Reference Dependent or Independent Labour Supply among Street Paper Sellers

In an intervention directed at sellers of a paper, we randomly reward some sellers per copy they sell.

Our group of subjects are sellers of the Swedish street paper Situation Stockholm (henceforth “Situation Sthlm” or “the paper”). Sellers buy copies from the paper’s office for 40 SEK each and sell them for 80 SEK. The paper is published monthly, except for one summer month. Sellers who have unsold copies when a new issue comes out may return the copies to the paper’s office and get 40 SEK per copy back.

In an intervention directed at sellers of Situation Sthlm, we reward sellers with ten SEK extra per copy they sell. To eliminate the possibility of cheating (e.g., sales between sellers where the issues can later be sold again), only sales made through an electronic payment method known as Swish qualify for the reward. Swish-sales already comprise the vast majority of all sales, and enable us to verify that the buyers are not other sellers, as well as to exclude buyers of multiple papers of the same issue. Sellers know these rules and should expect total sales to be checked against the difference between the number of papers they buy from Situation Sthlm and the number they sell back when a new issue comes out. Participating sellers also sign a contract that promises not to cheat. Participants do not know that they are in a study, having been told that the editorial office of the paper has simply received a donation.

A coin flip randomizes sellers. In concrete terms, a seller looking to buy copies of Situation Sthlm is asked whether he or she would like to flip a coin to determine whether he or she is to receive a ten-crown reward per copy sold through Swish. If the coin comes up heads, he or she gets the reward. Otherwise, he or she gets the reward the next issue.

2024-09-25

2024-11-27

Total number of sold papers of the issue

See the pre-analysis plan for details.

Log of Total number of sold papers of the issue
Total number of sold papers of the issue via the electronic system
Total number of sold papers of the issue via cash
Hours worked selling the issue
Number of days worked selling the issue
Total number of sold papers of the issue

See the pre-analysis plan for details.

Random assignment

Random assignment. 50% of the participants will be treated with the issue released on September 25, and the remaining 50% will be treated with the issue released on October 30.

Coin flip

Individual

No

150 sellers.

150 sellers.

75 treated in month 1, 75 control in month 1.

The sample will contain around 150 individuals, with half assigned to treatment and half to control. We know that the lags have very strong predictive power for the outcome, amounting to an R-square of 0.71, so we take that into consideration. At the conventional level of significance of 0.05, an R-square of 0.7, and a power of 0.8, our sample size would allow for a minimum detectable effect of 0.25 standard deviations.