Hacker News Reader

Comment by BurningFrog | Hacker News Reader

Alex Tabarrok pointed to the well done study, that of course never got anywhere near this level of attention: https://twitter.com/ATabarrok/status/969651065695227904

"Dubious. Cited paper uses self-reported data from 1100 drivers. Cook et al. use exactly measured data on 1.8 million drivers and calculate average wage after expenses of $13 per hour. https://web.stanford.edu/~diamondr/UberPayGap.pdf "

View on HN

It definitely does not look like you describe.

The original article is self-reported income, which i agree is crap and proof that it can be interpreted any way one wants, as is show in this thread: this is clearly PR damage-control from uber.

The one you linked is the other end of the spectrum. It is a math formula on values provided by uber and completely ignores the associated costs with the profession. It does not account for down time, it does not account for insurance, car depreciation etc. The $13 is just more meaningless data because all it measured in the paper is "income from uber, taking into account avg on-meter time with pool+surge pricing variations". (they do include the costs, but as "The primary costs drivers face are fuel, maintenance, depreciation, and fines for parking or moving violations. Zoepf et al. (2017) estimate median driver’s expenses are 32 cents per mile." which is very far from actual data)

IMHO both are utter garbage.

> does not account for down time, it does not account for insurance, car depreciation etc

Most jobs have those costs. You need a car to get to work; you pay for gas, insurance, tolls, etc. Time is wasted during transit/commute.

Are drivers special in this respect?

I assume they mean the extra costs in those areas due to driving, for example, driving for Uber will cause you to use your car much more, which can wear your car down faster.

Yes, insurance rates for ride share cars is higher because it's a commercial vehicle. And yes, most people commute an hour or 2 each day, not rack up hundreds of miles for 8+ hours a day. So yeah, the cost is not comparable.

> Yes, insurance rates for ride share cars is higher because it's a commercial vehicle.

Do Uber/Lyft drivers typically make their activities known to their insurance companies?

If they want cover in the event of an accident they do.

State Farm had a warning in their recent mailer for policy renewals that you are not covered under your auto policy for any ridesharing activities, although you could purchase a rider to insure you at additional cost.

Also, if you say 12 or 24 thousand miles a year and get into an accident with widely unexpected mileage on your car, they will suspect something is amiss and likely investigate.

State Farm also obtains your mileage data from "third party providers".

https://splinternews.com/how-car-insurance-companies-spy-on-...

> Do Uber/Lyft drivers typically make their activities known to their insurance companies?

Maybe not, but if they do there is substantial legal risk (not just for accident liability—for which they will be uninsured—but also for the legal consequences of driving without legally mandated insurance coverage.)

From a sample of 10, 8 were using private insurance and planning to lie in the event of an accident.

I have no way to know if this is at all represebtative. Moat expressed the rates were so high as to prevent any but full time drivers (40+ hours/week) from profiting with it.

I am certian Uber knows, as they choose not to verify this.

No you don't. I have never used a car to get to work. Most people I know do not drive to work.

You seem to directly contradict yourself. It doesn’t “completely ignores the associated costs” if they are deducting 32 cents a mile?

"...which is very far from actual data"

The study used a measure the poster did not think was accurate.

...which contradicts "completely ignores the associated costs" (emphasis theirs). Hell, the poster even said "ignores depreciation" in one sentence and says how the number includes depreciation in another sentence.

There's really no defending that paragraph.

Are you arguing that 32 cents per mile is too low a cost? It's lower than IRS reimbursement rules but it's pretty consistent with a calculation of depreciation/maintenance plus gas for a reasonably economical car, especially given that some costs like excise tax are indeed pretty much fixed. So long as those costs are also applied to miles driven when "on the clock" but driving to pick up a passenger (i.e. not being paid), it seems like a very reasonable ballpark number.

The Stanford paper estimates gross earning to be around $18 per hour and all expenses including insurance, downtime, gasoline and depreciation (but not fines) at $5 per hour. This gives net earnings of $13 per hour.

To get the $5 per hour in expenses they use 25 cents per mile for those expenses and 20 miles driven per hour.

This is all explained on page 36 of the Standford paper which I think is excellent but is primarily concerned with understanding the gender earnings gap not the distribution of net earnings.

The IRS mileage rate is $0.545 per mile, and is supposed to be a realistic average for car expenses.

Edit: per mile, not per hour. Oops! fixed the rate, too

$0.56 per mile not hour! (Though I see $0.545 per mile [1].)

[1] https://www.irs.gov/newsroom/standard-mileage-rates-for-2018...

It's supposedly more or less an average across car types and includes a percentage of fixed costs however. It's almost certainly high for a driver who would own a car anyway and has an economical vehicle that optimizes for Uber driving. It's going to be a range depending on the driver though 25 cents per mile (all miles, not just those with a passenger) is probably near the bottom end.

If I was an Uber driver, that's the rate I'd use for tax accounting purposes.

That's for any car though. Most Uber drivers are driving Priuses or similar and therefore have much lower costs.

The study would do well to use such a figure, as it is objective and sets a reasonable upper bound on the costs.

Why is it better to use a upper bound rather than a more realistic estimate?

Because it is 1) conservative and 2) a "realistic" figure has unknown variance.

Reasonable conclusions that don't stretch evidence get less attention

Thanks for the link! But I think you have an extra double quote at the end

Sorry, I was quoting his tweet in its entirety, which made the link unclickable.

Reply on news.ycombinator.com