[rodonn]

This paper has actual data on earnings from all Uber drivers in all cities in the US. They find that the average earnings are $20 per hour with a standard deviation of $5 (there is substantial variation across cities and especially across time, with late nights on weekends being especially high earning). http://www.nber.org/papers/w23296 The authors are very well respected economists from UCLA and Yale, though it should be noted that one author was a former Uber employee.

This is substantially higher than the estimates in this newer study unless you think that the per hour operating costs of a car are $16 per hour. The new study bases it's numbers on self reported values from a relatively small sample, which is much less accurate than calculating an average using Uber's actual full database.

Another interesting finding from the study I linked is that a majority of Uber drivers work 20 or fewer hours per week and place a substantial value on the flexibility that they get from being able to choose their own hours.

[dragonwriter]

> This is substantially higher than the estimates in this newer study unless you think that the per hour operating costs of a car are $16 per hour.

Fuel costs per hour could easily be a quarter or so of that, and fuel costs seem to be around 1/7 of total operating costs [0], so, yeah, $16+/hr isn't unreasonably high.

[0] https://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/pu...

[haldean]

I can't read the paper you link but in the abstract-ish thing you've given here, $20 is a mean. The MIT paper says $3.37 is a median. Both could simultaneously be true!

[rodonn]

Sorry about the paywalled link. Here is one that should be freely accessible. http://www.anderson.ucla.edu/faculty_pages/keith.chen/papers...

I agree that the mean and median could in principal be this far apart, but it would require a lot of skew in the data (small number of very high value trips -- seems possible).