Measuring Risk and Time Preferences Among the Urban Poor in Saudi Arabia

2019-01-27

2019-04-13

The risk and time preference parameters of poor urban households in Dammam

This study is based on a socio-economic household survey conducted with 496 households randomly selected from poor neighborhoods across the city of Dammam and an incentivized lab-in-the-field experiment conducted with a random subsample of 166 respondents.

The design of the lab-in-the-field experiment to elicit risk and time preferences is based on Tanaka et al. (2010) and has been applied by other studies, such as those conducted by Nguyen (2011), Liebenehm and Waibel (2014), and Ackert et al. (2020). More specifically, the lab-in-the-field experiment consisted of two parts: one to elicit participants’ risk preferences, and one to elicit their time preferences. Both parts were programmed in Survey Solutions, a software solution developed by the World Bank for computer-assisted interviews (https://mysurvey.solutions). In the risk experiment, participants made 35 choices between a risky and a less risky lottery. The probabilities of the lotteries were illustrated on the tablet with two colored balls. To further help participants to visualize the decision, interviewers showed participants an actual bag with different colored balls at the beginning of the experiment. The first 28 choices included only positive payoffs. The last seven choices also included negative payoffs to measure participants’ loss aversion. However, since participants received a fixed payment for taking part in the household survey, the possibility that participants suffered any overall loss for taking part in the experiment was excluded for ethical reasons. After completing the risk experiment, the time experiment was carried out. In the time experiment, participants were asked to make 75 choices between receiving a smaller amount today or a larger amount at various times in the future. After respondents had completed both experiments, 1 of the 110 choices was randomly selected, and the participant received a monetary reward according to the decision they had made.

The selection of the urban poor household sample in the city of Dammam followed a two-step procedure. In the first step, with the help of insider knowledge from charity organizations, we identified nine neighborhoods with a relatively high share of poor Saudi households. All the poor neighborhoods were located in central Dammam and were often referred to by the local community as “the old neighborhoods.” In the second step, we determined the dwellings within those nine neighborhoods through satellite image data and randomly marked every 10th dwelling to be interviewed. Since the study focused on Saudi nationals, dwellings occupied by foreigners were excluded. In total, 496 Saudi households were interviewed. The estimated baseline population in the nine neighborhoods was approximately 60,000 households or 300,000 people, of which 20,000 were Saudi households (130,000 people) and the remainder were foreign households.

In order to select participants for the lab-in-the-field experiment, we randomly selected every third household that was interviewed in the household survey. Hence, our analytical sample consists of 166 households that took part in both the household survey and the experiment. The household characteristics that we obtained from the household survey were used as control conditions in the study, whereas the experiment was used to measure households risk and time preferences.

Household Head

Yes

166 household heads that participated in both the household survey and the lab-in-the-field experiment

18,260 observation (110 choices per household head)

The sample size consist of 166 household heads and all household heads took part in the same experiment. However, due to the natural variation in the study area, some households were poor and others non-poor. In total, there were 58 poor and 108 non-poor households.