Try the Demo

Scientific model version

The model version is stored with the experiment and determines which engine executes it. Results of different versions are never mixed or re-interpreted.

v2.0.0 · current v1.0.0 · archived

Archived model: memoryless, myopic decisions; budget-cap scenarios only. Use it to reproduce or extend historical v1 records.

Experiment

Name and describe the experiment. The complete configuration is stored with the results.

Experiment name

Random seed

Description (optional)

Experimental design

Full factorial design: every population size × budget level × sensitivity level is a configuration cell, each simulated for all selected mechanisms with identical synthetic farmers and random draws.

Population sizes N

Comma-separated, 2–150 farmers. Protocol: 100, 250, 500.

Horizon T (periods)

Repetitions per cell

Protocol: 100 independent repetitions.

Budget scenarios (MU per farmer over the horizon)

MU / farmer

no cap — expenditure = sum of entitlements (frontier protocol)

Total budget cap = value × N. A cap is a limit, not an expenditure level: actual spending is reported separately.

Incentive sensitivity (reward coefficient β_R)

β_R =

Medium β_R = β_C (1 MU expected reward offsets 1 MU cost); Low/High = ×0.5 / ×2.

Incentive mechanisms

S0 is the mandatory control: improvement and cost-effectiveness metrics are paired against it within each repetition.

S0 · Control No financial incentive (always included)

S1 · Fixed Flat reward for verified adopters

S2 · Performance Proportional to verified improvement

S3 · Tiered Rewards by performance thresholds

S3R · Tiered, relative v2 variant: tiers on gain over own baseline

Model v2 protocol & behaviour

All values are hypothetical modelling assumptions (see the parameter dictionary). Setting η = 0, H = 1 and k0 = 0 reproduces the memoryless, myopic v1 decision rule as a sensitivity case.

Protocol

Sustained participation: k consecutive seasons

η persistence (habit / switching cost)

utility bonus for continuing; 0 = memoryless

H expectation horizon (seasons)

1 = myopic (v1)

φ expectation discount

weight per future season

π(0) initial payout expectation

1 = announced rewards believed

k0 prior seasons of initial participants

0 = start at baseline (v1)

h heterogeneity scale

× SD of α, B, C, s

r cost–baseline correlation

Gaussian copula

Reward-scale grid λ (Frontier)

Strictly increasing; λ = 0 is the control. The sensitivity protocol uses its own grid from the design.

Matched expenditure targets

MU per farmer over the horizon

Primary target

LHS design points K

LHS design seed

Factor ranges are taken from the preset (11 factors; see the parameter dictionary). Custom ranges can be supplied by importing a configuration JSON.

Defaults are documented modelling assumptions (see parameter table). They were not calibrated to observed data or tuned to favour any mechanism.

Behavioural coefficients

β_B benefit

utility per index unit

β_C cost

utility per M

ρ experience effect

benefit gain after adopting

M monetary reference

MU used for normalisation

Synthetic population SYNTHETIC

Initial adoption rate

share of adopters at t = 0

Attribute (truncated normal)	Mean	SD	Min	Max
Baseline propensity α logit
Perceived benefit B index 0–1
Adoption cost C MU / period
Incentive sensitivity s multiplier
Baseline performance q0 index 0–1
Gain potential g (qmax = q0 + g) index units

Performance dynamics & verification

γ improvement rate

gap to potential closed per adopting period

δ decay rate

gain lost per non-adopting period

σ_v verification noise

SD of measurement error

Budget allocation rule

When entitlements exceed the cap, all are scaled down pro rata.

Mechanism settings

S1 fixed reward F

MU

S1 minimum q̂

S2 rate k

MU / unit

S2 cap

MU

S3 tiers (threshold on verified performance → reward in MU per period; thresholds strictly increasing)

→

Demonstration presets

Behavioural coefficients

Synthetic population SYNTHETIC

Performance dynamics & verification

Mechanism settings

Run experiment