About The Authors

Wednesday, November 23, 2016

Why BEVs Won't Be Disruptive

Mike Smitka, Torino Italy

In 2030 I expect that Toyota, VW and GM will remain the top 3 global automotive producers (though not necessarily in that order). The flip side is that neither vehicle electrification nor autonomy nor Mobility 2.0 businesses will prove disruptive.

...Disruptive Technologies? Not in Automotive!...

Each of these purported threats have their own challenges as technologies and businesses. That is for other blog posts. All three however have a common feature: new technologies roll out slowly, and in the auto industry they roll out very slowly. Even with rapid commercialization, in 2030 only 1 in 10 vehicles on the road will be BEVs (battery electric vehicles).

New technology adoption and diffusion follows a logistics process: slow early on, then accelerating, and slow again towards peak. That is true in theory: few are willing to chance adopting a technology when no one they know has done so. Similarly, towards the peak those who have yet to adopt a technology have refrained not because (or not only because) they are obstinate but due to idiosyncratic circumstances. This is a robust empirical finding, dating back to Zvi Griliches' classic 1957 study of hybrid corn. Commercial hybrids were first developed in 1923. While half of Iowa farmers used such seeds by 1938, farmers in regions where corn was less widely planted continued to sow non-hybrid cultivars until the 1960s.

Automotive technologies are no different. Initial costs of a new technology will be high, while performance will still have room to improve. Historically many technologies appeared first as an option on luxury cars. If the uptake was good, one or more firms might make it a standard feature. As the volume rose, suppliers would reduce the price point, and OEMs would migrate it to high-volume products.

Feasible BEV Rollout Scenario

Year

Global new vehicle output

BEV share

BEV share vehicles on road

2020

100

1%

0%

2021

103

1%

0%

2022

106

2%

0%

2023

109

2%

1%

2024

113

3%

1%

2025

116

5%

1%

2030

134

41%

11%

2035

156

77%

35%

2040

181

81%

55%

This process is thus constrained by the commercialization process, by the standard "learning curve" and economies of scale effects, and by the time needed for the supply chain to add new capacity. It is also constrained by the new model development process, because it is highly unusual for a feature to be introduced in the middle of a model year. So the use of new technologies can only expand as models are redesigned, which for standard sedans is done a rolling 4-year cycle. That puts a limit on the pace of adoption. Furthermore, it may only be possible to introduce a radical technology with a new platform; those are developed on a rolling 6-10 year cycle. Drivetrains are also redesigned less often. And heavy trucks may not be fundamentally redesigned for as much as 20-30 years. (One major brand uses an H-frame first introduced in the 1960s, before the advent of the steel and aluminum alloys that are widespread in the passenger car market.)

[The bulk of the engineering for a standard passenger vehicle model takes place over roughly a 12-month period, with a smaller advance team working on model specifications at the front end of the process, and at the tail end a smaller team seeing the design through to SOP (the start of production). The full process thus spans 18-24s months. The rolling development cycle is thus due to staffing constraints in the development process, and the desire of the marketing and dealership end to have a steady stream of new models, but not a flood of them.]

In the past, even rapid rollouts of technology in the automotive space, such as when there is a "hard" regulatory deadline, has required over a decade. The fastest example of which I'm aware is the replacement of carburetors by technically superior fuel injectors, the latter necessary to meet emissions requirements. They had been used intermittently in racing from the 1950s, and began to appear on low-volume luxury cars in Europe in the 1970s. However, they were complex and costly mechanical contraptions. That changed with the introduction of microprocessor engine control units (ECUs), which also made fuel injectors much more effective. The first Motorola ECU was launched in 1980, and by 1990 GM had converted the last of its engines to the new technology. At the firm level, the rollout was over one decade, but for the industry as a whole it follows a logistics curve. Pulling off this fast introduction required huge investment. To facilitate the fast pace and not be hostage to Motorola, GM invested in its own semiconductor manufacturing operation; for a time it was the fourth largest chip maker in the world.

...15 years from now BEVs will still account for less than half of production. That's hardly disruptive!...

So what does it look like if you combine industry specifics with a logistics curve? First, by 2020 global production will be 100 million vehicles, and slowly increasing. Globally there will be perhaps 1 billion vehicles in operation, with 8% scrapped in a given year (at which rate the average vehicle on the road will be 11.5 years old). Finally, because large vehicles are unlikely to be BEVs, it's sensible to assume diffusion peaks at 80% of the market. You can read the numbers for yourself.

This is an excerpt from one section of a paper on new vehicle technologies that I presented this will at the "Toronto-Torino Conference" organized by the Munk School of Global Affairs at the University of Toronto, Collegio Carlo Alberto, and Politecnico di Torino. Along with wonderful food and wine, the conference also included a tour of the Torino assembly plant of Maserati.