MIT recently posted a sensationalist research article where the authors claim 74% of Uber and Lyft drivers earn less than the local minimum wage. Which begs the question, if the authors are right, why in the hell would hundreds of thousands of people do that? Are they all too stupid to do math, have no other options and are being taken advantage of by the Man, and the bright boys at MIT are showing them the way? I don’t think so. I think the bright boys at MIT don’t understand basic business, economics, or accounting.
The answer lies in tax efficiency, fixed cost and fixed asset allocation ratios, and marginal cost economics. One must look at it from the driver’s perspective, and ask why, if the authors’ survey data is right, is it still happening at scale and growing?
In my Uber riding experience, I ask drivers why they drive and how well they do and is this full time. Largely the responses are either: 1) I’ve been driving full time for a while and make good money, 2) I’m driving full time because I am between jobs and need to cover my car note, and make what I can until I get a new job, 3) I drive part time to a) fill my off hours, b) make some money on my commute, or c) drive until I’ve paid off my insurance and car note for the month then stop, or d) pay for a nicer car than I’d otherwise have.
Let’s start with the low hourly wage argument – The authors implicitly assume a couple of really, inaccurate assumptions that square the circle. Number one, they forget that of their $0.30/mile “cost”, only 20% is fully variable, gas, and the other 80% is fixed to mostly fixed (depreciation, insurance, maintenance) – their data, not mine. They basically argue that since “bulk of miles driven” are for ride sharing, all the costs should be allocated to the hourly wage of the driver. This makes no sense on multiple levels. Cars are fixed costs, drivers may or not be. This matters a lot because their total cost analysis is basically allocating all of that fixed cost on a variable basis when they are calculating driver profits. Drivers are assuming the opposite. Drivers are assuming that those car costs are largely fixed not variable. You can see that when they tell you things like “driving til my car costs are covered”. In fact, drivers are implicitly assuming that some of their labor hours and some of their fuel costs are ALSO fixed, which is quite interesting behavioral economics. The comments, I drive to cover costs on my commute, imply the driver is expecting to spend some or all of those hours driving and some of that gas money, and all of the other costs anyway. And a driver that says, I’m driving to fill my hours, cover my car note, or because I am between jobs is also arguing their labor costs are more fixed than variable – basically, “I was intending to be working anyway, so the “costs” to my household of the labor hour as non leisure were getting borne regardless” so I might as well take the highest revenue hour available at that time to cover them. What does this mean for profits? It means drivers are very, very rationally looking at most if not all of the costs that the MIT bright boys subtracted as fixed, and charging them to or spreading them across their full household base, not charging them to the hours driven. This is basic cost accounting stuff in any business. From a pure economics standpoint, if you have underutilized or absorbed fixed costs, you are better off taking any marginal revenue that exceeds your marginal cost, and your total profit, and net profit margin goes up. It’s called operating leverage. The only driver flaw in this allocation could be wear and tear, which they may be underestimating, but if it gets them a nicer car, and another they are just going to buy another one sooner, they probably don’t care, and neither does the economy.
One commentator in the TechCrunch article whines about the high 10-20% portion that Uber/Lyft take as holding down hourly wages. Ignoring the fact that even at this level Uber and Lyft aren’t yet making money, customer acquisition cost for any business (let alone what it costs a taxi driver without Uber) is usually $20-$100/retail customer (larger than the average Uber ride), usually a partially fixed upfront cost, and usually a material portion of the sale. If you think about it, the average retail gross margin is about 50%, meaning that the company who made the product gives up 50% of the customer value just to make the sale – on every sale. The average distributor gross margin in the 15-20% range, and virtually any business will give up 10% marginal commission to get a marginal customer without thinking. But Uber, at 15-20% does one better – they deliver you the technology to get the job done, they deliver you a customer at no upfront cost, a customer literally minutes from your “storefront” right now, let you pick between customers who have different price points (which is very hard for companies to do), they don’t require you even to commit minimum hours or even show up like any other customer or company, they handle all payment risk, they handle pricing, and enable you to get premium price in surge pricing without any extra work, and they pay you basically ASAP, not two weeks or monthly basis, let alone net 30, but we actually pay you in 90 because we are GE, don’t charge you Medicare and Social Security taxes, and allow you to create tax deductions for your own personal assets that wouldn’t otherwise exist. All for a small fraction of what it costs Fortune 500 companies with armies of people and top tier brands to acquire a customer and get their product to them. Yet, I should call this a bad economic deal for the driver?
We should probably even be looking at this as a net benefit, not cost. If you look at compensation surveys, employees will generally swap some amount of pay, perhaps as much as $5-10/+hour or 5-20%+ of compensation to get work complete flexibility in any job. Uber offers you that on Mark McGwire level steroids. So Uber drivers may be “grossing up” their labor rate by a fairly large amount, and moving in and out of the Uber market – especially if the alternative income opportunity is drive an extra average 30 minute commute to work (which the company does not pay for, to work another 5 hours a week getting from 35 to 40 (as most businesses try to do so they can afford the federally mandated time and a half).
Marginal Cost Economics – We should talk about the implicit idea the authors have that Uber driving is a bad deal for the marginal driver at the marginal labor hour. They note that 80% of drivers drive less than 40 hours a week, but then proceed to ignore the impact of the very rationale reasons why.
We have been inured to believing that for work hours, “overtime” should get paid more. But this simply is nuts. It’s just a modern labor law construct to stop abuse by corporations during the progressive era, and an incentive to get your employees to work extra for short spurts. But in fact, by hour 60 or 80 in a week, I guarantee we are all less productive per hour than we were at hour 20, and should in theory get paid less (and virtually every aspect of the economy and society does better if we pull more productive hours from people not currently working at a lower rate, than if we pull less productive extra hours from people already working at a higher rate). From a personal work choice perspective, you take your highest dollar first – only an idiot would choose to work a lower value wage first, and higher value one later. Of COURSE driving Uber part time should be lower $/hour than your main job, otherwise you’d have switched to full time, and so would everyone else (since the secret of Uber is reducing barriers to entry for new drivers), until the hourly rate was back down below the average full time labor rate again. Saying Uber drivers should get a higher rate for their last hour of work a week ignores every law of microeconomics. However, in the magic of gig economy, if might actually be freakanomically true in this unique case because Uber may be helping households absorb underutilized fixed costs and fixed assets in a highly tax efficient manner with very little friction, set up, and transaction costs.
Untaxed Driver Profits – The authors then argue that 74% of driver profit is in untaxed (and that this is bad for the economy/not fair etc.). They don’t seem to understand the definition of profit or how economies work. If most drivers are part of two income families, or are driving part time on top of another income (as they also mention, 80% of drivers are < than 40 hours/week), then driving makes perfect sense across a wide range of scenarios. From a driver’s perspective, you drive Uber as long as marginal revenue exceeds marginal costs (including the marginal labor rate you are willing to accept), or until you cover your fixed costs, and magically make your car note, depreciation, insurance, car repairs, and any gas you spent (that might have been expended anyway in the commute case) all become tax deductible, just like for any business. The authors want to charge this to the driver’s Uber income, at the nutty $3.37 hourly wage estimate they are using – newsflash 100,000 Uber drivers are not all so stupid and unemployable that they are working for $3.37 an hour – your analysis sucks. However Uber drivers appear to be doing the opposite – they are basically treating most of their expenses as fixed, and then ALSO sheltering higher tax rate income from their spouse or other job with Uber driving by shifting those expenses into the tax deductible category, and covering some fixed labor costs that they’d otherwise lose (from 0-c.80% of the labor hours depending on the driver). They are actually getting paid for those costs as tax efficient profit center, not charging them as expenses.
Meaning the Uber driver sees this as a very high revenue opportunity. We can always change the car tax deduction – but it would be very uncool for big businesses to get to deduct their cost of goods, but the small Uber driver business get told sorry, you’re with Uber, you can’t. In fact, that would get fixed real fast as Uber would just lease your car to you at below cost and take the tax deduction to equal offset. So is this bad for the economy? Ummmm, no. We are getting more marginal labor hours, more aggregate GDP and taxable income, more flexibility for workers, more use out of our installed vehicle base, which likely means more frequent and higher valued car purchases, fewer care note defaults, and using the same exact tax deal that every large business in America gets, just available to you and I Joe Schmoe only if we work more hours! You’ve got to be nuts to think this is a bad deal for the government or the economy. In fact, if Uber is successful enough to pull enough people /marginal work hours out of the workforce – maybe we’d finally see some wage inflation again.
30% of Drivers are Losing Money – Finally the authors claim that 30% of drivers are actually losing money once vehicle expenses are included. Quite possible, but not likely for very long, and it doesn’t likely matter. Two very interesting facts here. All of my analysis above says this is probably just an MIT cost accounting and tax allocation fantasy, nobody is that stupid to believe it except them. Second, they failed to analyze a timeline. I would bet that most Uber drivers who drive for a long time, make good money. Otherwise, they’d have stopped. It is likely that new drivers make less – why, well, some of the new ones aren’t very good, get bad reviews, aren’t efficient at finding riders, don’t have car and gas costs optimized for Uber, basically they are still little startups that haven’t learned the business yet. That’s true in ANY business. So it might be better to analyze, how well does a new driver do, how well does a good, experienced driver do, and how long and what does it take for a new driver to become smart and profitable in which markets. As a fun aside, I was once picked up for $12, 10 minute Uber ride home from a car dealership at 10 pm at night after I failed to negotiate the price I wanted on the car. My wife had dropped me off. The lady who picked me up lived literally yards from the dealership, usually drove Uber only on weekend nights in certain parts of town where she could get surge pricing, easy customers from 11 pm – 1 am before the drunks got out, to cover her auto costs on a very nice car and make going out money. She knew exactly what part of town and when she could make really high hourly rates, and when and where it was marginal. She was a college educated professional. She only took my ride because she was literally 3 minutes from the dealership, told me she was watching TV in her pajamas 5 minutes before, and her calculus, was effectively: $12 in her pocket or watch a bad rerun, right now. Who wouldn’t want to get paid $25-$30/hour plus a tax deduction to not watch bad reruns!
Let’s sum up with an example – In one hypothetical example a $15/hour (2x the minimum wage), employee currently employed 35 hours a week as a second household income (most families are 2 income) and offered a chance to work another 5 hours on Saturday afternoon at that job with a 30 minute commute. They might look at the same facts the authors describe and conclude: My alternative opportunity is $15/hour, less $3/hour for flexibility value (I’d rather work an extra hour a day on the way home, and value flexibility a lot more in my marginal hours as my Saturdays are family time), and after subtracting that and commute time and FICA, I make $9.30/hour cash pre tax or a measely $45/week cash. Screw that. In fact, in this example, the economy loses out on those labor hours and taxes.
Instead I drive 6 hours a week/24 a month for Uber effectively on my commute home or instead of watching TV, shift $300-500/month in car expense to tax deductible, saving up $100+ depending on my marginal tax rate, make $15-20/hour gross revenue, clearing $500-600/month, and keep my Saturdays. All with a total out of pocket of about $50 a month in gas, and costs I was going to spend anyway. That’s not a bad deal. It’s 2-2.5x the driver’s next best alternative, probably better than even overtime were it available, and more realistic given observed driver behavior than a fictional idea of 100,000 Americans choosing to work for $3.37/hr because they are too dumb to realize it until an MIT PhD points it out..
And just wait for driverless cars and more AI algorithms.