Two sided asymmetric information in labor markets.

The study will be conducted in urban areas of Patna, the capital of the state of Bihar, which is one of the poorest in the country. Casual workers in the city, and in other parts of India, often seek work at labor stands. These stands are public places which are spread across the city. Almost all workers seeking work at these stands work in construction or related industries. They arrive at the stand early in the morning and stay there till noon, expecting work, and firms hire them from the stand and take them to the work site. Firms hire workers from these stands. The contracts between the firms and workers range from a single day to several days.

Workers and firms do not know each other. The contractual environment is one of limited commitment and contracts are oral. Thus it is important for the worker to know if the firm is trustworthy and will not renege on its word at the end of work day. Similarly, firms are unaware of the productivity of workers. Ideally, firms and workers would want to engage in a long-term contract as it reduces search costs. However, asymmetric information about each other's type can prevent this from happening. Liquidity constraints on the both the firm and worker side can exacerbate this issue. Additionally, high levels of unemployment can allow firms to exploit workers by asking them to work for longer than they contractually agreed. Hence, workers would want to know ex-ante whether the firm is of the type which tends to exploit workers. This may result in workers not taking up contracts which allow firms to extract excessive labor from them.

HYPOTHESIS

Worker side
H1) Workers are less likely to take up jobs which offer back-loaded contracts.
H2) Offering insurance against wage-theft can increase the take-up of jobs by workers
H3) Liquidity constraints prevent workers from taking up back-loaded contracts.
H4) Workers prefer short contracts than long contracts and this is driven by risks of wage theft being higher in long contracts.

Firm side

F1) Firms prefer back-loaded contracts to contracts which pay workers daily.
F2) Firms prefer longer contracts to short contracts
F3) Alleviating firms liquidity constraints leads firms to hire more workers.

Worker-firm match

M1) Workers work for shorter hours in daily wage contracts
M2) Worker absenteeism is higher in daily wage contracts.

To test these hypothesis we recruit workers in several labor stands in Patna. Workers are offered contracts (which are randomized at an individual level) that vary the length, the payment structure and insurance.
We work with firms recruited through snowball sampling. We use a BDM procedure to elicit the preferences of firms for contracts. Each firm is offered to hire workers on 12 to 16 different types of contracts, one of which is implemented randomly.
Firms and workers which accept the respective contracts are matched with each other. We will measure productivity of workers after they are matched with the firm.

We list the treatment arms in detail and our hypotheses in comparing these treatment arms in the 'experimental design' section of this filing.

2024-02-01

2025-10-01

1) Job take-up on the worker side
2) Take-up of contracts on the firm side

We will construct the worker type variable in two ways and analyse how high workers differ from low-type workers in taking up contracts which are back-loaded (we hypothesize that high type workers are less likely to take up contracts which have wage theft risks).
1) Workers who work on fixed work contracts will be classified as high types as these contracts are highly valued
2) Workers who earned earned higher average income in the days before the survey will be classified as high type

Firms will be classified as large if they have higher number of construction sites/labour working under them. Firms which are managed by former masons might be considered as small.

1) Type of worker and take-up of jobs by different types of worker
2) Type of firm and acceptance of offers by different types of firms
3) Worker productivity at the work site
4) Worker-firm disagreements.

The workers are cross-randomized into treatments which offer a job, with or without insurance against wage theft, and with or without back-loading of wages.

The firms are offered contracts which provide compensation against worker separation or not, credit to pay workers or not, and with or without back-loading of wages.

Firms and workers which accept the contracts are matched with each other.

Throughout the experiment workers will be paid the going W for each day of work.

Workers are cross-randomized into following treatment arms:
T1: We offer contracts of different lengths (3 days vs 7 days)
T2: The contract will pay worker wages W daily or back-load the wage into latter parts of the contract. The total payment will remain the same.
T3: Contracts are either insured or not against wage theft by employers. We will provide zero pay insurance in the form of a written agreement.

We may include a treatment arm where we offer to hire workers in groups of 2 friends. An increase in take-up of this treatment would indicate that workers supply labor together to insure themselves against wage theft.



Each worker will be offered one or two contracts out of the 12 listed above. These contracts would be chosen randomly.

Firm side

We offer to hire one (and may hire more) workers for each firm.
Cross-randomized treatment arms will be:
T1: (daily wage) Firms will have to pay wages daily or back-loaded wages (smoothly or steeply)
T2: (Credit) firms will be provided credit to pay workers.
T3: (Guarantor) firms will be provided information about the worker before they are hired and compensation if the worker fails to turn up. They would be provided a written contract with all these guarantees listed.

Within each firm we will offer contracts for either 3 days or 7 days. This gives us a total of 24 types of contracts. We will elicit preference of firms for each of these contracts using an incentive compatible procedure for a set of 12-16 contracts.

Later, we might vary the the type of worker being offered (low vs high type).
Firms and workers which accept the contracts are matched with each other.

Worker side: A Unique ID with treatment cells randomly allocated to them is created in the office. Enumerators recruit the workers in the order of unique id.
Firm side: We randomize the order in which questions are asked. This is done in the office.

Randomization on the worker and firm side is done at the individual level.
Workers and firms which accept contracts within a treatment cell are matched with each other randomly to each other.

No

We will cluster at the individual level for the firms as we observe them hiring more than one workers, and because we offer them more than one contracts to hire workers on.

Workers: Approximately 2000. Each worker has a 25% probability of being given a job (for each contract they are offered). Firms: Up to 800 There is a possibility that we might not be able to reach 2000 workers. Similarly, number of firms may be below 800.

Workers
For the first 300-500 workers we will use the following sample size


1) Daily payment*uninsured* 3 day contracts: 15%
2) Steep Back-loaded* uninsured * 3 day contracts: 15%
3) Smooth Back-loaded* uninsured * 3 day contracts: 20%
4) Steep Back-loaded* insured * 3 day contracts: 20%
5) Smooth Back-loaded* insured * 3 day contracts: 20%
6) Daily payment*insured* 3 day contracts: 10%

For the remaining 1300-1500 sample we will offer each worker two contracts, one of which will be a 3 day contract and other will be a 7 day contract. The break up of 3 day contracts will be the same as above and the breakup of 7 day contract sample is below:


1) Daily payment*uninsured* 7 day contracts: 15%
2) Steep Back-loaded* uninsured * 7 day contracts: 15%
3) Smooth Back-loaded* uninsured * 7 day contracts: 20%
4) Steep Back-loaded* insured * 7 day contracts: 20%
5) Smooth Back-loaded* insured * 7 day contracts: 20%
6) Daily payment*insured* 7 day contracts: 10%


Firms:

Each firm will be offered 12-16 different contracts, one of which will be randomly chosen and implemented.