A particular ceiling on atmospheric CO₂ concentrations can be maintained through a variety of emission pathways. Over the past decade, there has been considerable debate over the characteristics of a least-cost pathway. Some have suggested that a gradual departure from the emissions baseline will be the most cost-effective because it reduces the pressure for premature retirement of the existing capital stock, and it provides valuable time to develop low-cost, low-carbon emitting substitutes. Others counter that a major flaw in analyses that support this line of reasoning is that they ignore learning-by-doing (LBD).

In this paper, we examine the impact of LBD on the timing and costs of emissions abatement. With regard to timing, we find that including learning-by-doing does not significantly alter the conclusions of previous studies that treated technology cost as exogenous. The analysis supports the earlier conclusion that for a wide range of stabilization ceilings, a gradual transition away from the ?no policy? emissions baseline is preferable to one that requires substantial near-term reductions. We find that the major impact of including learning-by-doing is on the costs of emission abatement. Depending upon the sensitivity of costs to cumulative experience, LBD can substantially reduce the overall costs of emissions abatement.