Biofortified Maize Impact Analysis
An ex-ante model of biofortification in Guatemala
Guatemala has one of the highest stunting rates in the world. Biofortified maize varieties are bred to contain higher levels of iron, zinc, and quality protein (lysine and tryptophan) so that an everyday staple food can deliver more nutrition to households.
This site documents the methodology used to estimate the potential impact of biofortified maize for both the children who eat the maize and for the farmers who grow it.
Model description
This framework is an ex-ante model: it projects the impacts of biofortified maize across the country in advance of large-scale rollout.
The ex-ante model is built as a sequence of distinct analytical problems, each solved using real Guatemalan data and chained together so that the output of one stage becomes the input of the next. The chain starts from reported household food consumption and ends in a nationwide simulation of impact. Every stage is documented in full on this site; this page shows how those stages connect.
The model addresses two questions:
- The nutritional dimension — how biofortified maize changes children’s nutrient intake, and how that change translates into improved growth and reduced stunting.
- The economic dimension — which farmers are likely to adopt the improved seed, and how adoption affects household income.
Both dimensions are combined in a two-dimensional simulation across 22 departments and a range of production coverage levels (the share of maize-growing area planted with biofortified seed).
From seed to national simulation
The intervention follows a short logical chain: biofortified seed increases the nutrient content of the maize that households consume, which raises children’s nutrient intake, which is linked to improved growth and lower stunting rates. Estimating the magnitude of each link from Guatemalan data is what the model does, and the diagram below shows the stages it is organised into. A parallel branch follows the agronomic consequences of the same intervention for farmers.
The two dimensions begin from the same point — the decision to plant biofortified seed — and run in parallel. The economic branch determines how biofortified maize adoption impacts farmer incomes and how much biofortified maize reaches households at each coverage level; the nutritional branch determines what that maize does for child growth. The national simulation combines both.
Two dimensions, modeled separately, simulated together
Nutritional dimension
Children’s daily food consumption is estimated from household survey data, the maize contribution is separated from the rest of the diet, and the effect of biofortification on nutrient intake is estimated.
Estimating the resulting change in child growth and stunting rates is the most demanding step in the model: growth depends on many factors beyond diet, and the link between nutrient intake and growth is indirect. For this reason, child growth and stunting reduction are estimated through two independent pathways:
- Covariate prediction — a statistical model relating each child’s nutrient profile to height-for-age, applied to a simulated child population.
- Meta-analysis (Gunaratna) — a published effect size for quality protein maize, applied to the Guatemalan child population.
Economic dimension
Guatemalan maize farmers are characterised from national survey data, calibrated against the official agricultural census.
The model then estimates which farmers are likely to adopt the biofortified seed and how adoption changes household income. Adoption determines how much biofortified maize is produced, and therefore how much reaches households at each coverage level — the link between an agricultural intervention and a nutritional outcome.
Why two pathways for stunting. The two pathways rest on different data sources and different modelling assumptions. The covariate pathway derives the relationship between nutrient intake and growth from the Guatemalan anthropometric survey; the meta-analysis pathway applies an externally published effect size for quality protein maize. Estimating the same outcome by two routes with independent stunting-estimation machinery, applied to the same upstream adoption and distribution scenario, allows the convergence of their results to be examined directly. With the 2028 seed at full production coverage, the two pathways converge on a reduction of 5.6 to 5.8 percentage points in child stunting prevalence.
How the model is built: six modules
The ex-ante model is implemented as six modules. Each module solves one problem in the chain and passes its result to the next. Modules 1, 3, 4 and 5 build and apply the nutritional dimension, module 2 covers the economic dimension, and module 6 combines both into the national simulation. Each module links to its full technical documentation.
Where the numbers come from
The model is built on national data sources.
| Dataset | Description | Year | Provider |
|---|---|---|---|
| ENCOVI | National Survey of Living Conditions | 2023 | INE Guatemala |
| SIVESNU | National Anthropometric Surveillance System | 2018 | SESAN / INCAP |
| ENSMI | National Maternal and Child Health Survey | 2014–2015 | MSPAS / INE |
| INE | Vital statistics: annual births | 2019–2023 | INE Guatemala |
| INCAP | Food composition tables | 2014 | INCAP Guatemala |
| WHO | Child growth standards (height-for-age) | 2006 | World Health Organization |
| MAGA | Agricultural production statistics | 2023 | Ministry of Agriculture |
| OSRM / OpenStreetMap | Departmental travel-time matrix (22×22) | 2026 | Open routing project |
Citation
@software{newseed_biofortification_guatemala,
title = {Biofortified Maize Impact Assessment Framework for Guatemala},
author = {Sánchez Tormo, Julia María and Palomo Llinares, Rubén and
Soto Méndez, María José and Adams, Katherine P. and Bowen, Curt},
year = {2026},
version = {4.0.0},
license = {AGPL-3.0},
url = {https://github.com/Azahar-Data-Insights/newseed-biofortification-guatemala-public}
}Contact
- Methodology and analysis: Azahar Data Insights — azahardata.com
- Client: Semilla Nueva — semillanueva.org