Experimental Design Details
Participants in the experiment are given a real-effort task and can earn money. The task consists of transcribing short paragraphs of scanned text. The text is in Latin. They trade-off payments in return for effort against leisure. Leisure consists of leaving the lab early. The task has been chosen so that it is relatively simple but boring and tedious, i.e. participants will have to incur some disutility to perform it. At the same time, I want to minimize the role of productivity in determining the amount of money you can get. Except for the last phase, productivity (i.e. how quickly and accurately you can type) will not determine the amount of money you will earn. However, all things equal, more productive individuals will gain more ‘leisure time’, i.e. they will leave the experiment earlier. To maximize salience, the experiment will use the language of ‘work’, ‘earnings’ and ‘benefits’. The expected duration of the experiment is up to 1.5 hours.
The experiment consists of 5 stages.
Stage 1: Instructions (up to10 min).
Sateg 2: Incentivised lottery to measure risk aversion (up to 5 minutes); participants choose between pairs of lotteries with different amount of riskiness; the lottery will represent a variant of the Holt-Laurie lottery but with probabilities kept fixes and increasing spread between the potential payoffs (as in Dichroutis and Lusk, 2016). This enhances the salience of the risk aversion (curvature of the utility function) versus the subjective probability weighting. Participants will choose between two lotteries (A &B) ten times. At the end of the experiment, one lottery is chosen randomly and played out. Lottery B is more risky than lottery A but the difference in expected payoffs between A& B is increasing. The ‘switching point’ , i.e. the moment lottery B is chosen over lottery A is the measure of risk aversion.
Stage 3: Real effort phase where participants choose between paid work and leisure (up to 48 minutes); participants are asked to choose between a real effort task and ‘leisure’ (i.e leaving the experiment early). The task consists in transcribing one paragraph of scanned text. In return, they receive a fixed payment subject to the transcription being accurate enough. Accuracy is measured as the editing distance (i.e. the minimum number of insertions, deletions and substitutions necessary to obtain the correct text). If editing distance is lower or equal to three, the worker is paid in full. If the editing distance is larger than three, the worker is not paid for that paragraph. The payment structure is such that there is no great monetary advantage associated with being more productive. This is so as to not confound willingness to work with productivity. Still, productivity will be measured in Stage 4.
Stage 3a.Initially, all participants are given two paragraphs to transcribe. This is so they become familiar with the task and are not paid [2 minutes per paragraph].
Stage3b. This is followed by 12 paid periods. At the beginning of each period, each subject is choose between “paid work” and an “out of work benefit”. If the participant chooses “paid work” she has two minutes to complete the transcription of the paragraph and is paid accordingly (see treatments below). The worker can move to the next stage as soon as she completed her task in the current period. The average required time to complete the task is well below 2 minutes. If she chooses an “out-of-work benefit”, she receives £1 and moves to the next round.
The outcome of interest is the decision to work/ not work. The three treatments differ in a) the availability of work 2) the pay rate.
Control
If the subject decides to work, she needs to transcribe 1 short paragraph and receives 1.5 £. If she does not work, she receives 1£ and moves on to the next period.
Treatment 1
If the subject decides to work, the computer decides whether work is available. There is a 50% chance no work is available, and if so the subject moves on to the next period. No work is performed and no payment is due. If there is work available, the subject transcribes 1 paragraph and receives 1.5£ for doing so.
Treatment 2
If the subject decides to work, the computer decides whether work is available. There is a 50% chance no work is available, and if so the subject moves on to the next period. No work is performed and no payment is due. If there is work available, the subject transcribes 1 paragraph and receives 3£ for doing so. The idea is that the expected value of choosing work in Treatment 2 and Control are the same.
Stage 3c. This stage of the experiment will vary the benefit payments. There are 12 rounds in which participants again decide whether to work or not. In the Control treatment, subjects will face a 50% probability of a “sanction” if they choose the out-of-work benefit twice in a row. If the participant chose to not work in the previous round and again in the current round, there is a 50% probability that the benefit will not be paid.
In the two treatment groups, two variations will occur.
Variation 1 is the possibility of incurring a benefit “sanction” described above.
Variation 2 offers participants who chose to work but subsequently were offered no work to apply for the benefit in the same round.
Half of participants in each treatment group will be randomly assigned to either Variation 1 or Variation 2.
The aim is to examine whether 1) making the benefit less attractive or 2) increasing the income smoothing effects of the benefit has an effect on labour supply and to compare the two effects.
Each time they choose, participants are presented with a screen that shows their choice in the last round, how much they earned in the last round and their current total earnings (excluding the show-up fee).
Stage 4: Productivity measurement (5 minutes); typing speed/accuracy may vary substantially across subject and may be correlated with willingness to work. To improve power, this stage will measure typing productivity. Subjects will spend 5 minutes typing as many paragraphs as they can; they earn £1.5 for each correctly transcribed paragraph (with an editing distance not larger than three); participants cannot move on before the 5 minutes elapse; this is to incentivise participants to use this time to type as many paragraphs as they can.
Stage 5 (up to 1 minute) The lottery is played out and participants are shown their total earnings (show-up fee plus experimental earnings). Participants are then given instructions about how to collect their earnings and thanked for their participation in the experiment.