MRIO Comparison | Confidence

The figures on this page show the consumption-based CO₂ emissions account (CBA; also called Carbon Footprint) of each country, with associated 2 standard deviation uncertainty bands as assessed using different regimes to estimate model reliability.The confidence intervals were established using a Monte Carlo technique, and repeatedly perturbing the MRIO transactions matrix using small random perturbations and calculating the countrywise CBA inventories from that disturbed IO table.

We used several different regimes for estimating the reliability of elements in the MRIO:

In Scenario 1 each the perturbation of each element was ±10% of the transaction's value.
In Scenario 2 elements large values were assumed to be more reliable and hence perturbed less (<1%) while small values were assumed to be less reliable and were perturbed more, up to 2000%
Scenario 3 used the same reliability regime as Scenario 2, but in Scenario 3 we normalized the territorial emissions inventory across models so that the only source of difference between models should be the description of the global economic structure and the sectorwise allocation of emissions, but not the total level of emissions.
Scenario 4 is similar to Scenario 3, except all CO₂ emissions were allocated exclusively to the Electricity Generation sector in each model. This removes sectorwise emissions allocation from the sources of discrepancy.
Scenario 5 is similar to Scenario 1, except transactions in the Final Demand block were allowed to fluctuate ±30%.

This work is described in full in our paper Convergence Between the Eora, WIOD, EXIOBAS, and OpenEU's Consumption-Based Carbon Accounts.

Complete numerical results are available in the Supplementary Information file for that paper.

We welcome everyone to use the charts and results to showcase the reliability/convergence of MRIO accounts for individual countries.

Daniel Moran and Richard Wood, August 2014