So what makes people do socially oriented tasks better? An interesting new paper by Nava Ashraf, Oriana Bandiera, and Kelsey Jack shows that money doesn’t matter and recognition makes a big difference.
The setting: Lusaka, the capital of Zambia, where there is one hair salon per 880 people (in case you were wondering, there is about 1 doctor per 14000 people in Zambia). Ashraf and co. are working with an NGO that is distributing female condoms via the hair salons (HIV prevalence in Zambia is 14.3%). The owners of the salons are invited to become volunteers who sell these condoms. The idea is that this is a somewhat new and less accepted technology and here are some folks who have a captive audience.
The treatment: these hair salon owning/managing volunteers are invited to a training on selling these condoms and at the end are assigned to four treatment groups:
1. Regular volunteers – they sell the condoms as per usual
2. The low powered incentives – these folks get a 10% margin on their condom sales (50 Kwacha, or roughly a penny)
3. The high powered incentives – these folks get a 90% margin on their condom sales (about 9 cents)
4. The “star” treatment. These folks got no money, but instead a thermometer, which could be posted on the wall and then got a star for each pack. If they sold over 216 packs, they got invited to a special ceremony (only 1 person did).
In addition, at the end of training, everyone was invited to play a modified dictator game. That is, in addition to the $8 show up fee, they were told that extra money to the tune of $2.50 had become available and they had a choice of keeping it or donating it to an HIV charity (all done in private). This gives them a proxy for motivation for the cause.
One interesting note on the setup. One of the things you might be worried about is spillovers. So what these folks did was they took a census of all of the hairdressers, including GIS locations. They then took the GIS information and laid it out on a grid of the city divided up into 650 sq meter squares. They then created buffer zones of 75m all around. Then they assigned treatment groups based on these grids (matching for characteristics), with no treatment being assigned to salons that fell into the buffer zones.
Data: One of the interesting features here is that a lot of the data comes from administrative data the program would have collected anyhow. The sales data comes from restock inventory data from the program’s distributors. They complemented this with a fairly short baseline census/survey (35 minutes) and a follow up with the hair salons and a short customer survey.
The results: Here is where it gets really interesting. It turns out that the folks who got the financial incentives – be they low or high powered did not sell any more condoms than the control volunteer group. Those stylists/salon managers who got the star treatment (recognition), sold twice as many: 14 vs. 7 for the other groups during the year long period of the program. In addition, those who contributed above the median amount in the dictator game also sold more (3.36 more), showing the added boost from being motivated by this issue in the first place. Indeed, this initial motivation is well leveraged by the star treatment – folks who donated above the median and were in the star group sold 10 packs, compared to those in the star group who donated below the median selling 4.3 packs (this difference is significant at the 10% level).
They also give us some insight into how this is happening. It turns out folks in the star treatment were more likely to post promotional materials in their salons and kept better records as compared to those in the other treatment arms. Customer response wasn’t being driven by the thermometer which measured individual sales – not only did very few customers report seeing this thermometer but Ashraf and co. distributed placebo thermometers (!) with average sales for everyone. Another interesting twist is that the star effect seems to have had more of an effect, the more peers there were in the neighborhood (they can use the variation in the density of salons, controlling for a bunch of things, to do this). So the star effect seems to be accentuated through some sort of comparison with neighbors.
So, in the quest to get better services out there, this gives us some lessons to think about. A couple of things to keep in mind. First, these hair stylists were volunteers in the first place. So maybe that’s why the financial incentives didn’t matter (but, hey, these are entrepreneurs after all…). Second, the task at hand, selling female condoms, is a side (social) business. But, at the end of the day, this really gives us something to think about and to take to other public service settings to try further. Any thoughts, ongoing work, or other interesting results in this vein out there?