In terms of cost and performance, infant technologies, such as solar photovoltaics (PV), are normally inferior to entrenched technologies. It is a Catch-22 situation since the diffusion on larger markets that would be needed to reduce cost is hindered by the high cost. Therefore it would make sense to subsidise PV to increase sales, which would increase experience and induce investments in larger factories, which in turn would drive down costs and the subsidies needed. The total costs of such a scheme does not have to be prohibitive if cost reductions with increased volumes are large enough. Over the last 20 years the cost of PV modules was reduced by 18-23% per doubling of cumulative production (a progress ratio of 0.77-0.82). For a progress ratio of 0.80 and an annual growth rate of 30%, the modelled annual subsidy peaks at $14 US billion, which corresponds to an additional electricity tax of no more than 0.1 US cents/kW h in OECD countries. A market support programme also creates institutional learning and increases the political power of the proponents of PV. The current federal German support programme is a product of learning and network formation in earlier market stimulation and research, development and demonstration (RDD) programmes of smaller scale. Hence, the current support programme is now likely to create not only economic virtuous circles that reduce costs, but also institutional virtuous circles that work for the survival and expansion of the programme itself. As the PV industry grows, care should be taken to maintain variety to reduce the risk of a premature lock-in of an inferior design. To maintain variety in the market place may prove costly when the market grows but variety creation at the level of RDD investments is fairly cheap. To increase the world expenditure on RDD of renewable energy technology by a factor of 10 would not cost more than $1 US/ton C or 0.02 US cent/kW h of electricity. (C) 2004 Elsevier Ltd. All rights reserved.