Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Can integrated agriculture-nutrition extension programs deliver cost-

effectively? Evidence from the dissemination of vitamin A rich orange-


fleshed sweet potato in Mozambique

Ricardo A. Labarta & Jan W. Low


International Potato Center (CIP)
Sub-Saharan Africa office

Introduction

In recent years biofortification of staple food crops has been increasingly seen as
complementary strategy to combat micronutrient malnutrition among the poor in
developing countries as the majority of their energy intake comes from staple foods
(Stein et al 2004, Bouis 2002). Although many countries have experienced successful
supplementation and fortification programs (Rogers & Coales 2001), very poor countries
with populations concentrated in rural areas have limited access to health services and
processed food (Qaim et al 2006), and rely heavily on home-produced food and
inexpensive foods from the market, limiting the reach of supplementation and
fortification efforts.

Orange-fleshed sweetpotato (OFSP) is a particularly promising food to use in this regard


because levels of pro-vitamin A carotenoids are high in many varieties (100-1600 ug
Retinol Activity Equivalents (RAE) per 100 g for varieties in use in Africa) (van
Jaarsveld et al 2005; Hagenimana et al. 1999) and has proved to be and efficacious
alternative to poor Southern Africa countries like Mozambique where the Vitamin A
deficiency is estimated in 71% in children of 6-59 months old (Aguayo et al 2005).

Programs that have achieved a desired level of impacts with the dissemination of
biofortified food crops have had an integrated agriculture and nutrition approach. They
have however implied costly highly intensive extension programs which make it difficult
to scale up (Low et al 2007). Recently there have been many efforts that look to save
costs in the implementation of food based approaches but that have raised the concern on
whether a less intensive extension program may imply a lower level of achieved impact.
The question that arises is whether there is a massive dissemination program that
provides large savings but that could retain the level of impact achieved by intensive and
integrated extension approaches.

This paper evaluates the cost-effectiveness of two alternative food-based approaches


“Towards a Sustainable Nutrition Improvement (TSNI)” and “Eat Orange” to increase
vitamin A intake through disseminating OFSP and seeking to change child feeding
behaviors. Using farmer level data form rural Mozambique and a detailed cost structure
of the implementation of TSNI and Eat Orange, we estimate the impacts of TSNI and Eat
Orange participation on the adoption of OFSP vines, knowledge on vitamin A and good
feeding practices and consumption of OFSP roots and other vitamin A rich food and
finally we calculate the level of impacts achieved per unit of money invested in
disseminated OFSP.

Results from this study suggest that although an intensive extension approach like TSNI
could generate the highest impact per direct beneficiary, a lighter project like Eat Orange
compensates a lower level of impacts per direct beneficiary by reaching a larger number
of final beneficiaries.

The dissemination of OFSP in Mozambique

TSNI was an intensive extension project with a strong research component that was
aimed to combat vitamin A deficiency and inadequate caloric intake in rural Mozambique
(Low et al 2005). This project introduced and promoted Orange-Fleshed, beta-carotene
rich sweetpotato (OFSP) varieties as a low cost source of vitamin A. The design of the
project aimed to deliver strong agricultural messages to grow and maintain sweetpotato
planting material and strong nutritional messages to principal caregivers of children
under five years old. These messages included the consumption of OFSP roots and other
vitamin A rich food. This project also paid attention to market strategies that can support
a bigger OFSP production and therefore a greater consumption of OFSP roots among
these farmers’ households. This project was executed during 18 months between January
2003 and June 2004. The activities were concentrated in the districts of Mopeia and
Namacurra in the Zambezia province

Eat Orange was designed as a continuation of the TSNI experience with the purpose to
scale up the dissemination of OFSP in rural Mozambique. The main objective was to
expand the area of coverage and reach a much broader number of final beneficiaries but
trying to keep the same level of OFSP adoption, knowledge of vitamin A and good
feeding practices and the same level of consumption of OFSP roots and other vitamin A
rich food. This project was also executed during 18 months between January 2005 and
June 2006. Eat Orange covered large parts of Mopeia, Nicoadala, Gurue and Milange.

The higher intensity of TSNI required agriculture and nutrition extensionists work
directly with small groups of farmers and mothers belonging to the same households.
This strong integration of agriculture and nutrition messages and the reduced number of
final beneficiaries were aimed to produce a high level of impacts. On the other hand, Eat
Orange introduced an innovation by incorporating agriculture and nutrition promoters
between the extensionists and the final beneficiaries. These promoters follow some of the
characteristics of the key farmers described in the well known Training and Visit
extension approach (Benor et al 1984)). Each promoter was trained by the agriculture or
nutrition extensionist and then this extensionist would deliver the agriculture or nutrition
messages to the final beneficiaries. Thus, rather than have each extensionist work directly
with a small group of farmers, they work with agriculture and nutrition promoters who
reach large number of final beneficiaries.
Table 1 describes the main differences between the implementation of TSNI and Eat
Orange. Here we observe the large differences in term of intensity per final beneficiary
and also the differences in terms of number of final beneficiaries attended. Thus while
TSNI was able to attend 274 households per agriculture or nutrition extensionist, Eat
Orange agriculture extensionist had to attend indirectly 2,386 households and the
nutrition extensionists up to 3,976 households on average.

Table 1. Differences in the intensity of extension between TSNI and Eat Orange

TSNI Eat Orange

Total level of investment (US$) 510,588 696057

Total number of agronomists 1 2

Total number of nutritionists 1 0

Total number of Ag extensionists 4 5

Total number of Nut extensionists 4 3

Total number of farmers group 53 182

Total number of direct beneficiaries 1094 11929

Average size of farmer group 21 66

Number of beneficiaries/agronomist 548 5965

Number of beneficiaries/Ag extensionist 274 2386

Number of beneficiaries/Nut extensionist 274 3976

Another thing to notice is the difference on farmer group size. While one extensionist
was working with group of farmers with about 21 members, a less trained promoter had
to deal on average with 66 households, which is clearly another indication of the different
intensity for delivering agriculture and nutrition messages. The difference is even more
notorious on the nutrition side as Eat Orange did not have a nutritionist in the team and
was operating with only one nutrition extensionist for half of the project. The other two
nutrition extensionists started their activities in the second half of the project.
The analytical methods

Benefits and costs of implementation of both programs TSNI and Eat Orange are
estimated and compared in this paper. We first provide descriptive statistics about the
outcomes that both programs achieved at the household level for adoption of planting
material, changes in the nutrition knowledge and consumption of OFSP and other vitamin
A rich foods. Next we estimate econometrically the benefits that can be attributed to the
participation of farmers in either TSNI or Eat Orange among these three set of outcomes.
Then we broke down all the costs involved in the implementation of both extension
programs and compared the potential impact that can be achieved per unit of money
invested in both extension programs.

The data analyzed comes from the final survey in the TSNI project that involved 741
households in the districts of Mopeia and Namacurra in the Zambezia province, and a
cross sectional survey of 240 sweet potato growers in the districts of Nicoadala, Milange
and Gurue where the Eat Orange project operated in the same Zambezia province. In both
cases the two data sets involved participants and non-participants in the extension
programs (Low et al 2005, Labarta et al 2007).

The econometric strategy starts by controlling for the endogeneity of the participation of
farmers in either TSNI or Eat Orange. This participation may be correlated with specific
characteristics of the final participants that could not be observed during the surveys (i.e.
linkages with existing organizations, more opportunities to participate in the extension
programs, etc). Additionally both programs were not randomly placed as both programs
set a group of conditions for implementing both projects (i.e. access to lowlands, good
access for extension personnel, etc). Not controlling for this endogeneity may lead to an
overestimation of the impact of each extension program (Wooldridge 2002, Revallion
2005).

For estimating the impact of TSNI and Eat Orange participation on the adoption of OFSP
planting material we separately specified bivariate probit models for each dataset. In both
cases the main dependent variable was the answer to whether farmers had adopted or not
this OFSP planting material. The second equation in both cases estimates the
determinants of participating or not in each of the two extension programs.

For estimating the impact of both extension programs on nutritional knowledge we


specified a system of two simultaneous equations using three stage least squares
procedures. While one equation controls for the determinants of farmer participation in
each extension program, the other equation estimates the impact of TSNI and Eat Orange
participation on a knowledge score. This dependent variable was built on the answers to
12 different questions regarding household knowledge about vitamin A, breast feeding
and other good feeding practices for children under five years old (Low et al 2005). The
higher the nutritional score the better the nutritional knowledge of the household.

Finally, the impact of TSNI and Eat Orange on the consumption of OFSP and other
vitamin A rich food was estimated also using three stage least squares on simultaneous
equations. In addition to the participation equation, we include five other equations where
the dependent variable measured the frequency of the consumption of pumpkin, ripe
papaya, dark green leaves, small fish and OFSP during the last seven days prior to the
day of the interview.

For all the model specifications the explanatory variables included proxies for input and
output prices, household exposure to extension programs, socio economic characteristics,
household assets and community fixed effects (See detailed variables in Table 3 of next
section).

Implementation costs of both TSNI and Eat Orange were recorded by World Vision
International (WVI), the institution running both extension programs. Costs were then
organized by major categories including fixed and variable costs. World Vision also
provided information about the total number of direct beneficiaries involved in the
project implementation. We estimated the indirect beneficiaries of these two experiences
from information gathered in the two surveys available (Low et al 2005, Labarta et al
2007). We finally estimated the cost of implementation for direct and indirect beneficiary
by dividing the total cost of implementation over the number of direct beneficiaries and
indirect beneficiaries.

The final step is to compare costs and benefits of both projects and determine which
provides the highest impact for the same amount of investment in the dissemination of
OFSP. This exercise was complex as both extension programs generated different level
of impacts and required different level of investments. The main target was therefore to
convert benefits and costs into comparable units.

In this final step, we first took the econometric estimates of the average impact per
household that both programs, TSNI and Eat Orange produced directly on OFSP
adoption, nutrition knowledge, consumption on OFSP and consumption of other vitamin
A rich foods. Thus, from the adoption model we took the estimate of the marginal
likelihood of adopting OFSP planting material (changes in the likelihood of OFSP
adoption due to farmers’ participation either in TSNI or Eat Orange). From the nutrition
knowledge model we took the marginal changes in the nutrition knowledge score due to
exposure to TSNI and Eat Orange programs (how much nutrition knowledge can be
improved after participating in any of the two dissemination programs). Finally from the
OFSP and other vitamin A rich food consumption models we took the estimates of the
marginal change in the frequency of consumption of OFSP, pumpkin, papaya, small fish
and green leaves (number of consumption days during the last week) due to participation
in either TSNI or Eat Orange. All these effects can be attributed directly to the delivery of
both extension programs.

The second task was to add to the direct impacts of the extension programs the effects
that the program produced indirectly on neighbors of TSNI and Eat Orange participants
and on households located outside both project areas. For this purpose we first took from
the original two surveys used in this paper the average number of households that were
influenced in terms of OFSP adoption & consumption and in term of nutrition messages
from each participant household (either TSNI or Eat Orange). Then we used the
econometric point estimate of changes experienced by influenced farmers (neighbors of
participant farmers directly involved in OFSP dissemination) on OFSP adoption,
nutrition knowledge and consumption of OFSP and other vitamin A rich food. As the
TSNI survey did not discriminate the neighbors of TSNI clients among non-participant
farmers, we assumed that the indirect effect on TSNI neighbors was at the same
proportion as the indirect impact of the Eat Orange project compared to the direct impact
of the same extension program.

For measuring the indirect impact outside the project areas we took from the survey the
number of households that were contacted by participant farmers (TSNI or Eat Orange)
through the markets where distribution of OFSP vines occurred. Then we assumed the
same per household indirect effect as the one produced on neighbors of farmers
participating in OFSP dissemination.

Finally, by multiplying the number of influenced household reached through TSNI or Eat
Orange by the point estimate of the impact per indirect household we calculated the total
indirect effect produced by each direct participant in TSNI or Eat Orange. Adding this
indirect effect to the previously estimated direct effect provides the total impact for each
of both projects.

The last task is to simply estimate how much of the total impact estimated can be
produced per unit of money invested. In this case we estimated the total impact achieved
per each 10 US$ invested in the implementation of both programs.
Results and discussion

After 18 months of implementation, TSNI and Eat Orange were able to show positive
outcomes on participant households related to the adoption of OFSP planting material,
the increase of nutritional knowledge and the increase in the consumption of fresh OFSP
and other Vitamin A rich products. In general the more intense TSNI produced better
outcomes than Eat Orange at the end of both projects. However, in spite of the lower
intensity of Eat Orange, farmer participants in this program achieved a similar percentage
of OFSP adoption and a similar level of nutritional knowledge. TSNI did produce a
markedly better outcome in the consumption of OFSP and other vitamin A rich food.

Table 2. Selected outcomes on OFSP adoption, nutrition knowledge and


consumption of vitamin A rich foods among participants of TSNI and Eat Orange

Percentage of households that: TSNI Eat Orange

Were able to keep OFSP vines after


the end of the project 85% 81%

Had more than 50% of sweet potato


area under OFSP 68% 41%

Could recognize sources of Vitamin A 55% 35%

Know vitamin A benefits on


Body 59% 59%
Eyes 40% 30%

Feed children under 5 with OFSP


more than 2 days per week 55% 49%

Use OFSP to prepare child porridge 74% 24%

Consume other vitamin A rich food 47% 31%


more than twice a week

This better performance of the TSNI project outcomes shown in the descriptive statistics
of both projects outcomes are now tested econometrically in order to estimate the various
impacts that can be attributed purely to the participation in the TSNI dissemination
programs and then compared with the impacts achieved by the Eat Orange program.
Table 3 shows the descriptive statistics of all the variables used in the econometric
models described previously.
Table 3. Descriptive statistics. Mean (standard deviation) of variables used in the
econometric models

TSNI Eat Orange

Dependent variables
Has adopted OFSP (dummy) 0.83 (0.04) 0.81 (0.03)
Level of OFSP adoption (0 min, 6 max) 4.80 (0.19) 4.23 (0.26)
Knowledge score (0 min, 12 max) 8.10 (0.37) 6.34 (0.19)
Frequency of OFSP consumption (days in last week) 3.08 (0.36) 2.79 (0.27)
Frequency of small fish consumption (days in last week) 0.92 (0.19) 0.16 (0.23)
Frequency of papaya consumption (days in lat week) 3.23 (0.17) 3.42 (0.26)
Frequency of pumpkin consumption (days in last week) 0.87 (0.16) 0.74 (0.18)
Frequency of dark leaves consumption (days in last 1.43 (0.25) 0.78 (0.18)
week)

Explanatory variables 0.10 (0.01) 0.03 (0.02)


De facto female headed household (dummy) 0.06 (0.04) 0.05 (0.03)
De Jure female headed household (dummy) 0.84 (0.41) 0.92 (0.37)
Man headed household (dummy) 34.03 (1.01) 36.9 (1.14)
Household head age (years) 4.17 (0.31) 4.45 (0.27)
Household head education level (years) 2.70 (0.13) 3.10 (0.14)
Number of men at home 2.50 (0.14) 3.10 (0.17)
Number of women at home 1.89 (0.13) 1.95 (0.09)
Number of children under five at home 0.17 (0.08) 0.33 (0.05)
Learned about vitamin A in radio (dummy) 0.85 (0.15) 0.81 (0.14)
Learned about vitamin A trough extensionist (dummy) 0.07 (0.03) 0.08 (0.07)
Learned about vitamin A in health clinic (dummy) 53.4 (9.37) 72.51 (8.33)
Distance to market (minutes) 0.53 (0.08) 0.58 (0.06)
Use bicycle to go to the market (dummy) 0.01 (0.01) 0.01 (0.01)
Use car to go to the market (dummy) 0.89 (0.09) 0.67 (0.01)
Main market, local 0 0.15 (0.01)
Main Market, international 0.11 (0.03) 0.18 (0.01)
Main Market, district 0.83 (0.13) 0.94 (0.03)
Have lowlands 0.10 (0.03) 0.53 (0.05)
Hire temporary workers 2.95 (0.65) 3.02 (0.15)
Number of crop plots 0.17 (0.06) 0.46 (0.03)
Have irrigation 0.55 (0.06) 0.90 (0.02)
Produce maize 0.95 (0.04) 0.72 (0.01)
Produce rice 0.04 (0.01) 0.56 (0.06)
Produce sorghum 0.99 (0.02) 0.60 (0.63)
Produce cassava 0.46 (0.03) 0.81 (0.04)
Produce beans
All the econometric models described in the methodological section were estimated using
STATA 9.0. In this paper we report only the point estimates of the effect of TSNI and Eat
Orange participation on the adoption of OFSP vines, changes of nutritional knowledge
and on consumption of OFSP and other vitamin A rich foodi as summarized in Table 4.

Table 4. Impacts of TSNI and Eat Orange participation on household outcomes


related to OFSP adoption, nutrition knowledge and consumption of vitamin A rich
food
Average direct impact per household
TSNI Eat Orange

On OFSP adoption (probability of adoption) 0.24 0.12


On nutrition knowledge (change in score) 2.6 1.9
On OFSP consumption (days of consumption) 1.3 1.2
On other vitamin A food consumption (days) 0.3 0.14

The econometric results shown in Table 4 confirm that TSNI participation was
responsible for achieving a greater impact per direct household than the participation in
Eat Orange on most of the outcomes evaluated. In terms of adoption of OFSP planting
material the participation in TSNI increased the likelihood of adopting OFSP by 24%
while the participation in Eat Orange increased the probability of adoption in 12%.
Likewise, the participation in TSNI increased the nutrition knowledge (measured by the
knowledge score) by 2.6 points while the Eat Orange participation by 1.9 points. Finally
both extension programs similarly increased the consumption of OFSP by 1.3 days and
1.2 days per week respectively. The consumption of other vitamin A rich food was much
higher among TSNI participants although the magnitude of the increase of the
consumption was small (only 0.3 more days per week).

In addition to the different impacts achieved on direct beneficiaries there were indirect
impacts that were achieved on households influenced by direct beneficiaries and other
households that had contact with direct beneficiaries through other means like markets.
As explained in the methodological section we used the number of indirect beneficiaries,
the marginal effect produced by these indirect beneficiaries or influenced farmers and
estimated the indirect effects of both extension programs and therefore the total aggregate
impact of TSNI and Eat Orange. Table 5 summarizes these calculations.

The greater impact of TSNI participation on the adoption of OFSP, nutrition knowledge
and consumption of vitamin A rich food is notably increased when we add the indirect
effects produced through the farmer to farmer and market diffusions of OFSP material
and knowledge. As showed in the previous table a more intensive project like TSNI can
produce up to almost four times more total aggregated impact per direct and indirect
beneficiaries than the aggregate impact achieved by a less intensive project like Eat
Orange. This effect was especially marked in the OFSP vines adoption and in the
consumption of other vitamin A rich food. However, the consumption of OFSP among
children under five years old, the main target in both extension programs, produced a
difference in favor of TSNI of about only 35%. It is important to highlight that the units
of the indirect effects were “translated” into direct beneficiaries units in order to be able
to aggregate direct and indirect effects in a comparable way.

Table 5. Individual and aggregate impacts of TSNI and Eat Orange participation

TSNI Eat Orange


Indirect beneficiaries Farmer to Farmer 4.2 1.7
Through markets 1.4 0.8

Average indirect impact per influenced household


On OFSP adoption (probability of adoption) 0.17 0.08
On nutrition knowledge (change in score) 1.10 0.80
On OFSP consumption (days of consumption) 0.14 0.13
On other vitamin A food consumption (days) 0.13 0.06

Total indirect impact per direct beneficiary


On OFSP adoption (probability of adoption) 1.07 0.28
On nutrition knowledge (change in score) 8.04 3.39
On OFSP consumption (days of consumption) 1.97 1.44
On other vitamin A food consumption (days) 0.93 0.25

Total aggregate impact per direct beneficiary


On OFSP adoption (probability of adoption) 1.19 0.32
On nutrition knowledge (change in score) 8.76 3.90
On OFSP consumption (days of consumption) 2.08 1.53
On other vitamin A food consumption (days) 1.03 0.29

Now the central question to be addressed is: are these differences in project outcomes
cost-effective? Should we recommend disseminating OFSP in a TSNI type project? To
answer this question we need to look at the cost of implementation of both projects and
estimate the average impact achieved per unit of money invested.

As explained before TSNI was a research oriented project with a high level of intensity
and a very concentrated area of operation that required a considerable amount of
investment in order to run the project. On the other hand, Eat Orange were targeted to use
a similar amount of investment but in order to reach a much higher number of direct
beneficiaries. As shown in Table 6 the cost of implementing Eat Orange was slightly
higher than the TSNI project, but if we consider the level of operation of both projects the
cost of implementation per each direct beneficiary was about seven times more expensive
in the TSNI project. But as the TSNI project was able to disseminate indirectly OFSP to
more influenced households, the cost per final beneficiary (including direct and indirect
beneficiaries) can be estimated as 79 US$ for TSNI and US$ 22 for Eat Orange.

Table 6. The cost of implementing TSNI and Eat Orange projects in Mozambique

TSNI Eat Orange

Total Cost of implementation (US$) 510,588 696,057

Fixed Costs 315,921 447,707


Supervisors salaries 191,070 153,045
Supplies & non capital equipment 19,167 51,092
Travel 7,430 20,224
Office rentals 20,466 25,128
Personnel training 1,412 6,963
Services 398 3,151
Capital equipment 44,352 92,228
Indirect costs (overhead) 25,028 74,061

Variable cost 194,667 248,350


Extensionists and casual labor 140,231 159,355
Per diems for field staff 19,226 30,238
Fuel & lubricants 17,872 38,533
Vehicle maintenance 17,338 20,224

Total Cost per direct beneficiary 467 61

Fixed Cost per direct beneficiary 289 39

Variable cost per indirect beneficiary 178 22

Total cost per direct+indirect beneficiary 79 22

Fixed cost per direct+indirect beneficiary 49 14

Variable cost per direct+indirect beneficiary 30 8

In both projects fixed cost implies a big proportion of the total cost of the
implementation. The fixed cost in TSNI was 62% of the total cost while the fixed cost of
Eat Orange was 64% of total cost. If we only count the variable cost implied for reaching
an additional household, the investment needed is as low as 30 US$ per final beneficiary
in TSNI and US$ 8 per final beneficiary in the Eat Orange project.

The next step is to combine the impacts achieved by both programs and the cost for
implementing the same both programs. On one hand, we found a project like TSNI
producing considerable much more impacts than a project like Eat Orange. But on the
other hand, Eat Orange was targeted to reach a much greater number of direct
beneficiaries than TSNI, reducing considerably the cost per final beneficiary. Table 7
summarizes the final step of the exercise where we put together impacts and costs of
disseminating OFSP in Mozambique.

Table 7. The cost effectiveness of TSNI and Eat Orange projects in Mozambique

TSNI Eat Orange

Expected impact per each US$10 of total cost


On OFSP adoption (probability of adoption) 0.14 0.13
On nutrition knowledge (change in score) 1.02 1.54
On OFSP consumption (days of consumption) 0.25 0.65
On other vitamin A food consumption (days) 0.12 0.11

Expected impact per each US$ 10 of variable cost


On OFSP adoption (probability of adoption) 0.36 0.35
On nutrition knowledge (change in score) 2.68 4.24
On OFSP consumption (days of consumption) 0.66 1.81
On other vitamin A food consumption (days) 0.31 0.31

Although there were stronger direct and indirect effects of TSNI given the higher
intensity in the implementation, the major advantage of Eat Orange is to be able to reach
around 11 times more direct beneficiaries, this translates into a project that requires about
4 times less investment (with large cost savings). As shown by table 7, per each 10
dollars invested, Eat Orange was able to reach the same level of impacts as Eat Orange in
OFSP adoption and in the consumption of other vitamin A rich food and a greater total
impact in nutrition knowledge and consumption of OFSP.

If we consider only the variable costs, we can really expect much better results from eat
Orange. So if we are going to expand an existing project (not considering fixed costs) we
can expect than an investment of US$10 in a project like Eat Orange would increase the
likelihood of OFSP adoption by 36%, increase the nutrition knowledge by 4.24 point
(measured by the knowledge score with a maximum of 12 points), increase the
consumption of OFSP by 1.81 days per week and increase the consumption of other
vitamin A rich foods by 0.31 days per week, on average on the new households
participating in the project.
Conclusions

Integrating agricultural and nutrition extension enhances the impact of food based
approaches aimed to combat vitamin A deficiency and to increase access to the intake of
vitamin A rich food. This paper provides strong evidence that in rural Mozambique
programs disseminating OFSP increased the adoption of this vitamin A rich crop,
improved the knowledge that participant households had on vitamin A and good feeding
practices and increased the intake of OFSP roots and other vitamin A rich food among
children under five years old. However, the level of impact varied according to the
intensity of each extension program.

TSNI was a very intensive extension program including both agriculture and nutrition
that used extensionists to directly work with small groups of farmers and was able to
produce up to four times more impact than the Eat Orange project. Eat Orange used
agriculture and nutrition promoters as intermediaries between extensionists and final
beneficiaries and reached more than 10 times the number of direct beneficiaries of TSNI
but achieving a lower impact per participant household. However this lower impact
achieved on Eat Orange participants was largely compensated by the cost savings Eat
Orange generated in disseminating OFSP to a broader number of final beneficiaries.
Overall the expected impact per unit of money invested in disseminating OFSP were
similar in both TSNI and Eat Orange extension programs.

Although a lighter extension approach like Eat Orange could be promising for
disseminating bio-fortified food crops, there is still the need to measure the sustainability
of the impacts of alternative extension approaches found in the short term. Disadoption
processes could be different after dissemination approaches with different degree of
intensity. This could have implications on the impacts on the long run. This is something
beyond the scope of this paper but that will require further attention in future research.
References

Aguayo V, Kahn S, Ismael C, Meershoek S. (2005) Vitamin A deficiency and child


mortality in Mozambique. Public Health Nutr. 8:29-31

Benor, D.; J. Harrison and M. Baxter (1984). Agricultural extension: The Training and
Visit System. World Bank, Washington D.C.

Bouis, H. (2002) A new tool for fighting micronutrient malnutrition. Symposium: Plant
breeding: A new tool for fighting micronutrient malnutrition. J Nutr. 132: 491S-94S

Hagenimana V, Carey EE, Gichuki ST, Oyunga MA, Imungi JK. Carotenoid contents in
fresh, dried and processed sweetpotato products. Ecology of Food and nutrition 1999;
37:455 – 473.

Labarta, R.; J.low; M. Andrade and S. Namanda (2007). Seed System Component End of
the Year Technical Report to Harvest Plus. Reaching End Users Project. International
Potato Center, Quelimane, Mozambique

Low, J.; M. Arimond; N. Osman; A. Osei; F. Zano; B. Cunguara; M. Selemane; D.


Abdullah and D. Tschirley (2005).Towards Sustainable Nutrition Improvement in Rural
Mozambique: Addressing Macro- and Micro-nutrient malnutrition trough new cultivars
and new behaviors. Final Report

Low JW, Arimond M, Osman N, Cunguara B, Zano F, Tschirley, D.(2007) A food-based


approach introducing orange-fleshed sweet potatoes increased vitamin A intake and
serum retinol concentrations among young children in rural Mozambique. J Nutr 2007;
137, 1320-1327.

Ravallion, M.(2005) Evaluating anti-poverty programs. World Bank Policy Research


Working Paper N. 3625, Washington D.C.

Qaim, M., A. Stein & J.V Meenakshi (2006). Economics of biofortification. Paper
presented at the meeting of the International Association of Agricultural Economics
(IAAE). Gold Coast, Australia.

van Jaarsveld PJ, Faber M, Tanumihardjo SA, Nestel P, Lombard CJ, Benade J. (2005)
Beta-carotene-rich orange- fleshed sweet potato improves the vitamin A status of
primary school children assessed with the modified- relative-dose-response test. Amer
Jour of Clin Nutr 81,1080-1087.

Wooldridge, J (2002). Econometric Analysis of Cross Section and Panel Data. MIT
Press, Cambridge, MA.

i
Details of each model specification and full regression model results can be found in Labarta et al (2007).

You might also like