Taxing the Poor Twice: Poverty, Bandwidth, and Utility from Consumption

i. Control
The control group will listen to short, emotionally neutral stories for approximately 10-15 minutes. This control group allows for a comparison with a group with similar levels of interactions with study staff and similar total time, but no activities intended to change bandwidth.

ii. Memorizing a sequence of visual stimuli
Participants in this condition will be asked to memorize an ordered sequence of blocks in different locations on a page.

iii. Scarcity Priming
In this treatment, participants will be read a story about a typical person in the city who is from a disadvantaged background. The story narrates both positive and negative financial issues that a person from a background similar to that of the participants would face in their daily lives (e.g. a large number of high interest loans). The story would be followed by a conversation with participants about similar concerns that they may face. The participants will clearly be told that they are free to not answer questions that they would prefer not to.

iv. Thirst
Participants in this condition will be given dry salty crackers. They will be asked to consume the crackers, but may, of course, choose not to. As the intention of this arm is to generate thirst, participants randomized to this condition will not be provided with water until the end of the experimental activities unless they specifically request it.

2019-08-22

2020-01-31

Score on the likert scale rating the value of the experiences

The primary outcome of interest is the valuation of the consumption experiences as measured via the likert scale measure. This measure is the primary outcome as it is simple and easy to respond to when bandwidth constrained which we believe will make it a more accurate reflection of the value of the consumption.

1) Bandwidth changes as measured by changes in performance on cognitive tasks
2) Willingness to pay (BDM) measures for additional consumption
3) Difference in willingness to pay (BDM) for anticipated while bandwidth constrained vs. consumption while not bandwidth constrained where the WTP is elicited before any changes in bandwidth are induced (i.e. anticipated enjoyment with or without bandwidth constraints)

1) Bandwidth measures: These measures are intended to be confirmatory of the efficacy of the first stage (i.e. we anticipate the treatments will reduce bandwidth).

2) WTP: These outcomes are considered secondary due to the concern that the BDM elicitation itself is cognitively taxing (complex) and that changes in bandwidth may result in a wedge between true value of consumption and stated willingness to pay. We will investigate this hypothesis by looking for suggestive evidence of heuristic decision-making in willingness to pay (e.g. lumpy outcome measures).

3) Difference in WTP: In advance of imposing any bandwidth constraints or undertaking any consumption, we will elicit the participants' WTP for consumption while either bandwidth constrained or while not bandwidth constrained (see additional details below). The purpose of this task is two-fold. First, it is used to determine whether people anticipate that bandwidth constraints reduce the value of consumption. Second, it serves as a test of the hypothesis that willingness to pay may be confounded by bandwidth constraints.

Sample and Recruitment: Participants will be recruited in person on the street in communities in Chennai. After a potential participant is approached, they will be a brief description of the study and be asked if they are willing to take a short survey to determine if they are eligible. If eligible and interested in participating, the participant will be referred to the study office to complete a vision screening and complete the informed consent process. If the individual wishes to participate and completes the informed consent they will be enrolled in the study and will complete the activities described below.

Participation has three major components:

First Component
In the first portion of the experiment, participants will be randomly assigned to one of the following treatment arms. These interventions include:

a. Control: The Control condition is the “no change” comparison group – in which participants will listen to a series of short stories with little emotional content. Participants do this activity rather than nothing to keep constant interactions with surveyors and the duration of the study.

b. Memorizing a sequence of visual stimuli: This treatment group would be asked to remember a sequence of blocks arranged in a page in a given order (e.g. top right, bottom left, center right, center top, bottom right), a standard lab intervention to tax bandwidth. This treatment will also be administered at the beginning and between each outcome measurement to maintain a high cognitive burden. Given the lower literacy and numeracy in this study population, we will use a visual sequence rather than a string of numbers.

c. Scarcity priming: Priming on financial concerns similar to those undertaken in Mani et al. and other studies examining the impact of bandwidth (Shah et al. 2012; Bartos et al. 2018). In this treatment, participants will be read a story about a typical person in the city who is from a disadvantaged background. The story narrates daily financial concerns that a person from a background similar to that of the participants would face in their daily lives. The story will be followed by a conversation with participants about similar issues they may be facing. The participants will clearly be told that they are free to not answer questions that they prefer not to. At the end of the treatment condition, participants in this condition are asked a question on how exactly they would manage a huge expense in the near future.

d. Thirst: Generation of thirst via consumption of dry salty crackers. Participants in this condition will be asked to eat a few dry salty crackers (e.g. saltines) to increase thirst. We will ensure that the participant does not have any form of allergy towards the ingredients present in the food item. Participants are free to decline to eat the crackers.

Prior to implementing the interventions, a baseline survey is conducted to collect demographic data. Participants are also asked to express their willingness to: 1) play a board game while simultaneously listening to an unpleasant noise and then sit quietly for a few minutes, or 2) play the same board game and then listen to the unpleasant noise. The total duration of these activities and the time spent listening to the noise is the same across these two versions (simultaneous vs. sequential) of the activity. The purpose of this task is two-fold. First, it is used to determine whether people anticipate that bandwidth constraints (the unpleasant noise) reduce the value of consumption. Second, it serves as a test of the hypothesis that willingness to pay may be confounded by bandwidth constraints.

Second Component:
In the second portion of the experiment, participants engage in two short tasks, to measure cognitive bandwidth.

The cognitive tests used to measure bandwidth for the participants are:
a. Psychomotor vigilance task (PVT) (measures the ability to sustain attention)
b. Raven's progressive matrices (measuring fluid intelligence)
The order of the cognitive tasks is randomized at the individual level.

Third Component:
In the third portion of the experiment, participants engage in a set of positive experiences and answer questions about their enjoyment (or lack thereof) of the activities. The order of these activities is randomized across individuals.
a. Watching short videos (comedy video)
b. Listening to music (short local songs)
c. Completing puzzles or easy games (e.g. jigsaw puzzles)
d. Eating a small food item (e.g. a bite or two of chocolate)

The order of these experiences is randomized at the individual level.

Measurements of the value of consumption include: (1) self-reported enjoyment on a Likert scale (primary) and (2) willingness to pay to engage in a similar activity again (secondary). If the participant does not wish to do the activity, they simply report a willingness to pay of 0 Rs.

The study participants would stay in the study office for approximately 3 to 4 hours to complete these tasks.

Randomizations were conducted in Stata. The original randomization files are preserved.

A. Treatment is randomized at the individual level to one of four arms: Control, Scarcity Priming, Memorizing a sequence of visual stimuli and Thirst. The sample is divided into 2 strata. One with above median financial stress and another with below median financial stress based upon a baseline measure of financial stress.

B. Within individuals, the order of the tasks listed below is randomized (to allow us to control for order effects):
a. Order of the cognitive tasks (Raven’s Matrices and PVT)
b. Order of the consumption activities (the likert measurement is done directly following the consumption activity)
c. Order of Willingness to pay for consumption choices
d. Order of willingness to pay elicitation for the simultaneous (game and noise at the same time) vs. sequential (game and noise in sequence) decisions.

C. Each participants is also randomly assigned to either complete (5% of the sample) or skip (95% of the sample) their choices in the simultaneous vs. sequential consumption activity. The individuals who complete their choices will be excluded from the primary analysis in a robustness check.

No

520 individuals.

520 individuals.

130 participants per experimental arm (control, memorization, scarcity prime, thirst)

0.35 unit increase in the likert scale measure when pooled across treatments