Uber versus taxi: A driver’s eye view

Josh Angrist, Sydnee Caldwell, Jonathan Hall 08 December 2017

a

A

In Boston and most other big US cities, taxi drivers must hold a medallion granting them the right to drive. Medallions are issued by city agencies like New York City’s Taxi and Limousine Commission. Medallions can be bought and sold, but because they are limited in number, they command a premium price. Boston’s Hackney Carriage Unit, for example, has issued only 1,825 medallions, a number that has been fixed since 1990, when it was increased from the original total of 1525 set in 1930. You need a medallion to drive a cab in Boston; even suburban medallion holders return empty from trips to Boston’s Logan airport (which is inside Boston city limits).

Not unusually for the US cab industry, some Boston drivers own a medallion, but most lease from someone else. Medallion owners may be investors, fleet owners, or former drivers who no longer wish to drive. In Boston in 2013, only 453 medallions were associated with owner-operated cabs. In Boston and elsewhere, most big city drivers lease by the shift (typically 12 hours) or weekly. After paying the up-front lease cost, drivers keep all trip receipts and tips collected while driving. As you might expect for an asset that until recently traded for around $700,000, leasing is expensive – before ride-share companies entered the market, Boston taxi drivers paid approximately $700 for a weekly lease. Even so, on the margin, a Boston cabbie drives for himself.

Rideshare companies such as Uber and Lyft have changed the taxi market dramatically. For one thing, the rideshare revolution reveals the extent to which the medallion system has restricted supply – in the summer of 2016, Uber alone had almost 20,000 active Boston drivers. Rideshare has undoubtedly made urban transportation cheaper and more convenient for millions of riders (as suggested by analyses in Cohen et al. 2016, and Hall and Krueger 2017). Should the drivers who provide urban transportation be happy too?

It’s all about that lease

Rideshare entry increases the number of drivers competing to take you to the airport or elsewhere, so cab driver income may have fallen as a result of increased supply of ground transportation. This depends in part on whether drivers or medallion owners get the rents from limited medallion supply. In any case, however, rideshare has something to offer drivers as well as riders. In a recent paper, we argue that the most important economic difference between Uber and taxis from a driver’s point of view is the need for medallion leasing or lack thereof (Angrist et al. 2017). The work done when driving for Uber or Lyft is similar to the work done when driving a taxi (driving people around town), and in both scenarios drivers choose their hours freely. But rideshare drivers needn’t pay up front for the right to drive. Rather, they pay a proportion of their trip receipts to Uber or Lyft, a tax on their earnings that Uber refers to as the ‘fee’.

Drivers who drive only a few hours should find leasing unattractive, but high-hours drivers earn more by paying a fixed lease rather than proportional fees. We use this insight to compare the economic benefits and costs of rideshare work and traditional taxi driving. Specifically, we ask how much Uber drivers must be compensated for the loss of rideshare work opportunities if the goal is to leave them as well off as they were when they were driving for Uber. Our comparison uses the economic concept of compensating variation or CV. CV is the cash payment that makes an Uber driver whole in a world where Uber disappears. For some high-hours drivers, compensation is negative – taxi is better than Uber. But most Uber drivers drive part time, and will therefore be worse off under compensation schemes that require a sizable lease. These drivers require considerable compensation to make-up for the loss of rideshare earnings opportunities.

The contrast between Uber and taxi is a metaphor for all sorts of work and pay arrangements. Taxi-style fixed-lease contracts are common in many settings. For example, many hairdressers and beauticians rent chairs in a beauty shop by the day or week. On the other hand, many franchise owners pay fees proportional to sales to the owners of a product or brand. In some occupations, workers may negotiate a combination of proportional fee and fixed-lease-type arrangements in return for the right to pursue the earnings opportunities in their field. Our Uber-versus-taxi analysis provides an economic framework that can be used to evaluate and compare these sorts of arrangements.

Contractual matters

Two parameters govern the rideshare–taxi face-off. The first is the driver labour supply elasticity. This elasticity tells us how much more drivers work in response to higher pay. Drivers who are unit elastic, for example, drive 10% more in response to 10% higher wages. Labour supply elasticities are relevant to the rideshare-taxi comparison because (other things equal) a taxi driver’s hourly wage is increased by not having to pay the rideshare fee. If the fee is 25%, for example, as it is for most Boston Uber drivers, fee removal raises wages by one-third, since drivers now keep a dollar of every fare dollar earned instead of 75 cents. Drivers who are highly responsive to pay benefit from this higher wage by driving more. 

The second parameter governing CV is something we call lease aversion. This captures the extent to which drivers are averse to the gamble implicit in the decision to buy a lease. Our analysis shows that many drivers who would likely benefit from leasing nevertheless fail to buy one. We describe these drivers’ behaviour by scaling nominal lease rates up to a number that explains our data. A driver who has a coefficient of lease aversion equal to 1.5, for example, responds to a lease offer of $100 as if it costs $150.

Labour economists have long debated the labour supply elasticity of cab drivers. Some have even said this key economic parameter is negative. For example, behavioural economists like Camerer et al. (1997; see also Thaler 2015) have argued that cab drivers actually drive less when their pay is high. This is said to result from ‘target earning,’ a pattern in which drivers are seen as setting a daily income target and then quitting when they reach it. Labour economists like Farber (2005) are sceptical of such claims because this behaviour is irrational and immiserating. Target earners end up much poorer than drivers who rationally exploit the low-hanging fruit of temporarily high fares on a rainy day or in response to busy events that boost demand for transportation.  

We estimated the rideshare labour supply elasticity by randomly assigning reductions in the Uber fee to a large sample of Boston Uber drivers (this experiment was pitched to drivers as an Uber promotion called the Earnings Accelerator).  Fee reductions amount to an increase in pay.  At the same time, we gauged the importance of lease aversion by offering randomly chosen drivers a virtual medallion – in return for paying an up-front payment of, say, $100, treated drivers avoid the Uber fee for a week. The elasticity and lease aversion estimates generated by our experimental treatments are then used to compute the CV required to convince Boston Uber drivers to lease a medallion and drive a taxi. Drivers who opted in to the Earnings Accelerator mostly did well as a result, saving an average of 126 dollars in fees after subtracting their lease costs.

Driven drivers prefer taxi

Our experiment reveals that drivers respond sharply to changes in their hourly earnings. A 10% increase in hourly earnings causes drivers to drive about 12% more (Boston Uber drivers are more than unit elastic). As can be seen in Figure 1, average driver behaviour in the week before our randomised fare increase (fee reduction) and the week after the increase was unchanged. This suggests the driver response to higher pay is consistent with the intertemporal substitution hypothesis, which says that workers faced with a temporary wage increase should work more when pay is high, leaving their labour supply in other periods unchanged. Figure 2 shows that the entire distribution of hours driven shifts up when fares increase; we find no evidence of behaviour suggestive of target earning.

Figure 1 Participation effects on labour supply: Opt-in week

Note: This figure reports treatment effects on hours, earnings, and an indicator of any Uber activity for drivers who opted in to the Earnings Accelerator. Reported are estimates for drivers who accepted the opportunity to drive fee-free. Effects are computed by instrumenting experimental participation with experimental offers as described in the text.

Figure 2 Distribution treatment effects: Opt-in week (fee-free driving)

Note: This figure reports estimated CDFs of potential hours driven in treated and non-treated states for drivers who participated in the Earnings Accelerator. Shown are estimates for drivers who accepted the opportunity to drive fee-free during the opt-in week. CDFs are estimated by instrumenting participation with experimental offers as described in the text, using a grid of 200 points. CDFS are smoothed using a five-point moving average.

The strong labour supply response to higher fares favours taxi. Drivers who are inclined to take advantage of higher fares benefit from the higher wages earned without the Uber fee.

Lease aversion

Offers of a taxi-style compensation contract reveal the extent of lease aversion. Drivers who drive a lot should prefer taxi because increased earnings from fee reduction more than cover the cost of our (inexpensive) virtual taxi medallions. Our taxi experiment reveals, however, that many high-hours drivers who would have benefitted from a contract that collects a fixed amount in place of the Uber fee failed to take advantage of this. Driver decisions as to whether to buy virtual medallion leases seem to be explained by a constant multiple of 1.5 – opt-in decisions are those we’d expect when drivers behave as though the offered lease is 50% more expensive than it actually is.

We interpret lease aversion as an expression of something behavioural economists call loss aversion. Gamblers are loss averse when their decisions to place bets are coloured by a tendency to suffer losses more heavily than they enjoy gains. Drivers offered a taxi contract must gamble that fare opportunities will be sufficiently strong to put them ahead of the default scenario in which they pay the proportional Uber fee. A world in which the pain of losing $100 exceeds the utility of $100 gained explains our data well. Perhaps surprisingly, simple risk aversion does not do the trick. Economists have repeatedly found that people are not nearly risk averse enough for risk aversion alone to explain the sort of behaviour we observe (primarily because the amounts involved are too small).

Compensating taxi

Our experimental results suggest that Uber drivers must be paid an average of $437 to move to a $400/week leasing scheme with a 25% higher wage. This is average CV for Uber drivers who now pay a 25% fee and are offered a $400/week lease (historical lease rates are higher, but rideshare is pushing these down). Without this additional compensation, only 2% of drivers would prefer fee-free leasing. The CV amount of $437 is a measure of the average financial gain realised by Uber drivers from the opportunity to drive lease-free. Simply put, most Uber drivers benefit from the opportunity to provide rideshare services in an amount that far exceeds their take-home pay. Our CV calculation implicitly maintains the level of ground transportation services provided; riders are just as well off either way.

Broader implications

The rise of the gig economy has given workers the opportunity to work under a variety of hours and compensation arrangements. Our analysis suggests most Uber drivers benefit from not having to pay up front for the right to drive. But we are not the first to discover this – the New York City Taxi and Limousine commission is all over it, having recently piloted an alternative proportion-to-fares leasing in place of the traditional fixed-fee arrangement.

References

Angrist, J, S Caldwell and J V Hall (2017), “Uber vs. taxi: A driver’s eye view”, NBER, Working Paper 23891.

Camerer, C, L Babcock, G Loewenstein and R Thaler (1997), “Labor supply of New York City Taxi Drivers: One day at a time”, Quarterly Journal of Economics 112(2): 407-441.

Farber, H S (2005), “Is tomorrow another day? The labor supply of New York City cabdrivers”, Journal of Political Economy 113(1): 46-82.

Farber, H S (2015), “Why you can’t find a taxi in the rain and other labor supply lessons from cab drivers”, Quarterly Journal of Economics 130(4): 1975-2026.

Fehr, E and L Goette (2007), “Do workers work more if wages are high? Evidence from a randomized field experiment”, American Economic Review 97(1): 298-317.

Hall, J V and A B Krueger (forthcoming 2017), “An analysis of the labor market for Uber’s driver-partners in the United States”, Industrial and Labor Relations Review.

Kahneman, D, J L Knetsch and R H Thaler (1991), “Anomalies: The endowment effect, loss aversion, and status quo bias”, Journal of Economic Perspectives 5(1): 193-206.

Oettinger, G S (1999), “An empirical analysis of the daily labor supply of stadium vendors”, Journal of Political Economy 107(2): 360-392.

Sydnor, J (2010), “(Over)insuring modest risks”, American Economic Journal: Applied Economics 2(4): 177-199.

Thaler, R H, A Tversky, D Kahneman and A Schwartz (1997), “The effect of myopia and loss aversion on risk taking: An experimental test”, Quarterly Journal of Economics 112(2): 647-661.

Wallsten, S (2015), “The competitive effects of the sharing economy: How is Uber changing taxis?”, Technology Policy Institute.

a

A

Topics:  Industrial organisation Labour markets

Tags:  rideshare, Uber, Lyft, taxi, taxis, taxi drivers, lease, lease aversion, loss aversion, disruption, gig economy, compensating variation

Ford Professor of Economics and co-director of the School Effectiveness and Inequality Initiative, MIT

PhD candidate, MIT Economics Department

Chief Economist and Director of Public Policy, Uber

Events