Your car values could go to zero and it would just be a rounding error in the grand scheme of your finances. :P
Very true. I wasn't that torn up when our previous pair of cars quit on us. We had a 1992 Buick Roadmaster wagon die at 200k (replaced by the Volvo) and a 2009 Hyundai Accent fizzled out at 262k (replaced by the Kia). Those together at the time they died would probably worth about as much as scrap value plus a Chipotle burrito. Maybe with guac if you include the phone chargers I abandoned in the Buick.
We know several companies are working on designing self-driven cars. How long do you think it takes before we do not all own private cars, but just call the cheap self-driven public car? Will that be within the lifespan of your future new car? I would think so. I would go for a used car. Even a slightly used car has less depreciation than a brand new car.
bingo - its going to come faster than most people think. GM has put in for approval for cars with out steering wheels or pedals to test in WA. and SF. The tech is right on the verge of the steep climb and adoption. tech disruptions dont come slowly they come quickly and overnight.
This is certainly something I consider when I get wistful about acquiring a new vehicle. I don't think my current 2008 car will last long enough to reach that transportation nirvana.
I'm a bit of a skeptic though when folks say it's right around the corner. That Pareto rule thing about the hardest 20% takes 80% of the time comes to mind. I think with lots of tech, it's easy to get a demo-ready version that looks for all the world like you're just about done, but in reality there's a whole lot of edge cases and error handling that hasn't been solved.
I've worked on software, and the first 80% is definitely easier than the last 20%. Getting from 99% to 100% takes at least as long as the first 80% in my experience. At least for one idiot like me working on his own.
However, the financial incentives for being the first out of the gate with autonomous vehicles/taxis is enormous. Uber knows their entire business falls apart if someone can undercut them, and autonomous taxis will absolutely be able to undercut them. There's a reason companies are throwing billions at developing this stuff. It's going to very interesting to see how it unfolds.
The AI for this kind of project relies a lot on tons of data collection to refine things, and as soon as autonomous taxis are rolled out, the amount of data they can collect will improve things even more. It will build on itself and compound. As b42 said, once the cat's out of the bag, things will move very very fast. The most disruptive part (autonomous Uber equivalent) won't even require consumers to buy anything, just download a new smartphone app. It could move even faster than previous recent innovations like the move from flip phones to smartphones or 35mm film to digital cameras. Those started out as toys for rich people before becoming ubiquitous, and this could skip even that step.
Due to all this, I don't pencil out any vehicle assumptions past the next five years or so. Once autonomous taxis are ubiquitous and cheaper than driving your own car, the used car market will crash pretty hard.
I like that the OP considers $2500 to be a "beater." The most I've paid for a car is $2000.
It's hard for me to justify tying up more than a few grand in vehicles. We currently drive a 2001 Volvo (paid $1800) and a 2005 Kia (paid $2000).
As long as you can get your depreciation at $0.05/mile or lower (meaning you get at least 20,000 miles out of each $1000 of capex on a vehicle), your opportunity cost is low, and your car is reasonably reliable and efficient, it's hard to go wrong.
Insurance can sink any thoughts of a newer car penciling out. I think we're paying around $45/mo for both cars together.
I really like this thinking with the miles per $. It is just a great quantitative way to break down the cost of a car. IMO the car is to get you from point A to B in reasonable comfort and safety, so the biggest factor is how many miles you get out of it (and not so much if your car talks to you or warms your ass or whatever new cars do these days).
This means on a $20K new car, every thousand, 20 of them, should get your 20K miles... Meaning a new car should get you 400K miles. Is that reasonable? I know cars have been known to get that far, but can it be expected? Maybe this is tough to do? This means even a $10K used car should get your about 200K miles, so a 10K used car with 50K miles will need to last to 250K miles.
Our 2005 Toyota Matrix currently has 205K miles, and we got it new for ~$15K. We always said we'd look to replace it at 200K miles, but now I'm tempted to push it to 250K or even 300K.
Really not sure what I want to replace it with at this point, anyway.
Our previous pair of vehicles (before the Kia and Volvo) cost us a combined $1700 and lasted a combined 54,000 miles, putting their capex cost per mile at just over $0.03/mile. Our total cost of maintenance and repairs (including some pretty stupid and expensive stuff) was $0.046/mile.
I only brush over it, but one key part you are missing is opportunity cost.
For a fairly el cheapo fleet like ours, it's inconsequential. If we say our 2001 Volvo and 2005 Kia are worth $3500 combined and we drive 25,000 miles per year on both combined (we do drive a lot), that puts our opportunity cost at 0.98 cents per mile if you have average investment yields of 7%. I'm not losing any sleep over that.
But for your example of a $20k brand new car, assuming you pay cash, and using the 7% number again, that costs you $1,050/year. If you drive 20,000 miles per year (very high number for a single vehicle), that's another 5.25 cents/mile in opportunity cost alone. So even if you are able to get your 400k miles (not impossible but far from something you could count on) out of that car, for the first year at least, you're paying even more in opportunity cost than you are in long-term capex per mile. Obviously this gets less dramatic with time as the value of the car decreases, but my point is that it throws off the calculations even more than you're imagining. The numbers are also a lot worse if the car doesn't last to 400k (very likely) or if you drive less than 20,000/year (also likely).
This also completely ignoring increased cost of insurance, taxes, registration, etc. And your cost of crashing the car is higher. Or if you don't feel like taking that risk, you get to add the cost of full coverage insurance. Either way your risk-adjusted cost is higher.
You have to take the entire picture into account. Even on places like this forum, opportunity cost is the most overlooked part of that.
The best way to compare total base vehicle cost (agnostic of fuel efficiency and insurance/taxes) is adding together capex per mile, total repairs and maintenance per mile, and opportunity cost. People tend to overestimate the cost of repairs and maintenance per mile of a decent older car, and underestimate the opportunity cost of a new car. Paying $550 to replace the gas tank of a 23 year old car with 176k miles on it at the time (true story) might feel insane in the moment, but paying a grand a year in opportunity cost on a 2018 car is completely invisible to most. Out of sight, out of mind.