Experimental Design
** Setting and Recruitment:
We propose to test our hypothesis with low-income workers in rural Odisha, one of India's least developed states. We selected the initial target villages using the following process. We sample participants from different social groups and analyze if the needs for flexibility differ across them. We aim to survey up to 10 participants from each of the two main social categories living in different hamlets in a village. Using the 2011 Census data, we identify villages with at least 40 households across two major social categories and at least 15 households in each of those social categories. In addition, we restrict our sample to those villages that have no more than 250 households, more than 15 km away from urban centers, and have more than 10 households that are engaged in casual labor.
After the initial selection, the recruiting team visits the target villages to verify and update the above information. Provided that the two major social categories are living in separate hamlets, we further restrict to villages that have physical demarcation dividing the two. We identify suitable work site locations that are close to a large cluster of target villages.
We conduct the study in villages that are within 10km of the worksites. In each sample village, local staff members visit every household in the two largest hamlets and explain that there is the opportunity for one male member to participate in a survey. Using a short screening questionnaire, we identify all male household heads who are aged 18-55 and currently residing in the village, who are not in regular employment (e.g. salaried work) , and who are interested in a six-week casual wage job. We randomly select a maximum of 10 people per hamlet (20 per village) to participate in the survey.
** Procedures:
Identifying the Willingness to Pay (WTP) for Flexibility: The choice experiment consists of two parts. The first part begins with survey questions about household characteristics and work experience over the past 30 days, including reasons for work absences. We will document the number of workdays missed by participants and categorize reasons for these absences into the following categories:
- Social duties (e.g., weddings, funerals, festivals, helping neighbors)
- Household duties (e.g., household business, child-rearing, firewood collection)
- Farm work (e.g., on their own or for close relatives)
- Illness
Thereafter, we will conduct a Becker-Degroot-Marschak (BDM) exercise involving a real 6-week employment opportunity. All job offers involve producing a basic household item at a manufacturing worksite, with a prevailing daily wage and a lump-sum attendance bonus at the end of six weeks. The offers vary only in terms of the attendance requirement, bonus amount, and distance to the village, as described in detail below. All participants will be informed that one of their choices will be implemented with a small probability, meaning there are real stakes associated with their decisions.
In the second part, we will conduct a survey module that captures norms around attending social events, such as the size and frequency of events, participation rates, and the participants’ dependence on social networks to further explore the role of social duties. Network dependence may stem from needs such as financial distress insurance or concerns related to reputation or self-image. We will assess the relative importance of these factors, noting the potential differences across agricultural lean vs. peak seasons. In addition, we will conduct various incentivized exercises to measure risk aversion and present bias to understand how behavioral factors unrelated to the four main categories outlined above affect WTP for flexibility.
** Illustration of the BDM Exercise:
* Contract Type 1 - Fixed Contract
This work arrangement requires the worker to show up every day, except for some fixed days. The worker can specify one day per week (e.g., Sunday) to take off without consequence. They can still work on these fixed holidays and earn wages, but if they take off any other day, they will not receive the lump-sum bonus at the end.
An alternative version of the fixed contract allows workers to pre-specify all the days they are willing to take off at the beginning of the work period. These pre-specified holidays do not have to fall on the same day of each week. This version helps us gauge the importance of expected versus unexpected causes of absences.
* Contract Type 2 - Flexible Contract
This work arrangement allows workers to take off days flexibly, up to a certain number, without having to pre-specify which days they want to take off. For example, a flexible contract with 6 days off allows them to take any 6 days off without consequence. If they exceed this limit, they will not receive the lump-sum bonus at the end.
* Worksite Type - Close vs. Distant
The same job and pairs of contracts are offered at two types of worksites. The close worksite is within commuting distance from the participants’ villages, while the distant worksite requires more than 3 hours of travel, preventing daily commuting. For the distant worksite, the employer provides transportation, lodging, and a lump-sum relocation payment. By comparing workers’ contract choices across the two worksites, we can determine whether workers value flexibility more when they are physically closer to their networks.
** Contract Choice
Workers will review multiple pairs of contracts and choose their preferred option. The main comparisons include:
* Close Worksite
- Fixed contract with 6 days off vs. Flexible contract with 6 days off
- Fixed contract with 6 days off vs. Fixed contract with 12 days off
- Fixed contract with 6 days off vs. Flexible contract with 12 days off
- Pre-specified leaves (6 or 12) at the beginning of the contract vs. Flexible contract with (6 or 12) days off)
* Distant Worksite
- Fixed contract with 6 days off vs. Flexible contract with 6 days off
- Fixed contract with 6 days off vs. Flexible contract with 12 days off
The offers are structured such that one contract offers the same fixed bonus amount (e.g., USD 40), while the other contract involves a price list with different bonus levels (ranging from USD 1 to USD 40). The order of contracts and bonus amounts (ascending or descending) will be randomized across participants. After the choices are made, we will determine if a participant receives a job offer through a lottery. If selected, there is a small chance that the offer chosen by the participant during the BDM exercise will be implemented. The choice experiment allows us to address the first two questions by measuring workers’ true WTP for flexibility and investigating factors that determine their decisions.
** Job Implementation
During the job implementation phase, we randomize participants into different types of contracts to understand the impact of work arrangements on workers’ job take-up, attendance, and productivity. For a subset of selected workers, instead of implementing their choices during the BDM exercise, we initially offer them a fixed contract which allows them to take off one fixed day per week (e.g., every Sunday) without losing the bonus, and measure their contract take-up.
Next, for a randomly selected 50% of participants, we upgrade their contract to a flexible one, allowing them to take off any 12 days. For those who refuse the rigid contract in the first place, we still upgrade them to a flexible job and again measure take-up. At the worksite, we measure daily attendance and productivity over six weeks, and administer weekly questionnaires to understand reasons for any absences.
This procedure allows us to (1) estimate the impact of contract type on job take-up in order to understand worker selection into formalized jobs; and (2) estimate the treatment effect of flexibility on productivity and earnings among those who were willing to take up the fixed contract, meaning the unobservable worker type is held constant. The job implementation phase will inform trade-offs firms face when offering different contract types.