This model is a structural snapshot of 2018–2021. It is not a present-day estimate.See limitations

Shock simulator

Move six external shock channels and the model computes how far each of the twelve islands falls. The damage function is calibrated on what actually happened in 2020, with only two free parameters.

Warning: First-round only: the loss does not feed back into the composite, and there is no time index. The GDP-loss figure is the calibrated output (two free parameters, fitted on the 2020 event, leave-one-out error 4.24pp on n=10). The composite delta and the most-vulnerable-sector call are an explanatory decomposition and are NOT independently calibrated. Islands with missing exposure channels are understated - check missing_channels.

Shock settings

The like-for-like comparison scenario required by the model spec.

Named scenarios
Custom
Free adjustment0 = no change
+30
% fall in tourism receipts
+30
% change in fuel import cost
+30
% change in food import cost
+30
% tightening of external finance
+30
% rise in freight cost / lead time
+30
cyclone / sea-level stress index
Why the slider ranges are what they are · how each channel's exposure is defined

The ranges are not arbitrary. The minimum and maximum on each economic channel is the historically observed range for that channel; the physical-climate channel follows a published IPCC pathway. The API rejects anything outside them with a 422. Several channels have a negative minimum because a shock can move favourably — a fall in fuel prices is a gain for an importer, not a loss, and that is exactly what happened in 2020.

Tourism demand [-20, 100] % fall in tourism receipts
Tourism receipts as a share of total exports, used directly rather than normalised: an 85% share means 85% of export earnings are at risk.
Fuel price [-60, 150] % change in fuel import cost
Share of final energy that is NOT modern renewable, i.e. 100 - modern_renew. Traditional biomass does not count as energy independence.
Food price [-30, 120] % change in food import cost
Low domestic primary production (MVI AFF, inverted) plus weak fishery (EPI Fish Stock Status, inverted).
External capital [-30, 100] % tightening of external finance
Reliance on externally sourced money: external debt, remittances and FDI.
Freight & logistics [-20, 120] % rise in freight cost / lead time
Remoteness (MVI REM) times trade openness — how much of the economy has to cross an ocean.
Physical climate [0, 100] cyclone / sea-level stress index
Direct climate exposure: low-elevation coastal population, disaster victims, and ND-GAIN vulnerability.
Model parameters

κ = 0.20 · scale = 0.3512

Contrast case: The simulator runs on the twelve-island panel only. Jeju is a contrast case that is never ranked, and 22 of the 28 indicators are undefined at sub-national level, so no exposure vector can be built for it — which is why it does not appear here.

GDP loss under this shock

This is the calibrated output. The two columns below are different units, so each carries its own scale: a 10% hit means something entirely different in Singapore than in Haiti.

Worst loss
37.9%
Maldives96.6%
Panel median
29.2%
n = 12
Worst − least
13.4 pp
GDP loss %0 – 37.9%
Loss per capita0 – US$17,525
#IslandGDP loss % scale max 37.9Loss per capita scale max US$17,525Data confidenceShow breakdown
  • 1MaldivesIndian Ocean (AIS)37.9US$4,15296.6%
  • 2VanuatuPacific33.6US$1,030100.0%
  • 3SeychellesIndian Ocean (AIS)32.2US$4,11993.1%
  • 4Papua New GuineaPacific30.4US$83396.6%
  • 5FijiPacific29.7US$1,788100.0%
  • 6JamaicaCaribbean29.2US$1,51989.7%

    This loss is UNDERSTATED: the island has no exposure data for the following channel, so that channel contributes zero — Tourism demand That channel is being shocked at the current setting.

  • 7MauritiusIndian Ocean (AIS)29.2US$2,955100.0%
  • 8Cabo VerdeAtlantic (AIS)28.6US$1,029100.0%
  • 9TuvaluPacific28.6US$1,13379.3%

    This loss is UNDERSTATED: the island has no exposure data for the following channel, so that channel contributes zero — Tourism demand That channel is being shocked at the current setting.

  • 10SingaporeAsia (AIS)27.7US$17,52589.7%
  • 11BarbadosCaribbean25.9US$4,78289.7%
  • 12HaitiCaribbean24.5US$299100.0%
Table view
IslandRegionGDP loss %Loss per capitaComposite changeSector that fails firstData confidenceMissing channelsRecovery
MaldivesMDVIndian Ocean (AIS)37.88US$4,152−4.61Food & primary production security96.6%NoneRecovered in 2–3 years
VanuatuVUTPacific33.59US$1,030−4.34Economic diversification & trade100.0%NoneNot recovered by 2024
SeychellesSYCIndian Ocean (AIS)32.18US$4,119−5.37Food & primary production security93.1%NoneNot recovered by 2024
Papua New GuineaPNGPacific30.38US$833−4.72Food & primary production security96.6%NoneNot recovered by 2024
FijiFJIPacific29.66US$1,788−5.12Energy security & transition100.0%NoneRecovered in 2–3 years
JamaicaJAMCaribbean29.24US$1,519−4.99Food & primary production security89.7%Tourism demandRecovered in 2–3 years
MauritiusMUSIndian Ocean (AIS)29.18US$2,955−5.60Food & primary production security100.0%NoneRecovered in 2–3 years
Cabo VerdeCPVAtlantic (AIS)28.63US$1,029−4.52Energy security & transition100.0%NoneRecovered in 2–3 years
TuvaluTUVPacific28.55US$1,133−3.02Energy security & transition79.3%Tourism demandRecovered within 1 year
SingaporeSGPAsia (AIS)27.73US$17,525−6.07Fiscal & financial capacity89.7%NoneRecovered within 1 year
BarbadosBRBCaribbean25.94US$4,782−5.43Food & primary production security89.7%NoneRecovered in 2–3 years
HaitiHTICaribbean24.53US$299−3.33Energy security & transition100.0%NoneNot recovered by 2024

The sector that fails first

The current setting beside the reference scenario (all channels at 30%). Where the two columns disagree, that island's weak point moves with the kind of shock it faces.

0 of the twelve islands fail in a different sector than they do under the reference scenario. · explanatory decomposition · not calibrated

IslandCurrent settingReference scenario
MaldivesFood & primary production security−10.96Food & primary production security−10.96
VanuatuEconomic diversification & trade−8.55Economic diversification & trade−8.55
SeychellesFood & primary production security−13.30Food & primary production security−13.30
Papua New GuineaFood & primary production security−8.46Food & primary production security−8.46
FijiEnergy security & transition−8.71Energy security & transition−8.71
JamaicaFood & primary production security−10.77Food & primary production security−10.77
MauritiusFood & primary production security−11.79Food & primary production security−11.79
Cabo VerdeEnergy security & transition−10.51Energy security & transition−10.51
TuvaluEnergy security & transition−10.08Energy security & transition−10.08
SingaporeFiscal & financial capacity−16.55Fiscal & financial capacity−16.55
BarbadosFood & primary production security−12.22Food & primary production security−12.22
HaitiEnergy security & transition−7.24Energy security & transition−7.24

M1 Food · M2 Energy · M3 Diversification · M4 Fiscal · M5 Digital · M6 Climate · M7 Ocean · M8 Human capital · M9 Dependence

2020 validation: predicted vs observed

The model held against what 2020 actually did. Only two parameters are free — a global scale and the damping exponent κ — and every exposure is fixed by the data.

Correlation r
0.816
fitted on n = 10
R-squared
0.666
Leave-one-out MAE
4.24pp
Rank MAE
1.4places
  • Used in the fit
  • Excluded from fit (hollow)
  • Perfect prediction (45°)
01020300102030Observed 2020 GDP loss (%)Model-predicted loss (%)MaldivesFijiBarbadosMauritiusCabo VerdeSeychellesJamaicaVanuatuSingaporePapua New GuineaHaitiTuvalu

A negative loss means the economy grew in 2020. Tuvalu is the one such case here.

Warning: Two islands are held out of the fit because tourism's share of exports is unavailable for them. Their predictions carry a zero tourism channel and are therefore understated. Jamaica · Tuvalu

The 2020 shock vector used for the fit
  • Tourism demand · % fall in tourism receipts+90
  • Fuel price · % change in fuel import cost−30
  • Food price · % change in food import cost+20
  • External capital · % tightening of external finance+25
  • Freight & logistics · % rise in freight cost / lead time+40
  • Physical climate · cyclone / sea-level stress index0

κ = 0.20 · scale = 0.3512

Fit note, verbatim from the API

Fitted on the 2020 COVID event, which is a shock VECTOR - fuel moved favourably for importers while tourism collapsed. Only two parameters are free (global scale, damping exponent kappa); all exposures are fixed by the data. Leave-one-out mean absolute error is 4.24 percentage points of GDP. Jamaica and Tuvalu are excluded from the fit because tourism share of exports is unavailable for them.

Table view

Error is predicted minus observed; a positive value means the model over-predicted the loss.

IslandPredicted %Observed %Error ppIn fit
MaldivesMDV24.2231.98−7.76Yes
FijiFJI16.2819.05−2.77Yes
BarbadosBRB11.9317.61−5.68Yes
MauritiusMUS14.5114.87−0.36Yes
Cabo VerdeCPV18.5514.78+3.77Yes
SeychellesSYC11.2610.72+0.54Yes
JamaicaJAM6.1910.20−4.01No
VanuatuVUT19.499.24+10.25Yes
SingaporeSGP7.675.39+2.28Yes
Papua New GuineaPNG6.053.88+2.17Yes
HaitiHTI3.353.37−0.02Yes
TuvaluTUV3.30-4.40+7.70No