Valuing the Time of the Self-Employed
How to cite
Baseler, Travis; Dupas, Pascaline; Snowberg, Erik; Agness, Daniel; Chassang, Sylvain. 2019. "Valuing the Time of the Self-Employed." AEA RCT Registry. AEARCTR-0004110. https://doi.org/10.1257/rct.4110-2.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
People's value for their own time is a key input in public policy evaluations---these should account for time taken away from work or leisure as a result of policy. Using rich choice data collected from farming households in western Kenya, we show that households exhibit non-transitive preferences. As a result, neither market wages nor standard valuation techniques correctly measure participants' value of time. Using a structural model, we identify the behavioral wedges in participants' choices, and find that distortions appear when households exchange cash either for time or for goods. Our model estimates suggest that valuing the time of the self-employed at 60% of the market wage is a reasonable rule of thumb.
Intervention
This study is not a randomized trial, and there are no interventions. We use a lab-in-the-field design to elicit farmers' willingness to pay across several numeraires. We randomly assigned which numeraire farmers would pay in in order to adhere to our elicitation protocol (which told farmers that their payment would be randomly chosen), but our main analysis does not exploit this random variation.
Each farmer in our sample was given three choices that used the BDM design of Becker (1964), as implemented in Berry et al. (2020). Participants were asked to state their preferences for some object, for example a lottery ticket for a pump, in some unit of payment, for example, hours of labor. After stating their preferences, a random price was drawn, and if their stated value was higher than the price, that is what they paid for the object. If their value was lower than the price, no transaction occurred.
Choice RW (Reservation Wage): We explained to each farmer that we were offering one-time, 2-hour jobs performing casual agricultural labor in a different village. We asked each farmer whether they would be willing to accept the job at 120 KSh per hour. If they answered "no,'' we asked about their reservation wage directly. If they answered "yes,'' we asked whether they would accept the job at incrementally lower wages until they changed their answer to ``no.''
Choice CB (Cash Bid): We explained to each farmer that we were selling lottery tickets offering 1-in-10 odds of winning a MoneyMaker pump. We collected willingness to pay in cash by asking the farmer whether they would be willing to pay a low price of 20 KSh, and then asking the same question for increasingly higher prices, until the farmer declined the offer.
Choice TB (Time Bid): As in Choice CB, we explained to each farmer that we were offering lottery tickets with 1-in-10 odds of winning a MoneyMaker pump. We collected willingness to pay in time by asking the farmer whether they would be willing to work 30 minutes for the ticket, and then asking the same question for increasingly higher amounts of time, until the farmer declined the offer.
Experimental design
We selected 18 villages for our sample from a set of villages sampled for a separate project which auctioned off Kickstart irrigation pumps. We selected all control villages that had not received any pumps, and used remaining pumps from that project to elicit willingness to pay in cash and time.
Each village was randomly assigned (by a pseudo-random number generator) to one of three groups: Cash, Cash + Day Work, or Task. Farmers in Cash villages received a lottery ticket price payable in cash only, and were not eligible for wage work. Farmers in Cash + Day Work villages received a lottery ticket price payable in cash only, and were eligible for wage work. Farmers in Task villages received a lottery ticket price payable in hours of work only, and were not eligible for day work.
Our main analysis does not exploit this random variation.
Design details
Cited by
No citation data available for this registration yet.