Thread by @econ_b

My paper "The Drivers of Social Preferences: Evidence from a Nationwide Tipping Field Experiment," with @UriGneezy, John List, and Ian Muir is now out as an NBER working paper. We analyze over 40 million tipping outcomes on the Uber app in 2017. THREAD

nber.org/papers/w26380

First facts:

16% of UberX trips are tipped

60% of riders who took at least 10 trips never tipped

Conditional on tipping, people tip a little more than $3 on average (about 26% of the fare), so the average tip, including cases where the rider did not tip, is about $0.50

Lower tips on Uber compared to e.g. restaurants could be a consequence of different cultural expectations, particularly since Uber did not encourage tips in its earlier days. Also, the tipping decision in most other contexts is more public, which could add social pressure.

Tips are lowest in big cities in the Northeast and California, and highest in less densely populated areas

A key result: rider characteristics explain 3 times more variation in the amount tipped than driver characteristics.

Knowing who the rider is on a trip is much more predictive of the tip amount than knowing who the driver is.

Rider characteristics positively correlated with high tips:

- a high rating
- less experience

Riders may initially overestimate tipping norms on Uber when they first join, then revise their tipping downward as they get more experience

Using a proxy for rider gender, we find that women tend to tip less than men. We also find that riders who use a default app language other than English tip less.

It's possible that perceived tipping norms or perceived pressure to tip vary by demographics.

On the driver side, we similarly find high ratings and less experience are associated with higher tips.

Declines in tips with experience could suggest that drivers try especially hard on their first few trips before getting comfortable. Or they decide tips not worth the effort.

Women get tipped more than men on average. The gap shrinks with age (controlling for time, location, and other covariates).

Both male and female riders tip women more.

Male riders are more responsive to age of female drivers than age of male drivers.

Men tip younger women seven cents more than older women, tip younger men about 3 cents more than older men. Female riders tend to not vary tips as much based on the age of the driver.

Riders from ZIP codes that are high income and have a lower share of African American or Hispanic residents tend to tip more.

Drivers from low income areas and areas with a higher share of Hispanic residents get tipped less on average.

Tips seem to be positively correlated with trip quality, e.g. on-time pickup, fewer hard brakes and accelerations

Those are some of the big facts in our paper. We also use natural and experimental variation to study social preferences in the field:

What happens when the same rider sees the same driver twice?

They tip 27% more on average the second time.

Two reasonable explanations: strategic reciprocity, or some kind of behavioral or norms response

Under strategic reciprocity explanation, when a rider matches with the same driver twice, they realize that repeat interactions happen more often on Uber than they originally thought.

That means they should tip all of their drivers more to avoid bad encounters in the future.

Testable implication: riders should tip other drivers more too.

Not borne out in the data–only the single trip with a repeat interaction gets tipped more.

More likely repeat interactions lead to higher tips because of social connection built between the rider or driver, or different norms governing repeat interactions and single encounters.

A reasonable conjecture is that conversation leads to greater social connection

An implication: If rider or driver is less comfortable using English, we might expect conversation to be less likely, which could hinder social connection and mitigate the repeat interaction effect.

When the rider or driver uses a default app language other than English, increase in tips from repeat interaction is similar to the result for the general pool of Uber users.

If higher tips happen on repeat interactions because of social connection, this is at least suggestive evidence that conversation is not the mechanism through which social connection is built.

Finally, we study defaults.

Most riders tend to choose one of the default options when they do tip on Uber. We analyze an experiment Uber ran on defaults.

While higher defaults did lead to slightly higher tips, the effect of defaults is much smaller on Uber than what Haggag and Paci (2014) suggests for taxis.

On Uber, tips happen privately, while in taxis tips happen in front of the taxi driver.

Our results suggest that defaults can be more influential in setting tipping norms when the tipper's actions are monitored.

Defaults and monitoring are complements.

I hope you take a look at the full paper if you are interested in learning more!

nber.org/papers/w26380

Also take a look at, "Design and Analysis of Cluster-Randomized Field Experiments in Panel Data Settings," with Hortacsu, List, Muir, and Wooldridge.

nber.org/papers/w26389

We run simulations to study best practices for analyzing cluster-randomized experiments w/ panel data.