Autonomy and Specificity in Agricultural Technology Adoption: Evidence from Mexico

The experiment consists of a control group and four treatment arms
that combine soil analysis and recommendations (either at the plot or
cluster level) with flexible or inflexible in-kind grants. All
treatment arms were offered a soil analysis report, a set of input
recommendations and a package of agricultural extension services
designed to help them implement the recommendations. In addition,
three of the four treatment arms were offered an in-kind grant of
2,000 pesos (roughly 150$ usd at the time of the intervention).

In particular, arm T1 received individualized or plot level
soil analyses with input recommendations and an inflexible
in-kind grant. Arm T2 was the same as T1 except that the soil
analysis and recommendations were \averaged (so that
recommendations were identical for all farmers in the
cluster). Comparing T1 with T2 allows us to estimate the effect of
varying the level of the specificity of the soil analysis and
recommendations (conditional on receipt of an inflexible grant). Arm
T3 differed from T2 in that the in-kind grant was
flexible and so farmers in T3 could use the in-kind grant
to purchase any inputs of their choosing sold by a local agro-dealer
rather than just the recommended ones. A comparison between T2 and
T3 thus measures the effect of autonomy as defined above. The
conditions for arm T4 are the same as those for T2 (or T3)
except that no grant was provided. Comparisons between T4 and T2
(or T3) measure the effect of providing the in-kind grant
(conditional on receipt of localized information and
recommendations). Finally, a control group of farmers C did not
receive any interventions during the experiment, but instead received
soil analyses and recommendations -- but no extension services -- the
year after the intervention ended. Comparisons between C and T4
estimate the effect of providing localized soil analyses,
recommendations and extension services. In later years, after the
intervention had ended and control farmers had also received soil
analyses and recommendations, comparing C and T4 provides an
estimate of the effect of agricultural extension services paired with
of analyses and recommendations but with qualifications, since the
control group receives the soil analysis and recommendations one year
later.

2014-01-01

2017-12-30

1) Compliance with fertilizer recommendations
2) Adoption of new practices
3) Persistence of the adoption of new practices

1) Measurements of kilos of Urea, diammonium phosphate (DAP), and Potassium Chloride (KCl) used, relative to recommended.
2) Standardized index of“new” agricultural practices introduced by the intervention. We classify six practices as ``new'' since they
were uncommon at baseline and recommended by the extension workers
3) We measure standardized index of new practices 2 years after the end of the experiment.

4) Yields (tons of maize per hectare)
5) Profit

4) Self reported tons of maize per hectare harvested. We also harvested a subset of plots ourselves to check the quality of this self reported measure.
5) Profit = Yields * Price of Maize as sold by farmer - costs, including labor and input costs in all stages of planting, fertilizing and harvesting.

The five experimental arms are a combination of three components: (a) soil analysis and recommendations (individual or averaged), (b) extension services and (c) and in-kind grants (conditional or unconditional) as outlined in the introduction. To summarize:
T1: Individualized or plot-level soil analysis, input recommendations, extension services; in-flexible in-kind grant.
T2: Averaged or cluster-level soil analysis, input recommendations, extension services; inflexible in-kind grant. (i.e. same as T1 but with averaged instead of individual soil analyses and recommendations).
T3: Averaged soil analysis, input recommendations, extension services; flexible in-kind grant.(i.e. same as T2 but with a flexible instead of inflexible in-kind grant).
T4: Averaged soil analysis, input recommendations; extension services. (i.e. same as T3 or T2 but with no in-kind grant).
C: No intervention. Control Arm received averaged soil analysis and recommendations they ear after the intervention ended (in early 2016).

We chose to include agricultural extension services in all treatment arms because a pilot in the same study area with a comparable sample had suggested limited value of the soil analysis and recommendations without the extension services, as farmers appeared to greatly value the ability question and discuss the recommendations with the extension agents. Grants were provided in-kind (rather than in cash) for three reasons. First, because agro-dealers did not typically stock fertilizers in the blends required by the recommendations we partnered with Agropecuaria Amozoc, who agreed to offer the tailored high-quality fertilizer packages to farmers as long as we guaranteed a minimum volume of sales. Second,the in-kind grant was intended as a “push” to farmers to experiment with higher quality in-puts (fertilizers manufactured by a reputable high-quality firm). Finally and perhaps most importantly, cash grants were simply not possible because organizations such as QFD were not allowed by law to disburse cash grants.

Randomization by computer at office

farmer level randomization, within geographical strata.

No

Aimed for 900 individual farmers

We aimed for a sample of close to 900 farmers, we have 678 total.

We have close to 135 observation per arm