A learning by doing multiplier accelerates the transition to photovoltaic cells

Abstract

Abating emissions of carbon dioxide depends in part on how quickly the levelized cost of electricity (LCOE) from photovoltaic cells (PV) achieves grid parity without policy interventions. Reaching this threshold is accelerated by learning by doing, which reduces the LCOE generated by PV. The resultant cost reduction generates a positive feedback loop and increases the demand for PV; but previous analyses ignore this feedback and therefore overlook a critical learning by doing multiplier effect, in which increasing the cumulative production of PV modules lowers their price, and lower PV module prices increase purchases of PV modules, which increases cumulative production of PV modules, and so on. We quantify the learning by doing multiplier effect with a cointegrating vector autoregression (CVAR) model that captures the simultaneous relation between the price for and cumulative production of PV modules. The learning by doing multiplier effect amplifies the static effects of learning by nearly a factor of ten and eliminates the simultaneous equation bias in previous estimates of unidirectional learning curves. This multiplier effect enhances the ability of policies, such as a carbon tax, to lower the costs of PV, increase cumulative production, and lower carbon emissions. Together, these results suggest that grid parity is closer than indicated by unidirectional learning curves.