Coping with Uncertainties in Technological Learning

To date, optimization models of endogenous technological change commonly deal with uncertainty in technological learning with risk-factor methods, i.e., by adding expected risk costs resulting from overestimating technological learning rates into an objective function with a subjective risk factor. This paper argues that another way of coping with the uncertainty, risk-constrained methods that have been ignored in existing literatures, could be more practicable (at least as a supplement) for decision support. With a simplified model, this paper explores technology development paths generated by two risk-constrained methods, and compares the two risk-constrained methods with a risk-factor method. Our study shows that comparing with the risk-factor method, the two risk-constrained methods also generate an S-shaped technology diffusion pattern, which accords with historical observations, and they can result in earlier as well as later adoption of an advanced but currently expensive technology, depending on different combinations of uncertainty levels of the technology learning rate and the upper limit on the expected risk cost. Another finding of our research is that two totally different technology development paths can both be optimal solutions, which implies that with early policy interventions there is the possibility that an economy could be led to a low-carbon economy with little additional cost.