Learning from the success of battery electric vehicles

For anyone interested in accelerating the pace of innovation in net zero technologies, the transition to electric vehicles (EVs) is a particularly rich case to learn from. Eugenie Dugoua and Marion Dumas explain why.

This is the story of two technologies, both strongly supported by public research and development (R&D), but one – fuel cell electric vehicles or FCEVs – failed, while the other – battery electric vehicles or BEVs – eventually succeeded. FCEVs are a failure in that after many years of R&D and enthusiasm, they did not materialise at any significant scale. On the other hand, BEVs are diffusing rapidly and cost parity with conventional vehicles is within reach.

We recently published a study that analyses what happened to precipitate this situation, in particular the series of innovation activities, investments and policy choices. The story we uncovered suggests that innovation policies could accelerate low-carbon innovation in a number of sectors that are still lagging behind by targeting all the technologies that complement each other in creating the new low-carbon system and coordinating those efforts internationally.

When global uncertainty met serendipitous innovation from the electronics industry

Our analysis of the EV case surfaces and solves two puzzles. First, why were EVs not produced commercially before 2008 when authorities had been providing public support for R&D for both battery electric vehicles and fuel cell electric vehicles since the early 1990s? Second, we show that carmakers globally were collectively focused primarily on FCEVs in the 2000s. Why, then, did BEVs become the favoured replacement for internal combustion engines? 

Both technologies offered a path away from internal combustion engines. Each came with strengths and weaknesses and its own set of challenges. Critically, both required entirely different infrastructure. BEVs rely on rechargeable batteries and need a network of charging stations, while FCEVs use hydrogen fuel cells and require a hydrogen production and distribution system. For policymakers and carmakers, the challenge extended beyond weighing the merits of each technology: it required anticipating and aligning with the collective choices of other countries and manufacturers.

The automotive industry is highly integrated on a global scale, with complex supply chains and shared markets. This meant that for either technology to succeed, it needed widespread adoption not only by multiple car manufacturers, but also by suppliers, energy providers and policymakers across different countries. And no country or carmaker would have wanted to develop one technology if it was not going to be the globally preferred technology. In our study, we therefore argue that the choice between BEVs and FCEVs represented a global coordination problem.

The lack of resolution to this problem meant there was a great deal of uncertainty around which low-carbon technology would dominate. For several decades, as this uncertainty persisted, carmakers and policymakers focused on R&D, but without producing either cars or charging points. National policies aimed at encouraging R&D into either FCEVs or BEVs were not determinant in global carmakers’ choice of which technology to focus on. Indeed, national policies are not formulated at the right scale to ease the concerns of companies operating globally. What was determinant was the wave of innovation in lithium-ion batteries, driven by the booming demand for smartphones and laptops. This advancement in battery technology external to the automotive sector suddenly made BEVs more viable and attractive. It provided the necessary know-how and supply chain that incumbent car manufacturers, but also new entrants like Tesla, could leverage.

This serendipitous boost from an unrelated industry tipped the scales in favour of BEVs. Around 2008, there was a noticeable shift in the industry’s focus, with battery-related patents surging while fuel cell patents declined. This trend was further reinforced when policymakers across major markets began aligning their support behind BEV technology around 2010, including through investments in charging infrastructure. This appears to have been the tipping point in which everyone’s expectations aligned on BEVs becoming the replacement for the international combustion engine, thus lowering technological uncertainty. It is after this tipping point that investments really took off and commercialisation truly started.

This then solves the two puzzles. FCEVs failed despite carmakers’ R&D efforts because complementary technologies to produce and distribute hydrogen were missing, while the complementary technologies for BEVs came about through the electronics industry. Twenty-plus years of R&D did not lead to new products until 2010 because a global alignment on technology choice was needed to enable complementary investments in infrastructure.

Policy lessons – support for complementary technologies and catalysing ‘learning spillovers’

The EV transition case puts in question several heuristics that seem pervasive in discussions about innovation policy.

First is the idea that technologies pass through different stages in a linear sequence: through niche applications to initial diffusion and, after some upscaling, system integration. This sequence corresponds well to solar photovoltaics (PV), which started with niche markets (off-grid, subsidised rooftop panels), and could initially penetrate energy systems with little friction. Now that solar PV is highly performant and affordable, investing in solutions to integrate this technology into the wider system (smart grids, grid storage and so on) is widely viewed as worthwhile.

This linear sequence model implies that we can focus on supporting a given target technology without worrying about other technologies and integration into a bigger system until later. Yet, the case of FCEVs shows that this may not always be the case. If the complementary elements are necessary at the outset to obtain a useful technology, then a sequential approach is not possible. Innovation policy needs to simultaneously support all complementary technologies, otherwise bottlenecks will arise which may lead to ultimate failure.

Second is the tendency to develop innovation support programmes for each technology of interest, rather than a portfolio strategy that considers learning dynamics across technologies. Progress on a technological problem in one area helps make progress in another by teaching us something. Economists call this ‘learning spillovers.’  As shown by the EV case, these learning spillovers are critical for the success of technologies.  But R&D policies are rarely designed with the aim of catalysing them. For example, public R&D was focused on fuel cells in cars, rather than all applications of fuel cells (such as shipping, machinery and industrial processes). This limited the progress of fuel cells both in cars and elsewhere.

How governments can do better: strategic, coordinated choices

Much can be learnt from a current initiative in the United States that aims to radically improve the strategic intelligence behind innovation and technology policy. This initiative was started by Professor Fuchs at Carnegie Mellon and enlists the efforts of interdisciplinary academics throughout the country to build a National Network for Critical Technology Assessment. The NNCTA is developing responsive analytics to address the systemic questions raised above, and more. The tools being prototyped aim to make better technology choices by understanding how technologies address public goals or missions (defence, climate change, wellbeing, health and so on), and identify the synergies and trade-offs between technologies and between all the public missions that technology policies are trying to advance. They also identify bottlenecks in infrastructure, equipment and skills. Such a coordinated approach would be very likely to accelerate low-carbon innovation by addressing the systemic barriers that often hold it back.

This kind of approach of course calls into question another prevalent idea, which is that innovation policy should remain technology-neutral. The belief is that the distributed intelligence of market actors is more likely to identify the better solutions than policymakers could. However, it is difficult to ensure support for complementary technologies in infrastructure in a technology-neutral portfolio that funds technologies broadly. In addition, and this is a key point of the EV story, technological neutrality maintains the uncertainty about which technology the industry is transitioning to. When this uncertainty prevents the scaling-up of investments, that slows down innovation. As the NNCTA initiative shows, strategically choosing technology portfolios can also draw on distributed intelligence, even if it does not rely on the market to pool this intelligence together.

Finally, the EV story calls into question governments’ usual practice of developing innovation policy at the national scale. This approach leads to a fragmented policy landscape for industries that are often global in scope. Again, this reinforces uncertainty and discourages a more vigorous and determined effort. As articulated by Simon Sharpe in Fives Times Faster, institutions are currently missing to support innovation and problem-solving for decarbonisation in each specific sector, at the scale at which they operate, which is usually international.

Why this case matters so much

The case of solar PV has been very well studied and is a fantastic example of how to grow new low-carbon industries. However, the case of EVs may be more helpful in understanding how to promote low-carbon technological change among existing global-scale producers of tradable goods, such as cars, ships, planes, steel or cement. Our study thus provides a counterpoint that should help enrich and nuance the design of innovation policies.

The authors’ paper ‘Coordination dynamics between fuel cell and battery technologies in the transition to clean cars’ was published in the journal PNAS 121:27 on 24 June 2024.