Tag Archives: Consumer decision making for EVs

Testing Tony Seba’s EV Predictions 13 (What Makes Us Buy a Car?)

“What makes us buy a car?” Well something obviously does since global sales of cars and light vehicles reached around 100 million a year in 2017. Moreover, if we can answer that first question then we are in a better position to answer this question: “What makes us choose an electric vehicle rather than an internal combustion engine (ICE) vehicle?”.

These questions come from the demand side. Up to now, all my posts have basically been dealing with the supply side. That is asking questions such as “Who will sell EVs?, “Who is  investing in EV production?”, “Will there be enough lithium?”. My answers to those types of question leads me to believe that the supply side can keep us on Tony Seba’s EV penetration S-Curve through to at least the mid-2020s. But will anyone buy these EVs even if the auto makers build them?

Hoping to shed some light on that question, I have been toiling over reports by companies and organisations that forecast how many vehicles will be sold in the future and what percentage of those vehicles will be EVs. In September 2017, David Roberts at Vox wrote a nice piece pulling all those forecasts together:


A reminder: I have taken Tony Seba’s death of ICE vehicles by 2030 forecast to mean 95% market penetration (130 million EV sales). From the middle chart above we can see what everybody else thinks:

  • Exxon Mobil:   10% by 2040
  • British Petroleum:   10% by 2023
  • Norway’s Statoil:   40% by 2040
  • Goldman Sachs:  9% by 2040
  • OPEC:   12% by 2040
  • Total SA:   22% by 2030
  • Bloomberg New Energy Finance (BNEF):  55% by 2030

So they don’t agree with Tony. But my interest with them is less in the forecasts themselves (a topic for a future post), but rather how they were arrived at. Keep in mind that what I did in the chart below is not make a forecast but rather fit a few data points into a logistic function to produce an S curve:

Seba Central Scenario

To repeat, this is a data fitting exercise. I’ve just taken Tony Seba’s point forecast of 95% penetration by 2030, the current penetration and sales of EVs, and added an S curve that looks reasonably sensible. But it is not a true forecast.

There are a couple of ways to forecast: 1) you can create a model which is purely empirical and takes an historical time series and then projects it forward, or 2) you can build a microeconomic model in which your forecast variable is a function of variety of other variables. Hoping to learn what methods and methodologies these esteemed organisations and companies have been using to arrive at their EV penetration rates, I trawled through the back pages of their reports looking for references and helpful appendices, but there were none to be found (at least as far as I could find; I am still working my way through these reports).

I get the distinct impression that the forecasts drop down like manna from heaven. It’s basically a game of pick a number that looks roughly right based on your organisation’s attitude toward EVs and then construct a vague argument surrounding that forecast number. You could argue, perhaps, that it is impossible to do anything more rigorous given the profusion of variables that come into the purchase decision, stretching from GDP growth rates and the oil price at the macro level to eco fashion and periodic EV safety scares at the micro level. But I still think we can do better than that, by adopting a somewhat different approach.

From the existing academic literature based around ICE purchase decisions, we know that consumers have a fixed budget and are presented with a basket of goods that they can spend the budget on (or save). You can consider that basket of goods as consisting of cars and non-car goods. A consumer will buy a car rather than the alternative of non-car goods if the car gives him or her a higher degree of utility (happiness or pleasure). The sources of utility derived from a car can be thought of as threefold:

  • Mobility
  • Aesthetic
  • Status signalling

So cars aren’t just a means of transport: they are also like large pieces of jewellery which give one pleasure through personal contemplation of them and through displaying them to other people.

Finally, the budget constraint is temporal: it stretches from the present to the future. So the purchase decision is not just bounded by the current available budget but also budgets through time. Accordingly, the purchase decision takes into account future costs captured by

  • Fuel
  • Maintenance
  • Depreciation and replacement

This is all pretty much common sense even if I have used economic terms, which may be unfamiliar for some. But I hope we can pull out some insights that will impact on EV adoption rates. The next bit is important: I don’t have to build a complicated multivariable model and input millions of data points into a computer to derive some insights. From basic theory, I know how the model will act in certain conditions. Let’s look at one such set of conditions.

First, we start with a budget constraint existing now in an ICE dominated world. Consumers have allotted to spend an amount of money now and in the future on a means of mobility whether EVs exist or not. This is where it gets interesting and why I feel the adoption forecasts by the likes of Exxon Mobile and the International Energy Agency make no sense whatsoever.

If any individual consumer is faced with an EV that is a) cheaper than an ICE, b) has lower ongoing costs than the ICE, c) has better mobility characteristics than the ICE, d) is prettier than the ICE and e) signals status better than the ICE, then that consumer will always choose the EV. That is because the consumer will get a higher level of utility from the EV purchase. (Alternatively, the consumer will spend less money buying an EV yet get the same amount of utility as buying the ICE, so then having extra money to spend on non-car goods.)

In fact, I can even relax these conditions a bit. Let all the factors be equal except one where the EV is better. That market will also flip 100% EV too. Don’t believe a market can do that? Look what happened with film and digital cameras.

I admit to some brakes on the transition: the consumer has to be aware of how the variables have changed in EV’s favour. That requires active information searches or exposure to information through advertising and marketing. The psychology and marketing literature covers that aspect of decision-making well.


Moreover, a more important reason why markets don’t flip overnight (ignoring the supply side) is because they go through an intermediate phase where some of the factors differentiating between products are positive and some negative. And given that consumers derive different utilities from each factor, the market will move in phases from one state to another as each individual faces a different decision-making process.

For example, purchasers of the first Tesla model S were deriving most of their utility through signalling that they a) were an early adopter of new technology and so were hip, b) had superior eco credentials and c) had immense wealth since they could afford the car. The rest of the population didn’t have the budget to buy the car and/or was less interested in the status signalling or, maybe, just didn’t like the look of the car. So the market didn’t tip the first day the Model S went on sale.

This brings me to why I think consensus forecasts for EV penetration are barking mad. As of 2018, EVs have already exceeded ICE vehicles with respect to a number of the variables. They pose less of a future budget constraint because they a) are cheaper to run, b) have lower maintenance costs and c) have a longer potential road life as they have fewer parts to wear out.

So we are left with the upfront price of an EV versus an ICE (which is the current budget constraint), the mobility functions, aesthetic and status signalling.

For EVs, nearly all of these factors relate to the size and cost of the battery pack. If the battery pack price can come down enough, the drive mechanism of an EV will be cheaper than an ICE, which means the overall vehicle will be cheaper. Note that engine power delivered by a battery is already better for an EV than an ICE. If the energy density of batteries improves enough, then the mobility characteristics such as range of the EV will be able to at least match the ICE. If the battery pack shrinks enough, this will improve mobility characteristics such a storage and opens up a range of new form factor possibilities that impact aesthetics.

There is one aspect of the aesthetic that I think an EV can’t match; that is, the joy brought to some people through owning a piece of precision engineering. A high end ICE is a marvel of engineering and an ICE engine has an aesthetic all of its own. The residual ICE market for me will be cars as luxury watches. Rolex, Tag Heuer, IWC, Breitling and Breguet of Switzerland still sell mechanical time pieces, but their market share is tiny. That to me is the fate of the ICE due to the cold, hard logic of the EV surpassing the ICE on  every other variable that goes into a consumer decision-making process.

So the battery holds the key. That will be the subject of a future post.

For those of you coming to this series of posts midway, here is a link to the beginning of the series.