable: use of extension services = 1 if use at least one extension service, 0 otherwise Regressor All South East Center South From 0.15 to 0.3 0.1022 –0.0293 0.185 0.2012*** (0.0839) (0.1244) (0.1751) (0.0716) From 0.3 to 0.6 0.224*** 0.0877 0.5032** 0.3151*** (0.0872) (0.1138) (0.2294) (0.0996) From 0.6 to 1.2 0.2105** –0.0168 0.6133** 0.2906*** (0.1053) (0.1456) (0.2755) (0.1036) From 1.2 to 2.4 0.1066 –0.0364 0.4209 0.251** (0.1208) (0.1741) (0.3441) (0.0996) More than 2.4 0.0643 –0.1091 0.3741 0.2903* (0.1185) (0.185) (0.3196) (0.1654) Crops Maize 0.162 –0.0135 –0.1175 0.2711 (0.1156) (0.1436) (0.0788) (0.19) Beans –0.042 0.0521 –0.1801 0.056 (0.0796) (0.1148) (0.2977) (0.1702) Bananas 0.0178 –0.0608 0.2198* –0.2016** (0.0796) (0.1377) (0.1197) (0.0804) Coffee 0.2652** 0.4298*** 0.056 –0.0409 (0.1145) (0.1614) (0.1161) (0.1294) Mangoes –0.0435 –0.1938* 0.0488 0.0677 (0.0793) (0.1169) (0.1545) (0.1314) Interactions Female with Education Female*None omitted omitted omitted omitted . . . Female*Literate –0.0471 0.0197 –0.1455** 0.0009 (0.0459) (0.0998) (0.0573) (0.0613) Female*Elementary 0.0327 0.0219 –0.031 0.0853* (0.0403) (0.0837) (0.054) (0.0504) Female*High_school –0.0016 –0.0304 –0.0674 0.0847 (0.0585) (0.0868) (0.1222) (0.0844) Female*Professional –0.0811 0.0483 –0.287 0.0162 (0.1287) (0.2884) (0.3616) (0.1111) Female*University –0.0728 –0.4293 0.3177 –0.0541 (0.2147) (0.306) (0.5322) (0.2964) Female with Agricultural training Female*Empirical omitted omitted omitted omitted — — — — Female*Occasional –0.1052* 0.008 –0.1502* –0.133 (continued on next page) 28 Determinants of Agricultural Extension Services: The Case of HaitI Table A.6: Regression results – with interactions (continued) Determinants of usage of agricultural extension services Dependent variable: use of extension services = 1 if use at least one extension service, 0 otherwise Regressor All South East Center South (0.0564) (0.1032) (0.0815) (0.1002) Female*Technical 0.0034 0.5042 –0.0793 –0.4944*** (0.2827) (0.319) (0.1605) (0.1243) Female*University –0.5944** omitted* 0.0365 –0.6876*** (0.302) . (0.5143) (0.2532) Female with Farm size Female*Less_than_0.15 omitted omitted omitted omitted — — — — Female*From_0.15_ 0.0272 0.0655 –0.2219 0.0618 to_0.3 (0.0463) (0.0629) (0.1686) (0.0492) Female*From_0.3_ 0.0551 0.1265*** –0.3*** 0.0583 to_0.6 (0.0479) (0.0493) (0.1109) (0.0507) Female*From_0.6_ 0.0386 0.1323* –0.312** 0.0216 to_1.2 (0.0522) (0.0785) (0.1543) (0.053) Female*From_1.2_ 0.1315* 0.1365** –0.1661 0.0588 to_2.4 (0.071) (0.0623) (0.1761) (0.0665) Female*More_than_2.4 0.271** 0.014 0.0962 –0.2324* (0.1313) (0.0937) (0.1316) (0.124) Female with Crops Female*Maize 0.0161 0.0317 –0.0276 0.0121 (0.0486) (0.0679) (0.1069) (0.0517) Female*Beans –0.06 –0.0735 0.0764 –0.1378* (0.0694) (0.0494) (0.1407) (0.0777) Female*Bananas 0.0134 0.0476 0.0661* –0.0438 (0.0331) (0.0556) (0.0378) (0.0467) Female*Coffee 0.0251 0.0445 –0.1548** 0.0131 (0.0527) (0.0582) (0.0738) (0.1021) Female*Mangoes 0.0555 0.0296 0.0936 0.0415 (0.0388) (0.0523) (0.094) (0.058) Intercept –1.6591*** –1.6922*** –2.3708*** –1.9122*** (0.3768) (0.237) (0.4546) (0.4046) Observations 300100 85743 123814 90523 Pseudo R2 0.1823 0.0763 0.3016 0.1536 Source: Authors. * p-value < 0.1, ** p-value < 0.05, *** p-value < 0.01, omitted* = predicts perfect failure. Determinants of Agricultural Extension Services: The Case of HaitI 29 About the series: The LCSSD Occasional Paper Series is a publication of the Sustainable Development Depart- ment (LCSSD) in the World Bank’s Latin America and the Caribbean Region. The papers in this series are the result of economic and technical research conducted by members of the LCSSD community. The series addresses issues that are relevant to the region’s environmental and social sustainability; water, urban, energy and transport sector development; agriculture, forestry and rural development; as well as cross-cutting topics related to sustainable develop- ment such as climate change; logistics; crime and violence; and spatial economics. While all papers in this series are peer reviewed and cleared by the LCSSD Economics Unit on behalf of the Director of LCSSD, the findings, interpretations, and conclusions expressed in this paper, as in all publications of the LCSSD Occasional Paper Series, are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations or to members of its Board of Executive Directors or the countries they repre- sent. The World Bank does not garantee the accuracy of the data included in this paper and accepts no responsibility for any consequences of their use. For environmental sustainability reasons as well as ease of distribution, the primary vehicle for the dissemination of this series is electronic, with papers available on the following web- site: www.worldbank.org/lcssdpapersseries