Under pressure: the impact of women's mental load on labor productivity and occupational choices. Evidence from Kenya.

We want to examine whether women's mental load, i.e., the management behind household activities, negatively affects cognitive abilities and labor productivity through a lab-in-the-field experiment. The treatment consists of priming in the mind of the participants thoughts related to mental load. They will then be asked to perform an effort task for 30 minutes during which they will have to divide black from red beans, and to complete as many Towers of Hanoi as possible with 4 and 5 disks. In the last slot, they will be asked to choose whether they rather perform the Tower of Hanoi with 5 disks or the beans again. This last information will be used to test whether mental load leads individuals to self-select into less cognitive-demanding tasks (i.e., sorting beans) because their mind is already overloaded by thoughts related to the management of the household activities. We will also test whether mental load leads individuals to make an income-maximizing choice based on information they have from the previous two rounds.
The choice of the two effort tasks is slightly gendered: while dividing black from red beans is a gendered task for women, and requires low cognitive skills, the Tower of Hanoi is slightly gendered towards men.

2022-05-18

2022-06-16

There are three primary outcome variables of interest: labor productivity, self-selection into less cognitive-demanding jobs, and optimal choice of income maximization based on previous productivity levels.

Labor productivity will be measured as the income earned by the participants in the effort task. These two different tasks allows also for disentangling the impact of mental load on more automatic tasks (i.e., dividing beans) and on more proceduralized tasks (i.e., the Tower of Hanoi). The second outcome of interest will be a dummy variable which will be equal to 1 if the participant chooses to perform the Tower of Hanoi with 5 disks, and 0 if she chooses to sort beans in the third round of the effort task. The third outcome of interest will be a dummy variable equal to 1 if the individual choose the income maximizing option in the third round or not. Based on the productivity levels in the first two rounds, we will build the dummy variable which is equal to one if the participant makes an irrational choice, i.e., if she/he chooses for the third round to perform the task in which she/he was less productive during the first two rounds of the effor task.

Secondary outcomes will be the average number of moves needed to complete the Towers of Hanoi and the average time needed to complete each Tower. We will also look at the grams of beans divided in each round, and the separate income of each task.

Participants in the treatment and in the control group will be exposed to the same activities but in different orders. Participants in the treatment group will be first exposed to the treatment before being asked to perform an effort task. After the effort task, they will be exposed to a Digit Span Task to measure their attention level, they will have to self-assess their own stress level, and they will finally answer basic socio-economic questions. Participants in the control group, on the other hand, will be asked to compile the socio-economic questionnaire first, and then to perform the effort task. After the effort task, they will be exposed to the Digit Time Span and they will have to self-assess their own stress level. They will also have to complete a simplified version of the Raven's matrixes. Finally, they will be exposed to part of the treatment.
The treatment consists of three parts: first, participants will have to compile a time use module under the form of time diaries; then they will be asked to recall their most pressing worries and thoughts related to the household; and finally they will have to link these thoughts and worries within the schedule of a typical workday. To do this last exercise, they will have to look first to a short video with role models (i.e., inhabitants of Kibera) talking about their worries and thoughts in relation to a normal weekday. The scope of the video is twofold: first, it should trigger thoughts related to mental load by having participants identify with the subjects in the video; second, it should help participants to better understand the nature of the exercise we are asking them to engage in.
The effort task will last 30 minutes. The 30 minutes will be divided into three-time slots of 10 minutes each. During the first two slots, participants will have to divide black from red beans and to complete as many Towers of Hanoi as possible with four disks. For the last time slot, participants will have to choose whether they would rather complete as many Towers of Hanoi as possible with 5 disks or divide black from red beans again.

To ensure that all participants spend the same amount of time in the experiment, they will all be exposed to the same activities. Participants in the control group are asked to complete the socioeconomic questionnaire prior to the effort task to make sure that any potential differences in the outcomes of interest are not due to differences in cognitive fatigue. On average, the socioeconomic questionnaire takes the same amount of time as the treatment.
We identify two potential mechanisms through which mental load can impact productivity. The former operates through the attentional constraint: because attention is a limited resource, we hypothesize that mental load, by charging individuals with constant daily thoughts about the management of household activities, reduces their attention at the workplace and, consequently, their productivity. The latter channel involves stress: mental load can increase stress levels, which can in turn either increase or decrease productivity. To test for these two mechanisms, after the effort task we collect data on self-reported stress, by adopting a stress scale validated in Sub-Saharian countries, and we ask participants to complete the Digit Span Task, a psychological task widely used to measure working memory and sustained attention. We randomize the sequence of the modules for each participant to ensure that the structure of the questionnaire has no effect on the responses to these questions. We will then use this information to check whether these two factors mediate the impact of mental load on labor productivity.
Participants in the control group will also be required to complete a simplified version of Raven's matrixes. We will use this information to check for correlations between the performance in the Tower of Hanoi and individuals' cognitive abilities, as well as to check whether higher cognitive functions predict the endogenous choice into a more automatic task (i.e., sorting beans) or into a more proceduralized and more cognitive demanding task (i.e., the Tower of Hanoi with 6 disks). We do not expose participants in the treatment group to this module for two main reasons. First, we are not interested in checking whether mental load reduces fluid intelligence, but rather attention. As we already ask them to perform the Digit Span Task, we do not want to overload them with the Raven's matrixes. Indeed, participants in the treatment group will be exposed to a series of activities that require focus, concentration, and are more emotionally charging than participants in the control group. Second, a reasonable concern could be that, as the Tower of Hanoi is correlated with individual problem solving skils, we would need to control for that in our empirical analysis. To be able to control for fluid intelligence and problem-solving skills we would need to collect this information before any other experimental activity. However, this could impact the performance in the effort task, as the Raven's matrixes can increase cognitive fatigue. As randomization should ensure that cognitive abilities are normally distributed in our sample, we decided not to include this module for participants in the treatment group.

Randomization done from the office through phone calls.

The unit of randomization are individuals. We sample individuals with children under the age of 10, living in urban areas in Nairobi.

No

We will sample a total of 700 individuals, divided into equal shares between women and men.

700 individuals.

We will have 350 participants in the treatment group and 350 participants in the control group. Participants will be randomly assigned to the treatment or control group by the survey instrument.