I am quite sure your model is good, but I think the drop in ICE sales is doing to happen earlier, as we just discussed. I don't see that much growth in ICE sales; I don't see how it happens. If the current 3.1% growth rate is maintained we only get to 90.4M peak *vehicle* sales by 2022, some of which are EVs -- the rate of growth car sales would have to go back up, and I don't see why it would.
I believe the "put off buying a new car -- there will be EVs soon" dynamic may be much larger than we had guessed.
Or it may partly be dynamic of "put off buying a new car -- there will be robotaxis soon", which lots of people seem to believe (though I don't think it's true, here the belief is what matters).
Or maybe cars really are lasting longer than they used to.
I agree. My model is not designed to anticipate a massive Osborne effect. I am assuming a 3.6%/y growth rate for total car sales. Then I've got a logistic model that that divvies up the market share between ICE, BEV and PHEV. To model an Osborne scenario, I could suppose some production ramp for EVs alone, then use the market share model to extrapolate ICE sales. So for example, if my model says that EV market share is 5% in a particular year and EV production is 5M, then we back into ICE sales of 95M. But if EV production is just 4.5M, we get ICE sales of 85.5M. It is a bit strange to think of a market functioning this way, but it would produce the desired effect. That is, we want somehow to represent the idea that production constraints in EV translates into suppressed growth for the entire market.
There may be some other way to model pent up demand for EVs that suppress actual sales. For example, Suppose total demand is 100M with 5% allocated to EVs. But only 4.5M EV are produced. Thus, 5M demand minus 4.5M supply leaves 0.5M unsatisfied demand for EVs. So total actual sales is just 99.5M (95M ICE, 4.5 EV), but the other 0.5M unsatisfied demand is making due with used cars for a year. This demand carries over to the next year. Suppose next year there is 7.5% demand for EV on a total demand for 103.6M vehicles. So specific demand for EVs is now 0.075*103.6+0.5= 8.27M, including the 0.5M carry over of pent up demand.
Let's suppose production ramps up 60% to 7.2M EV produced. So unsatisfied EV demand is now 8.27 - 7.2 = 1.07M. Meanwhile demand for ICE is 103.6*(1-0.075)=95.83M, and total sales are 103.03M on total demand of 103.60M. So we see with this example, that the growth rate in total auto sales can be suppressed for lack of EV production, meanwhile the pent up demand for EVs allow EV makers to command a premium for their product and motivate rapid growth in production.
Had EV production only grown by 30%, half of 60%, then EV production would have been just 5.85, so total sales would have been just 101.68. And pent up EV demand goes to 8.27-5.85=2.42M more than twice what it would be had production grown 60%. So we see that this pent up demand can grow quite fast if production growth is anemic.
In this cycle total EV demand with 8.27M with 0.5M pent up. To have satisfied this, production would need to grow 8.27/4.5=1.84, or 84%. Even modest amounts of pent up demand can motivate rapid ramp up in production. Also in this scenario Total auto sales would be 104.1M.
To summarize the scenarios, total demand grows 3.6% to 103.6M in this year, excluding pent up EV demand. But auto sales can range from 101.68M, 103.03M to 104.1M depending on EV production ramping up 30%, 60% or 84%. Perhaps a model along these lines could explain how eventually growth in the total auto market comes to depend critically on how much EV production can ramp up in a given year. I think it is transparent that the massive sales increase in the Model 3 is in part a function of pent up EV demand. Indeed, every reservation holder was signaling pent up demand in a very concrete way. My guess is that most of these reservation holders were simply holding onto their existing cars while waiting for their Model 3 to come to be built. So they held onto "used" cars for an extra year or two.
If pent up EV demand does become huge, what does this do for the used car market. I think in the short run it supports used car prices. But when EV production surges to catch up with this pent up demand, the situation can quickly change. People start to dump their old ICE onto the use car market as they finally drive away with the EV they've been waiting for. This potential for turbulence could motive ICE buyers to lean a little more towards buying or holding a used ICE vehicle. But I don't think EV makers can catch up fast enough with pent up demand to really undermine used car prices. The effect of pent up demand for EVs should be to suppress the ratio of average new ICE price to average used car price. This is really bad news for ICE makers because they are competing both with EVs and used vehicles for sales.
These are all things we've discussed quite a lot, but the challenge is how to express all this in parsimonious mathematical model. Modeling pent up EV demand introduces a latent variable, which cannot be directly observed. This really ups the complexity of the model, but may be necessary to represent an Osborne effect.