With greater certainty in anthropogenic influence on observed changes in climate there is pressure for agreements to abate emissions of greenhouse gases. These agreements are often formulated in terms of emission targets that need to be met according to a timetable. In ICAM (Integrated Climate change Assessment Model) we have developed a realistic representation of the stochastic features of and uncertainties in natural systems and socio-economic activities. This representation is then used to evaluate the relative strengths and weaknesses of different climate policy frameworks.

In this simulation environment target frameworks are specified and artificial adaptive agents used to prospectively meet prior agreements on abatement targets. Unlike optimization frameworks, these targets are not specified as constraints, and if the prospective search towards the target is poorly executed by the agent, the objective is not met. The uncertainties and stochastic nature of the natural and human systems make prospective policy tuning difficult and targets are rarely met. Hence, the traditional measure of a "better policy" being the least cost strategy to the target is not applicable for two reasons:

(i) the target is often met in an approximate fashion;
(ii) uncertainties affect the costs of different paths towards the target.

Therefore, a new paradigm of "better policy" is being developed to differentiate between alternative frameworks for definition of climate change mitigation objectives.

In an initial search, I have defined three objectives for good policy design reflecting my understanding of the political economy policy design. These objectives are that the policy should:

(i) offer smooth and predictable signals to decision-makers ­ so that their investments have a consistent trend towards the overall objectives of reducing GHG emissions and vulnerability to climate fluctuations/change;
(ii) be robust against uncertainties ­ so that along the way as we change our target (a likely outcome given current uncertainties) we make small revisions to the signals used to shape investments by decision-makers; and,
(iii) give high assurance of meeting the stated target, so that given the uncertainties in our understanding and control over natural and human systems we still manage to meet stated prior objectives with an accuracy that will enhance public confidence in the political system.

In devising climate policy we are interested in reducing the impacts of climate change. Emission levels, atmospheric concentrations of GHGs, and global average temperature change are examples of proxy measures often used to indicate the potential level of impacts. However, none are representative of actual impacts. Responding to the impacts is likely to occur, but in terms of revisions to previously stated targets in terms of these proxy measures. Therefore, a key question is which of these or other proxy measures of climate change offer us the best opportunity of meeting the three objective of good policy outlined above?