Incentives and Natural Experiments

People respond to incentives

Incentives have side effects

Why before-and-after misleads

Natural experiments and difference in differences

A city offers ride-sharing scooters a 1 dollar per ride subsidy, paid to the scooter company. Which outcome is the most predictable first-round response, before any second-round adjustments?

A town builds a new tram line and afterwards reports that its downtown retail sales rose 6 percent compared with the year before. The town credits the tram. What is the most serious weakness in this claim?

Tram town retail sales rose from 200 to 224 million dollars after the new line opened, while sales in an otherwise similar town with no tram rose from 200 to 206 million dollars over the same period. Compute the difference-in-differences estimate of the tram effect in millions of dollars and write the number.

A daycare frustrated by late pickups introduces a fine per late minute, and late pickups increase rather than decrease. Which explanation fits?

Treated city sales rise from 100 to 104 while control city sales rise from 100 to 101. What is the estimated effect of the treatment?

Treated town retail sales rise from 12 to 15 million dollars, while comparable control town sales rise from 10 to 11 million. Compute the difference-in-differences estimate in millions, and write the number.