Learning dependent subsidies for lithium-ion electric vehicle batteries Schuyler Matteson, Eric Williams ⁎ Golisano Institute for Sustainability, Rochester Institute of Technology, Rochester, NY, USA article info abstract Article history: Received 19 November 2013 Received in revised form 13 December 2014 Accepted 26 December 2014 Available online 9 February 2015 Governments subsidize diffusion of a variety of energy technologies believed to provide social benefits. These subsidies are often based on the idea that stimulating learning and industry development will lower costs to make the technology competitive, after which point the subsidy can be removed. We investigate two questions related to the design of subsidy programs. One question is how net public investment changes with the time interval over which subsidies are reduced, i.e. semi-annually, annually, etc. Governments prefer to reduce subsidies more often to lower public costs, producers prefer longer time periods for a more stable investment environment. The second question addressed is uncertainty in learning rates. Learning rates describe the fractional cost reduction per doubling of cumulative production; slower learning implies more government investment is needed to reach a cost target. We investigate these questions via a case study of subsidizing electric vehicles (EV) in the United States. Given the importance of lithium battery cost in the price of an EV, we gather historical data to build an experience curve that describes cost reductions for lithium-ion vehicle batteries as a function of cumulative production. Our model assumes vehicle batteries experience the same learning as consumer electronics, yielding a learning rate of 22%. Using learning rates ranging from 9.5–22%, we estimate how much public subsidy would be needed to reach a battery cost target of $300/kWh battery. For a 9.5% learning rate, semi-annual, annual and biannual tapering costs a total of 24, 27, and 34 billion USD respectively. For 22% learning, semi-annual, annual and biannual tapering costs a total of 2.1, 2.3, and 2.6 billion USD respectively. While the tapering does affect program cost, uncertainty in learning rate is the largest source of variability in program cost, highlighting the importance of finding realistic ranges for learning rates when planning technology subsidies. © 2015 Elsevier Inc. All rights reserved. Keywords: Experience curve Lithium batteries Subsidy policy Electric vehicle 1. Introduction Electric vehicles (EV) are thought to provide a variety of benefits to society, including decreased congestion emissions, improved environmental performance in regions with clean grid electricity, and have recently shown safety improvements over internal combustion vehicles due to increased crumple zones and low centers of gravity (Loveday, 2013). Due to these potential benefits, federal and state governments in many countries offer subsidies and/or tax credits toward the purchase of EVs. In the United States, for example, the federal incentive is an income tax deduction of up to $7500 depending on the battery capacity of the vehicle and is set to end after the first 200,000 EV purchased (IRS, 2013). EV subsidies are an example of broader government efforts to promote the development of energy technologies viewed as socially desirable, such as photovoltaic modules (Dinçer, 2011), and fuel cells (Brown et al., 2007). In general, a diffusion subsidy is set up to support a technology at a specified level over a fixed time period (Kimura and Suzuki, 2006) or target production level (IRS, 2013). At the expiration date and/or production level, a decision is made to cease or decrease the subsidy. A recent, more innovative system in Europe sets an explicit schedule for annual subsidy decreases (degression) for feed-in-tariffs (FIT) for renewable energy (del Rio, 2012; Munoz et al., 2007; Wand and Technological Forecasting & Social Change 92 (2015) 322–331 ⁎ Corresponding author. E-mail address: exwgis@rit.edu (E. Williams). http://dx.doi.org/10.1016/j.techfore.2014.12.007 0040-1625/© 2015 Elsevier Inc. All rights reserved. Contents lists available at ScienceDirect Technological Forecasting & Social Change