The Amazon rainforest is the world's largest rainforest, and thus can be considered a global public good. Currently, at least 16 percent of the area has been lost to deforestation, and without new preservation plans 40 percent of the area is expected to be deforested by 2050. Avoiding deforestation and the resulting loss of biodiversity and ecosystem services provide benefits to both local households and households worldwide. The latter benefits seems to account for the majority of the total global benefits, because the number of affected households these benefits are aggregated over is very large. As it is very time consuming and costly to assess these global non-use values in stated preference surveys in all countries worldwide, benefit transfer exercises and expert assessment in Delphi Contingent Valuation (CV) surveys have been conducted. We test the reliability of these two approaches for predicting distant beneficiaries´ willingness-to-pay (WTP) for Amazon Rainforest preservation plans by comparing these estimates to a new CV survey of 300 Norwegian households. The CV survey found a mean WTP of 110 € (NOK 1100) per household per year to avoid further forest and biodiversity loss (which was the most ambitious of two alternative preservation plans).

Whereas benefit transfer from a North American CV survey of the same scenarios, both in terms of unit transfer with income adjustment and value function transfer, resulted in transfer errors of up to several hundred percent; the Norwegian experts in the Delphi CV survey predicted the outcome of the population CV survey well. Transfer errors were as low as 12 % in one model, and in all models below 35 %. Thus, this study provides evidence that the Delphi CV method could be a valid, as well as very time and cost effective, technique for assessing benefits of global public goods to distant beneficiaries.