Precompute Biofortification Scenarios

Module 6: Scenarios

Why Precompute Scenarios?

The Shiny dashboard requires real-time response to user interactions. Computing biofortification impacts on-the-fly would involve:

Resolving each farmer’s conversion ratio along the continuous adoption curve at the internal adoption index associated with every national production coverage level.
Looking up pre-computed individual profiles (baseline intake, QPM intake, baseline stunting, QPM stunting) for the 4,065 children of the synthetic population or the 4,134 ENCOVI children, depending on the pathway.
Allocating biofortified maize across departments using the gravity-based redistribution from the market baseline.
Aggregating survey-weighted metrics across 22 departments and 11 production coverage levels.

This pipeline takes approximately 3-5 minutes per scenario combination. The framework evaluates a grid of 18 scenario combinations: three seed variants (2022, 2026 and 2028 seed), three subsidy regimes (current, 50% reduction, 75% reduction), and two stunting methodologies (covariate prediction and meta-analysis). Computing the grid on demand would take longer than an interactive session allows, so every scenario is computed in advance and the dashboard reads the result.

This page documents the simulation methodology with the baseline scenario combining the 2022 biofortified seed variety with the current subsidy regime as of April 2026 as the canonical reference. Alternative seed variants and subsidy regimes follow the same methodology with different upstream parameters.

Simulation Architecture

The scenario simulation pipeline generates two complementary datasets for each combination of seed × subsidy × pathway:

Output	Rows	Use case
Departmental scenarios	242 (22 × 11)	Geographic comparisons, departmental filtering
National scenarios	11	National-level indicators, overview metrics

The two share the same simulation logic and differ in the aggregation level: the departmental dataset aggregates metrics by departamento, and the national one computes weighted means directly at the Guatemala level.

Direct Aggregation Principle

National metrics are computed from individual-level data using survey weights, not by averaging departmental summaries. Averaging the summaries would weight each department equally regardless of its population, which is where Simpson’s paradox enters.

Pre-Computed Individual Profiles

A defining feature of this module is the separation of responsibilities with respect to the upstream nutritional and stunting modeling:

Module 6 receives pre-computed individual profiles (baseline intake, QPM intake, baseline stunting, QPM stunting, delta HAZ) from Modules 4 and 5. Module 6’s role is exclusively to (i) determine which children receive biofortified maize at each production coverage level using the hierarchical priority assignment, and (ii) aggregate the resulting individual outcomes to departmental and national level. The nutritional modeling itself (intake calculation, bioavailability adjustment, stunting prediction) is upstream of this module.

This separation is important for two reasons. First, it makes Module 6 fast and reproducible — no re-fitting of models, no recomputation of nutrient intakes, just lookup and aggregation. Second, it isolates the decisions that change across pathways (which stunting methodology is applied) from the decisions that change across production coverage levels (which farmers adopt and which children consume biofortified maize).

The source of the individual profiles depends on the pathway:

Covariate prediction pathway (PCO) — Profiles come from the synthetic population built in Baseline and QPM Profiles. Stunting is predicted with the GAM fitted in Module 4.
Meta-analysis (Gunaratna) pathway (PMA) — Profiles come from the ENCOVI children carrying the hierarchy of Biofortification Priority Assignment. The stunting impact uses the Gunaratna et al. (2010) meta-analysis effect size.

Simulation Pipeline

For each production coverage level (0% to 100% in 10% increments) the pipeline executes five sequential steps:

Step 1: Determine Each Farmer’s Conversion Ratio

Production coverage is implemented as a continuous adoption process driven by the per-farmer parameters calibrated in Module 2. Each production coverage level is evaluated at its internal adoption index p_internal, the position on the national adoption ranking that plants that share of maize-growing area with biofortified seed. Every farmer with entry point p_entry ≤ p_internal is in the adoption pool, and converts a fraction of their maize land, and of their potential biofortified production, determined by the profile-specific sigmoid:

\[ r_i(p_{internal}) = \text{sigmoid}(p_{internal}, p_{entry,i}, \text{floor}_{profile}, k_{profile}) \]

The effective biofortified production of farmer i at that production coverage level is:

\[ \text{bio}_i(p_{internal}) = \text{production\_biofortified}_i \times \text{weight\_calibrated}_i \times r_i(p_{internal}) \]

This formulation expresses adoption as a smooth conversion intensity per farmer. The conversion curves and their calibration are documented in the Market Baseline Calculation page, together with the production coverage scale that associates each level with its internal adoption index.

Step 2: Calculate Economic Metrics

For adopting farmers (those with p_entry ≤ p_internal), the script calculates survey-weighted aggregates:

n_farmers_bio — Total adopting farmers (with conversion ratio applied).
land_bio_ha — Biofortified land area (hectares).
production_biofortified_total — Total biofortified production (quintales).
income_increment_total — Additional farmer income (Q).
cost_seeds_increment_total — Change in seed costs (Q).

At national level, land_bio_ha over the total maize land reproduces the production coverage level of each scenario, within the association error of the production coverage scale.

The seed cost component depends on the active subsidy regime: under the current regime the per-area subsidy is funded entirely by New Seed, while under the reduced regimes part of the cost becomes producer subsidy absorption and part is passed to the farmer. The decomposition by actor is documented in Module 2.

Step 3: Retrieve Departmental Coverage

Consumption coverage percentages at each production coverage level are retrieved from Market Baseline Calculation, where the gravity-based redistribution has already been applied. Both stages therefore read the same coverage trajectory.

\[ \text{Coverage}_d(p) = \frac{\text{Biofortified production consumed in } d \text{ at } p}{\text{Total maize consumption in } d} \]

Consumption coverage can exceed local production share when departments receive redistributed surplus from saturated neighbours through the gravity model.

Step 4: Assign Biofortified Maize to Children

Children in the target population receive biofortified maize based on departmental consumption coverage and household priority. This is the methodologically complex step of the pipeline.

Hierarchical Assignment Logic

ENCOVI (farmer data) and SIVESNU (child anthropometry) cannot be linked at the household level — there is no common identifier. We cannot directly identify children of adopting farmers who consume biofortified maize from own-production.

The hierarchical assignment uses observable proxies in the target population (head of household occupation, urban/rural area) to approximate the probability that a child’s household would access locally-produced biofortified maize. This is a probabilistic assignment based on plausible consumption patterns, not a deterministic link to specific adopting households.

Priority hierarchy. The two pathways order children differently, because the information available about each child’s household differs between them.

The covariate prediction pathway works on the synthetic population, where the household is characterised by the occupation of its head and by area of residence:

Priority	Group	Rationale
1	Cereal farmers (`Granjero cereales`)	Direct access to biofortified production
2	Other farmers (`Granjero otros`)	Agricultural households with market access
3	Agricultural workers (`Trabajador del campo`)	Rural employment in the farming sector
4	Rural, other occupations	Rural households without a direct farming connection
5	Urban households	Dependent on commercial distribution

The Meta-analysis (Gunaratna) pathway works on the ENCOVI children and reuses the household-level hierarchy built in Biofortification Priority Assignment, which is based on observed agricultural production rather than on occupation: maize producers with self-consumption first, then maize producers without, then other agricultural households, then rural and urban.

Assignment algorithm. For each department at each production coverage level, children are sorted by priority and, within a priority level, by a random tie-breaker. The tie-breaker is drawn per child in the covariate pathway and per household in the meta-analysis pathway, so siblings there enter together. The cumulative survey weight is then accumulated down that order and compared against the departmental target:

\[ W_{\text{target},d} = \text{coverage}_d \times \sum_{i \in d} w_i \qquad \text{flag}_i = \mathbb{1}\left[\sum_{j \preceq i} w_j \le W_{\text{target},d}\right] \]

The quota is therefore expressed in represented children rather than in sampled records, which keeps the assigned share consistent with the departmental consumption coverage under the survey design. Farmer households enter first, urban households only once rural demand is met, and the ordering inside each priority level carries no further structure.

Step 5: Lookup Pre-Computed Profiles and Aggregate

For each child, the script retrieves the pre-computed individual profile produced upstream:

If flag_biofortified = 1, use the QPM profile (intake under biofortified maize consumption, stunting outcome under QPM).
If flag_biofortified = 0, use the baseline profile (intake under conventional maize, baseline stunting outcome).

Individual-level outcomes are then aggregated using survey weights to produce departmental and national metrics: mean intakes, inadequacy prevalence, stunting prevalence, mean HAZ, and cases averted.

Heterogeneous Effects

Heterogeneity in the response to biofortification enters upstream, at the level of the individual profiles. In the covariate prediction pathway the GAM admits a non-linear relationship between the nutrient index and HAZ; over the fitted range the estimated smooth turns out close to linear, so each child’s change in HAZ follows mainly from the size of their own change in the index. In the Meta-analysis (Gunaratna) pathway the effect size is applied to height growth velocity, modulated per child by their protein adequacy. Both routes produce per-child outcomes, which is what allows the aggregation at this stage.

Simulation Results

The following sections summarise the pre-computed national scenarios, organised in three parallel blocks. The first two blocks document the stunting and nutrient outcomes under the two pathways (covariate prediction and meta-analysis), with the current subsidy regime held fixed and the three seed variants compared. The third block documents the economic outcomes (which are invariant to the pathway), with the full grid of three seed variants and three subsidy regimes.

Scenarios Under Covariate Prediction Pathway

The covariate prediction pathway estimates per-child stunting outcomes using a GAM model fitted in Module 4 on the synthetic population generated in Module 4. For each child, the model captures the non-linear relationship between the nutritional index and the height-for-age z-score (HAZ), and the HAZ change is the difference between its fitted values at the baseline and QPM index positions. Aggregating individual outcomes across all children yields the national progression below.

This block compares the three seed variants under the current subsidy regime. The seed variant primarily affects yield change factors per segment, and therefore the volume of biofortified production available at each production coverage level; subsidy regimes (held constant here) modulate the order in which farmers enter the adoption pool but converge at the endpoints. See the Economic Progression section for the full subsidy × seed grid on the economic side.

2022 Seed — Covariate Prediction

Table 1: National scenario progression — 2022 seed, covariate prediction pathway, current subsidy

National Metrics by Production Coverage Level — Seed 2022. Pathway: Covariate prediction | Subsidy: current regime
Production Coverage	Farmers Bio	Coverage	Stunting (%)	Mean HAZ	Zinc Inad. (%)	Iron Inad. (%)
National Metrics by Production Coverage Level — Seed 2022
Pathway: Covariate prediction \| Subsidy: current regime
0%	0	0.0%	48.6	−2.026	52.8	53.8
10%	209,884	12.6%	47.6	−2.004	51.7	53.2
20%	391,415	22.3%	46.9	−1.985	51.0	52.7
30%	504,524	32.6%	46.2	−1.966	50.1	52.4
40%	582,959	42.7%	45.6	−1.947	49.0	51.7
50%	624,536	53.1%	44.9	−1.926	47.7	51.2
60%	663,209	63.1%	44.2	−1.908	47.1	50.6
70%	687,040	71.7%	43.9	−1.895	46.6	50.2
80%	741,392	78.1%	43.5	−1.885	46.0	49.9
90%	751,118	82.7%	43.3	−1.879	45.6	49.7
100%	783,474	85.5%	43.2	−1.877	45.5	49.6

Table 2: National nutrient intake progression — 2022 seed, covariate prediction pathway, current subsidy

Nutrient Intake by Production Coverage Level — Seed 2022. Weighted mean bioavailable nutrients — Pathway: Covariate prediction | Subsidy: current regime
Production Coverage	Coverage	Absorbed Intake			Change vs Baseline
Nutrient Intake by Production Coverage Level — Seed 2022
Weighted mean bioavailable nutrients — Pathway: Covariate prediction \| Subsidy: current regime
Production Coverage	Coverage	Zinc (mg)	Iron (mg)	Protein (g)	Zn Δ%	Fe Δ%	Prot Δ%
0%	0.0%	1.121	0.513	15.44	0.0%	0.0%	0.0%
10%	12.6%	1.131	0.519	16.35	+0.9%	+1.2%	+5.9%
20%	22.3%	1.138	0.524	17.13	+1.6%	+2.1%	+10.9%
30%	32.6%	1.146	0.529	17.92	+2.3%	+3.1%	+16.1%
40%	42.7%	1.154	0.534	18.68	+3.0%	+4.1%	+21.0%
50%	53.1%	1.162	0.539	19.52	+3.7%	+5.1%	+26.4%
60%	63.1%	1.169	0.544	20.26	+4.3%	+6.1%	+31.2%
70%	71.7%	1.175	0.547	20.81	+4.8%	+6.8%	+34.8%
80%	78.1%	1.179	0.550	21.21	+5.2%	+7.3%	+37.4%
90%	82.7%	1.181	0.552	21.47	+5.4%	+7.6%	+39.0%
100%	85.5%	1.183	0.552	21.57	+5.5%	+7.7%	+39.7%
Zinc and iron values are absorbed (TAZ via Miller equation, Fe at 5% absorption). Protein is PDCAAS-adjusted.

2026 Seed — Covariate Prediction

Table 3: National scenario progression — 2026 seed, covariate prediction pathway, current subsidy

National Metrics by Production Coverage Level — Seed 2026. Pathway: Covariate prediction | Subsidy: current regime
Production Coverage	Farmers Bio	Coverage	Stunting (%)	Mean HAZ	Zinc Inad. (%)	Iron Inad. (%)
National Metrics by Production Coverage Level — Seed 2026
Pathway: Covariate prediction \| Subsidy: current regime
0%	0	0.0%	48.6	−2.026	52.8	53.8
10%	207,403	13.9%	47.4	−2.002	51.5	53.1
20%	385,331	24.6%	46.8	−1.981	50.7	52.6
30%	500,704	35.6%	46.1	−1.961	49.8	52.3
40%	585,579	46.9%	45.2	−1.938	48.4	51.5
50%	633,774	57.4%	44.6	−1.919	47.2	50.7
60%	662,533	68.2%	44.0	−1.901	46.7	50.3
70%	698,760	77.3%	43.5	−1.887	46.1	50.0
80%	732,692	84.9%	43.3	−1.877	45.5	49.6
90%	749,656	90.3%	43.2	−1.872	45.2	49.5
100%	783,474	93.9%	43.2	−1.869	44.9	49.5

Table 4: National nutrient intake progression — 2026 seed, covariate prediction pathway, current subsidy

Nutrient Intake by Production Coverage Level — Seed 2026. Weighted mean bioavailable nutrients — Pathway: Covariate prediction | Subsidy: current regime
Production Coverage	Coverage	Absorbed Intake			Change vs Baseline
Nutrient Intake by Production Coverage Level — Seed 2026
Weighted mean bioavailable nutrients — Pathway: Covariate prediction \| Subsidy: current regime
Production Coverage	Coverage	Zinc (mg)	Iron (mg)	Protein (g)	Zn Δ%	Fe Δ%	Prot Δ%
0%	0.0%	1.121	0.513	15.44	0.0%	0.0%	0.0%
10%	13.9%	1.132	0.519	16.44	+1.0%	+1.3%	+6.5%
20%	24.6%	1.140	0.525	17.30	+1.7%	+2.3%	+12.0%
30%	35.6%	1.149	0.530	18.12	+2.5%	+3.4%	+17.3%
40%	46.9%	1.157	0.536	19.02	+3.2%	+4.5%	+23.2%
50%	57.4%	1.165	0.541	19.85	+3.9%	+5.6%	+28.5%
60%	68.2%	1.172	0.546	20.55	+4.6%	+6.4%	+33.1%
70%	77.3%	1.178	0.549	21.13	+5.1%	+7.2%	+36.9%
80%	84.9%	1.183	0.552	21.56	+5.5%	+7.7%	+39.6%
90%	90.3%	1.185	0.554	21.76	+5.7%	+8.0%	+40.9%
100%	93.9%	1.187	0.554	21.89	+5.9%	+8.1%	+41.8%
Zinc and iron values are absorbed (TAZ via Miller equation, Fe at 5% absorption). Protein is PDCAAS-adjusted.

2028 Seed — Covariate Prediction

Table 5: National scenario progression — 2028 seed, covariate prediction pathway, current subsidy

National Metrics by Production Coverage Level — Seed 2028. Pathway: Covariate prediction | Subsidy: current regime
Production Coverage	Farmers Bio	Coverage	Stunting (%)	Mean HAZ	Zinc Inad. (%)	Iron Inad. (%)
National Metrics by Production Coverage Level — Seed 2028
Pathway: Covariate prediction \| Subsidy: current regime
0%	0	0.0%	48.6	−2.026	52.8	53.8
10%	212,717	15.1%	47.4	−2.000	51.5	53.1
20%	387,123	27.2%	46.7	−1.976	50.6	52.4
30%	500,057	39.3%	45.9	−1.954	49.2	51.9
40%	586,628	51.3%	45.0	−1.931	48.1	51.2
50%	633,641	62.8%	44.3	−1.910	47.0	50.5
60%	666,985	72.9%	43.7	−1.894	46.5	50.2
70%	703,083	82.4%	43.3	−1.880	45.7	49.8
80%	725,591	90.3%	43.2	−1.872	45.1	49.5
90%	751,475	96.4%	43.1	−1.867	44.7	49.5
100%	783,474	100.0%	43.0	−1.863	44.5	49.3

Table 6: National nutrient intake progression — 2028 seed, covariate prediction pathway, current subsidy

Nutrient Intake by Production Coverage Level — Seed 2028. Weighted mean bioavailable nutrients — Pathway: Covariate prediction | Subsidy: current regime
Production Coverage	Coverage	Absorbed Intake			Change vs Baseline
Nutrient Intake by Production Coverage Level — Seed 2028
Weighted mean bioavailable nutrients — Pathway: Covariate prediction \| Subsidy: current regime
Production Coverage	Coverage	Zinc (mg)	Iron (mg)	Protein (g)	Zn Δ%	Fe Δ%	Prot Δ%
0%	0.0%	1.121	0.513	15.44	0.0%	0.0%	0.0%
10%	15.1%	1.132	0.520	16.50	+1.0%	+1.3%	+6.9%
20%	27.2%	1.142	0.526	17.48	+1.9%	+2.6%	+13.2%
30%	39.3%	1.151	0.532	18.38	+2.7%	+3.7%	+19.0%
40%	51.3%	1.160	0.538	19.34	+3.5%	+4.9%	+25.3%
50%	62.8%	1.169	0.544	20.21	+4.3%	+6.0%	+30.9%
60%	72.9%	1.175	0.548	20.85	+4.8%	+6.8%	+35.0%
70%	82.4%	1.181	0.552	21.45	+5.4%	+7.6%	+38.9%
80%	90.3%	1.185	0.553	21.75	+5.7%	+7.9%	+40.8%
90%	96.4%	1.188	0.555	21.99	+6.0%	+8.2%	+42.4%
100%	100.0%	1.190	0.556	22.15	+6.2%	+8.5%	+43.5%
Zinc and iron values are absorbed (TAZ via Miller equation, Fe at 5% absorption). Protein is PDCAAS-adjusted.

Scenarios Under Meta-Analysis (Gunaratna) Pathway

The Meta-analysis (Gunaratna) pathway estimates per-child stunting outcomes using the effect size from Gunaratna et al. (2010), which reports a 9% increase in height growth velocity for children consuming Quality Protein Maize relative to conventional maize. This effect is the common reference effect, modulated for each child by the protein-adequacy factor defined in Module 5, and applied to the ENCOVI children population produced by 05_06, with stunting outcomes derived from the resulting HAZ trajectories.

The Meta-analysis (Gunaratna) pathway functions as a methodological sensitivity check for the covariate prediction pathway. Both pathways express child growth on the same height-for-age z-score (HAZ) scale, but reach it by different routes: they use different populations (an ENCOVI-based synthetic population enriched with SIVESNU anthropometric donors, against the ENCOVI child population directly) and different stunting models (GAM-based vs effect-size-based), while sharing the same upstream farmer adoption mechanics, the same hierarchical priority assignment, and the same gravity-based redistribution. Comparing the two pathways across the same seed variants provides a range of plausible impact estimates under different methodological assumptions.

2022 Seed — Meta-Analysis

Table 7: National scenario progression — 2022 seed, Meta-analysis (Gunaratna) pathway, current subsidy

National Metrics by Production Coverage Level — Seed 2022. Pathway: Meta-analysis (Gunaratna) | Subsidy: current regime
Production Coverage	Farmers Bio	Coverage	Stunting (%)	Mean HAZ	Zinc Inad. (%)	Iron Inad. (%)
National Metrics by Production Coverage Level — Seed 2022
Pathway: Meta-analysis (Gunaratna) \| Subsidy: current regime
0%	0	0.0%	48.6	−1.976	53.6	54.2
10%	209,884	12.6%	47.6	−1.955	52.1	53.4
20%	391,415	22.3%	46.7	−1.940	50.9	52.8
30%	504,524	32.6%	46.0	−1.924	49.8	52.4
40%	582,959	42.7%	45.3	−1.909	48.7	51.9
50%	624,536	53.1%	44.6	−1.892	48.1	51.5
60%	663,209	63.1%	43.8	−1.876	47.4	51.2
70%	687,040	71.7%	43.5	−1.864	46.7	50.9
80%	741,392	78.1%	43.2	−1.855	46.3	50.7
90%	751,118	82.7%	43.1	−1.849	46.0	50.5
100%	783,474	85.5%	43.0	−1.845	45.9	50.3

Table 8: National nutrient intake progression — 2022 seed, Meta-analysis (Gunaratna) pathway, current subsidy

Nutrient Intake by Production Coverage Level — Seed 2022. Weighted mean bioavailable nutrients — Pathway: Meta-analysis (Gunaratna) | Subsidy: current regime
Production Coverage	Coverage	Absorbed Intake			Change vs Baseline
Nutrient Intake by Production Coverage Level — Seed 2022
Weighted mean bioavailable nutrients — Pathway: Meta-analysis (Gunaratna) \| Subsidy: current regime
Production Coverage	Coverage	Zinc (mg)	Iron (mg)	Protein (g)	Zn Δ%	Fe Δ%	Prot Δ%
0%	0.0%	1.120	0.511	15.43	0.0%	0.0%	0.0%
10%	12.6%	1.131	0.518	16.48	+0.9%	+1.3%	+6.8%
20%	22.3%	1.139	0.523	17.31	+1.7%	+2.4%	+12.2%
30%	32.6%	1.148	0.529	18.22	+2.4%	+3.5%	+18.1%
40%	42.7%	1.155	0.535	19.11	+3.1%	+4.7%	+23.9%
50%	53.1%	1.163	0.540	19.90	+3.8%	+5.7%	+29.0%
60%	63.1%	1.170	0.544	20.50	+4.4%	+6.4%	+32.9%
70%	71.7%	1.175	0.547	21.01	+4.9%	+7.1%	+36.2%
80%	78.1%	1.178	0.549	21.31	+5.2%	+7.4%	+38.1%
90%	82.7%	1.181	0.550	21.51	+5.4%	+7.7%	+39.5%
100%	85.5%	1.182	0.551	21.62	+5.5%	+7.8%	+40.2%
Zinc and iron values are absorbed (TAZ via Miller equation, Fe at 5% absorption). Protein is PDCAAS-adjusted.

2026 Seed — Meta-Analysis

Table 9: National scenario progression — 2026 seed, Meta-analysis (Gunaratna) pathway, current subsidy

National Metrics by Production Coverage Level — Seed 2026. Pathway: Meta-analysis (Gunaratna) | Subsidy: current regime
Production Coverage	Farmers Bio	Coverage	Stunting (%)	Mean HAZ	Zinc Inad. (%)	Iron Inad. (%)
National Metrics by Production Coverage Level — Seed 2026
Pathway: Meta-analysis (Gunaratna) \| Subsidy: current regime
0%	0	0.0%	48.6	−1.976	53.6	54.2
10%	207,403	13.9%	47.5	−1.953	52.0	53.3
20%	385,331	24.6%	46.5	−1.937	50.6	52.7
30%	500,704	35.6%	45.7	−1.920	49.4	52.3
40%	585,579	46.9%	45.1	−1.903	48.6	51.8
50%	633,774	57.4%	44.3	−1.885	47.9	51.5
60%	662,533	68.2%	43.6	−1.869	46.9	51.0
70%	698,760	77.3%	43.3	−1.856	46.3	50.7
80%	732,692	84.9%	43.0	−1.845	46.0	50.5
90%	749,656	90.3%	43.0	−1.839	45.8	50.2
100%	783,474	93.9%	42.9	−1.834	45.6	50.1

Table 10: National nutrient intake progression — 2026 seed, Meta-analysis (Gunaratna) pathway, current subsidy

Nutrient Intake by Production Coverage Level — Seed 2026. Weighted mean bioavailable nutrients — Pathway: Meta-analysis (Gunaratna) | Subsidy: current regime
Production Coverage	Coverage	Absorbed Intake			Change vs Baseline
Nutrient Intake by Production Coverage Level — Seed 2026
Weighted mean bioavailable nutrients — Pathway: Meta-analysis (Gunaratna) \| Subsidy: current regime
Production Coverage	Coverage	Zinc (mg)	Iron (mg)	Protein (g)	Zn Δ%	Fe Δ%	Prot Δ%
0%	0.0%	1.120	0.511	15.43	0.0%	0.0%	0.0%
10%	13.9%	1.132	0.518	16.59	+1.0%	+1.5%	+7.5%
20%	24.6%	1.141	0.524	17.50	+1.9%	+2.6%	+13.4%
30%	35.6%	1.150	0.531	18.47	+2.6%	+3.9%	+19.8%
40%	46.9%	1.158	0.537	19.39	+3.4%	+5.0%	+25.7%
50%	57.4%	1.166	0.541	20.13	+4.1%	+6.0%	+30.5%
60%	68.2%	1.173	0.546	20.80	+4.7%	+6.8%	+34.9%
70%	77.3%	1.178	0.549	21.28	+5.1%	+7.4%	+38.0%
80%	84.9%	1.182	0.551	21.61	+5.5%	+7.8%	+40.1%
90%	90.3%	1.185	0.552	21.79	+5.7%	+8.0%	+41.3%
100%	93.9%	1.186	0.553	21.91	+5.9%	+8.2%	+42.0%
Zinc and iron values are absorbed (TAZ via Miller equation, Fe at 5% absorption). Protein is PDCAAS-adjusted.

2028 Seed — Meta-Analysis

Table 11: National scenario progression — 2028 seed, Meta-analysis (Gunaratna) pathway, current subsidy

National Metrics by Production Coverage Level — Seed 2028. Pathway: Meta-analysis (Gunaratna) | Subsidy: current regime
Production Coverage	Farmers Bio	Coverage	Stunting (%)	Mean HAZ	Zinc Inad. (%)	Iron Inad. (%)
National Metrics by Production Coverage Level — Seed 2028
Pathway: Meta-analysis (Gunaratna) \| Subsidy: current regime
0%	0	0.0%	48.6	−1.976	53.6	54.2
10%	212,717	15.1%	47.4	−1.951	51.8	53.3
20%	387,123	27.2%	46.3	−1.933	50.2	52.5
30%	500,057	39.3%	45.6	−1.914	49.1	52.1
40%	586,628	51.3%	44.6	−1.895	48.1	51.6
50%	633,641	62.8%	44.1	−1.877	47.4	51.2
60%	666,985	72.9%	43.5	−1.862	46.6	50.8
70%	703,083	82.4%	43.1	−1.849	46.0	50.5
80%	725,591	90.3%	43.0	−1.839	45.8	50.2
90%	751,475	96.4%	42.9	−1.831	45.5	49.9
100%	783,474	100.0%	42.8	−1.825	45.2	49.9

Table 12: National nutrient intake progression — 2028 seed, Meta-analysis (Gunaratna) pathway, current subsidy

Nutrient Intake by Production Coverage Level — Seed 2028. Weighted mean bioavailable nutrients — Pathway: Meta-analysis (Gunaratna) | Subsidy: current regime
Production Coverage	Coverage	Absorbed Intake			Change vs Baseline
Nutrient Intake by Production Coverage Level — Seed 2028
Weighted mean bioavailable nutrients — Pathway: Meta-analysis (Gunaratna) \| Subsidy: current regime
Production Coverage	Coverage	Zinc (mg)	Iron (mg)	Protein (g)	Zn Δ%	Fe Δ%	Prot Δ%
0%	0.0%	1.120	0.511	15.43	0.0%	0.0%	0.0%
10%	15.1%	1.133	0.519	16.70	+1.1%	+1.6%	+8.3%
20%	27.2%	1.143	0.526	17.73	+2.1%	+2.9%	+14.9%
30%	39.3%	1.153	0.533	18.78	+2.9%	+4.2%	+21.7%
40%	51.3%	1.161	0.538	19.70	+3.6%	+5.4%	+27.7%
50%	62.8%	1.170	0.543	20.45	+4.4%	+6.4%	+32.6%
60%	72.9%	1.176	0.547	21.08	+4.9%	+7.2%	+36.7%
70%	82.4%	1.181	0.550	21.48	+5.4%	+7.7%	+39.3%
80%	90.3%	1.185	0.552	21.79	+5.7%	+8.0%	+41.3%
90%	96.4%	1.188	0.553	21.98	+6.0%	+8.3%	+42.5%
100%	100.0%	1.189	0.554	22.09	+6.1%	+8.4%	+43.2%
Zinc and iron values are absorbed (TAZ via Miller equation, Fe at 5% absorption). Protein is PDCAAS-adjusted.

Economic Progression

The economic dimension is invariant to the stunting pathway: the same farmers adopt the same land under the same conversion curves regardless of whether the downstream stunting impact is estimated via covariate prediction or meta-analysis. The economic block therefore consolidates the full grid of nine combinations — three seed variants (2022, 2026 and 2028 seed) and three subsidy regimes (current, 50% reduction, 75% reduction) — drawing from the covariate prediction pathway outputs.

The subsidy regime affects the seed cost structure faced by farmers. Under the current regime the per-area subsidy is funded entirely by New Seed, at 242 Q/mz of direct commitment, with no producer subsidy absorption and no extra cost to the farmer. Under the reduced regimes the reduction is split equally between producer subsidy absorption, through improved producibility, and the farmer, through a higher biofortified seed price. The complete decomposition by actor is documented in Module 2.

The seed variant affects yield change factors per farmer segment. The 2022 variety reflects current field performance; the 2026 and 2028 varieties reflect expected improvements from ongoing breeding pipelines. Higher yield change factors translate into higher per-farmer income increments and larger production volumes at the same production coverage level.

2022 Seed

Table 13: Economic impact progression — 2022 seed, current subsidy regime

Economic Impact by Production Coverage Level — Seed 2022. Farmer adoption, land use, and economic outcomes — Subsidy: Current
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2022
Farmer adoption, land use, and economic outcomes — Subsidy: Current
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	209,884	101,218	108,666	115,703	8,302,994	346.77
20%	391,415	112,071	279,343	231,266	14,756,727	612.02
30%	504,524	122,404	382,119	346,965	21,526,970	883.89
40%	582,959	131,972	450,987	462,510	28,232,333	1,151.46
50%	624,536	146,986	477,550	578,252	35,107,368	1,422.31
60%	663,209	163,033	500,176	693,606	41,689,775	1,690.35
70%	687,040	174,186	512,854	809,323	47,398,942	1,897.39
80%	741,392	188,662	552,729	925,076	51,615,154	1,946.96
90%	751,118	188,662	562,455	1,040,684	54,715,113	1,718.97
100%	783,474	188,662	594,812	1,156,213	56,504,051	1,544.18
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

Table 14: Economic impact progression — 2022 seed, 50% subsidy reduction

Economic Impact by Production Coverage Level — Seed 2022. Farmer adoption, land use, and economic outcomes — Subsidy: 50% reduction
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2022
Farmer adoption, land use, and economic outcomes — Subsidy: 50% reduction
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	212,838	102,208	110,630	115,619	8,287,577	336.54
20%	394,682	113,168	281,514	231,248	14,837,395	595.66
30%	507,968	125,566	382,402	346,871	21,705,699	859.32
40%	580,629	134,114	446,515	462,484	28,631,181	1,129.03
50%	620,065	149,569	470,496	578,116	35,928,800	1,416.90
60%	658,961	166,746	492,215	693,684	42,915,642	1,675.95
70%	689,400	184,128	505,271	809,199	48,742,448	1,855.70
80%	736,697	188,662	548,035	924,791	52,160,216	1,811.54
90%	757,104	188,662	568,441	1,040,352	54,793,550	1,596.84
100%	783,474	188,662	594,812	1,156,213	56,504,051	1,443.42
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

Table 15: Economic impact progression — 2022 seed, 75% subsidy reduction

Economic Impact by Production Coverage Level — Seed 2022. Farmer adoption, land use, and economic outcomes — Subsidy: 75% reduction
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2022
Farmer adoption, land use, and economic outcomes — Subsidy: 75% reduction
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	215,199	105,125	110,074	115,663	8,293,102	330.68
20%	396,644	115,316	281,328	231,252	14,852,304	585.19
30%	507,849	126,167	381,682	346,782	21,803,950	848.16
40%	579,854	135,687	444,167	462,472	28,809,265	1,116.64
50%	619,261	151,011	468,250	578,073	36,213,744	1,406.55
60%	660,522	171,484	489,038	693,787	43,367,747	1,656.04
70%	698,276	186,935	511,342	809,135	49,013,000	1,824.25
80%	733,222	188,662	544,560	925,058	52,497,070	1,718.87
90%	759,510	188,662	570,848	1,041,041	54,840,838	1,532.74
100%	783,474	188,662	594,812	1,156,213	56,504,051	1,393.87
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

2026 Seed

Table 16: Economic impact progression — 2026 seed, current subsidy regime

Economic Impact by Production Coverage Level — Seed 2026. Farmer adoption, land use, and economic outcomes — Subsidy: Current
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2026
Farmer adoption, land use, and economic outcomes — Subsidy: Current
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	207,403	117,099	90,304	115,665	9,192,098	468.49
20%	385,331	127,156	258,174	231,145	16,253,570	857.64
30%	500,704	134,334	366,370	346,957	23,561,847	1,244.18
40%	585,579	141,480	444,099	462,466	30,996,885	1,654.53
50%	633,774	150,967	482,807	578,122	37,926,920	2,023.81
60%	662,533	157,734	504,799	693,817	45,102,351	2,433.34
70%	698,760	170,623	528,137	809,405	51,142,437	2,738.51
80%	732,692	187,008	545,685	925,020	56,138,458	2,953.91
90%	749,656	188,662	560,994	1,040,856	59,701,387	2,875.90
100%	783,474	188,662	594,812	1,156,212	62,103,890	2,623.58
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

Table 17: Economic impact progression — 2026 seed, 50% subsidy reduction

Economic Impact by Production Coverage Level — Seed 2026. Farmer adoption, land use, and economic outcomes — Subsidy: 50% reduction
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2026
Farmer adoption, land use, and economic outcomes — Subsidy: 50% reduction
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	208,247	115,187	93,060	115,612	9,221,236	465.39
20%	389,943	126,105	263,838	231,189	16,303,607	843.57
30%	506,262	134,454	371,808	347,012	23,766,276	1,229.77
40%	590,319	143,505	446,815	462,459	31,223,713	1,630.43
50%	630,995	151,828	479,167	578,073	38,609,341	2,017.18
60%	666,369	162,247	504,122	693,760	45,853,000	2,416.19
70%	696,929	174,062	522,867	809,383	52,057,770	2,717.08
80%	740,067	188,662	551,405	924,974	56,712,483	2,864.81
90%	743,932	188,662	555,269	1,040,290	60,270,870	2,696.10
100%	783,474	188,662	594,812	1,156,213	62,103,890	2,522.82
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

Table 18: Economic impact progression — 2026 seed, 75% subsidy reduction

Economic Impact by Production Coverage Level — Seed 2026. Farmer adoption, land use, and economic outcomes — Subsidy: 75% reduction
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2026
Farmer adoption, land use, and economic outcomes — Subsidy: 75% reduction
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	209,541	113,870	95,672	115,598	9,257,096	466.47
20%	391,417	125,049	266,368	231,296	16,307,262	835.99
30%	507,161	133,902	373,260	346,813	23,811,492	1,219.81
40%	589,908	143,520	446,388	462,461	31,307,814	1,616.29
50%	630,068	152,736	477,331	578,148	39,003,880	2,018.08
60%	668,955	165,544	503,411	693,670	46,304,340	2,404.85
70%	695,433	174,656	520,777	809,295	52,422,563	2,705.83
80%	737,910	188,662	549,248	924,885	56,954,628	2,814.05
90%	747,432	188,662	558,770	1,040,606	60,357,292	2,629.55
100%	783,474	188,662	594,812	1,156,213	62,103,890	2,473.27
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

2028 Seed

Table 19: Economic impact progression — 2028 seed, current subsidy regime

Economic Impact by Production Coverage Level — Seed 2028. Farmer adoption, land use, and economic outcomes — Subsidy: Current
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2028
Farmer adoption, land use, and economic outcomes — Subsidy: Current
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	212,717	147,894	64,823	115,551	9,997,239	538.38
20%	387,123	151,998	235,125	231,255	18,014,197	1,057.73
30%	500,057	154,758	345,299	346,886	25,999,726	1,549.28
40%	586,628	156,912	429,716	462,464	33,889,573	2,062.70
50%	633,641	159,711	473,930	578,181	41,494,605	2,576.68
60%	666,985	164,192	502,794	693,681	48,192,734	3,020.48
70%	703,083	172,002	531,081	809,555	54,510,028	3,409.98
80%	725,591	178,060	547,531	924,901	59,708,063	3,720.39
90%	751,475	188,563	562,912	1,040,472	63,759,307	3,720.26
100%	783,474	188,662	594,812	1,156,212	66,544,562	3,479.42
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

Table 20: Economic impact progression — 2028 seed, 50% subsidy reduction

Economic Impact by Production Coverage Level — Seed 2028. Farmer adoption, land use, and economic outcomes — Subsidy: 50% reduction
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2028
Farmer adoption, land use, and economic outcomes — Subsidy: 50% reduction
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	212,015	144,078	67,937	115,623	10,028,482	535.85
20%	387,096	148,710	238,386	231,299	18,019,883	1,042.14
30%	500,994	152,493	348,502	346,867	26,029,458	1,524.19
40%	586,895	155,683	431,212	462,472	34,013,302	2,033.15
50%	633,482	159,469	474,013	578,182	41,657,590	2,540.62
60%	666,828	165,873	500,955	693,775	49,069,616	3,022.66
70%	698,553	172,012	526,541	809,316	55,442,726	3,407.83
80%	742,251	188,270	553,981	925,066	60,320,500	3,646.74
90%	749,473	188,563	560,910	1,040,575	64,077,790	3,602.91
100%	783,474	188,662	594,812	1,156,212	66,544,562	3,378.67
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

Table 21: Economic impact progression — 2028 seed, 75% subsidy reduction

Economic Impact by Production Coverage Level — Seed 2028. Farmer adoption, land use, and economic outcomes — Subsidy: 75% reduction
Production Coverage	Adopting Farmers			Ha	Prod Bio (qq)	Net Income Change (M GTQ)
Economic Impact by Production Coverage Level — Seed 2028
Farmer adoption, land use, and economic outcomes — Subsidy: 75% reduction
Production Coverage	Total	Hybrid	OPV	Ha	Prod Bio (qq)	Net Income Change (M GTQ)
0%	0	0	0	0	0	0.00
10%	213,571	143,410	70,161	115,623	10,004,836	531.75
20%	387,210	148,238	238,972	231,021	18,002,499	1,032.41
30%	501,855	152,313	349,542	346,870	26,055,404	1,512.77
40%	589,640	155,899	433,741	462,478	34,037,767	2,015.79
50%	632,008	159,585	472,422	578,143	41,964,838	2,537.92
60%	668,000	166,657	501,342	693,791	49,347,074	3,010.16
70%	700,764	174,246	526,519	809,300	55,681,683	3,385.35
80%	742,975	188,563	554,412	925,078	60,560,749	3,605.08
90%	746,643	188,563	558,080	1,040,686	64,411,399	3,516.40
100%	783,474	188,662	594,812	1,156,213	66,544,562	3,329.11
Net Income Change is the change in net income, with the seed cost difference included. Land converted at 0.7 ha/mz.

Key Outputs

The pre-computed scenario datasets carry three families of variables.

Economic Variables

Variable	Description	Unit
`n_farmers_bio`	Adopting farmers (survey-weighted, conversion ratio applied)	count
`land_bio_ha`	Biofortified land area	hectares
`production_biofortified_total`	Total biofortified production	quintales
`income_increment_total`	Additional farmer income	Q
`cost_seeds_increment_total`	Change in seed costs	Q
`income_increment_mgtq`	Net income change, seed cost difference included	M GTQ

Nutritional Variables

Variable	Description	Unit
`mean_zinc`	Mean absorbed zinc (TAZ via Miller equation)	mg/day
`mean_iron`	Mean absorbed iron (5% absorption)	mg/day
`mean_protein`	Mean PDCAAS-adjusted protein	g/day
`*_inad_pct`	Nutrient inadequacy prevalence	%
`*_change_pct`	Change vs baseline	%

Stunting Variables

Variable	Description	Unit
`stunting_pct`	Stunting prevalence	%
`mean_zlen`	Mean height-for-age z-score (HAZ)	SD units

Departmental vs National: Key Differences

Aspect	06_02 Departmental	06_03 National
Rows	242 (22 depts × 11 levels)	11 (1 region × 11 levels)
Identifier	`departamento`	`region = "Guatemala Total"`
Quantiles	Included (distribution boxplots)	Excluded (require individual data)
Use	Geographic filtering, departmental comparisons	Overview tab, national indicators
Aggregation	`.by = departamento`	No grouping (national weighted mean)

Both datasets share identical variable schemas for compatibility with downstream Shiny processing.

Methodological Assumptions

Adoption Mechanism

The continuous adoption mechanism assumes that farmers with higher adoption scores (better technology fit, larger scale, prior hybrid experience) enter the adoption pool earlier along the national adoption ranking, and therefore at lower production coverage levels. Within each farmer, the conversion intensity follows the profile-specific sigmoid calibrated against empirical adoption anchors in Module 2.

Consumption Coverage Assignment

The hierarchical assignment assumes that farmer households prioritise self-consumption of their own production, that local markets receive biofortified maize before commercial distribution, that urban access depends on formal supply chain reach, and that within each priority group access is random (no further stratification).

Stunting Response

The covariate prediction pathway assumes that the GAM-based relationship between nutritional index and HAZ fitted on the synthetic population generalises to the scenario simulation, and that estimated effects are heterogeneous (stronger marginal benefits for nutritionally deficient children). The Meta-analysis (Gunaratna) pathway assumes that the Gunaratna et al. (2010) effect size on height growth velocity is transportable to the Guatemalan ENCOVI children population. Both scenario applications are model-based ex-ante projections and report point estimates without propagated uncertainty. The covariate prediction pathway derives its response from observational survey data; the meta-analysis pathway imports its reference effect from randomized trial evidence.

Limitations

Key Limitations

No household-level linkage — Children cannot be directly linked to specific adopting farmer households; assignment is based on occupational proxies.
Uniform biofortification factors — All biofortified maize is assumed to have identical nutrient increases regardless of variety, soil conditions, or processing.
No uncertainty quantification — Point estimates only; confidence intervals would require bootstrap or Monte Carlo simulation.
No supply chain modelling — Distribution is assumed to follow the priority hierarchy and the gravity model without modelling transport costs, market access, or price differentials.
Cross-sectional stunting model — Predictions represent point-in-time associations, not longitudinal growth trajectories.

Summary

The scenario precomputation step generates the analytical backbone for the Shiny dashboard. For each of the 18 combinations in the seed × subsidy × pathway grid, it produces departmental (242 rows) and national (11 rows) scenarios with consistent variable schemas.

Key methodological elements

Continuous adoption — Per-farmer conversion intensity along profile-specific sigmoid curves, evaluated at the internal adoption index that reproduces each production coverage level, produces smooth saturation trajectories.
Pre-computed individual profiles — Module 6 receives baseline and QPM profiles from Modules 4 and 5 and is responsible only for assignment and aggregation, not for the underlying nutritional and stunting modelling.
Hierarchical priority assignment — Realistic biofortification distribution based on household occupation and geography, calibrated to departmental consumption coverage from the gravity-based redistribution.
Dual pathway — Two parallel stunting methodologies (covariate prediction and meta-analysis) provide a range of plausible impact estimates under different assumptions.
Survey-weighted aggregation — All metrics use calibrated weights for population-representative estimates.

These pre-computed scenarios enable instant dashboard response while maintaining full methodological transparency. The 18 combinations are exported as parquet files for direct consumption by the Shiny dashboard, which is the product of results layered on top of this methodological backbone.