Experimental Design Details
Following are the details of the surveys.
1. Interactions Survey to compare donations by beggar type: The interaction survey is designed to capture charitable interactions between beggars and passers-by to compare the amount of donations toward beggars with and without items. Surveyors work in pairs and immediately after observing an interaction, one of them goes to the beggar and the other goes to the donor asking them about the amount of money (and item, if any) that was exchanged. This survey is intentionally kept short and takes less than 15 minutes. To ensure credible measurement of charitable transfers, both the beggar and the giver are asked about the transfer amount immediately post-interaction. Moreover, beggars' participation is incentivized (flat participation fee of 50 INR). The survey plans to cover 600 interactions across Delhi, with 300 each for beggars with and without items.
2. Beggars' Survey to measure beggars' types:
In this survey, we will interview the beggars with and without items across crowded streets in Delhi including religious sites, metro stations and marketplaces. The survey documents the beggars' socio-economic background, experience with the labor market, economic values, aspirations, and migration status. Moreover, in a lab-in-the-field experiment, we collect incentivized measures of their preference for working for money, free-riding, and honesty, along with basic numeracy (ability to count to 100). The enumerators will interview every beggar they observe in the randomly assigned street to them. Despite being on the street, we make a conscious effort to ensure that the beggars are interviewed privately without being overheard.
3. General Population Survey to compare beliefs about beggars with and without items:
This survey involves incentivized belief elicitation regarding beggars' preferences for working for money, free-riding, and truth-telling, as well as their proficiency in basic numeracy. These responses are then compared with the actual average responses of the beggars. Each participant is randomly assigned to report beliefs about one of two types of beggars: those with or without items, following a between-subjects design. Following the belief elicitation task, participants answer questions about their socio-economic and family background and meritocratic beliefs and preferences. Additionally, beliefs about the other type of beggar are elicited, allowing for within-subject comparisons of beliefs about the two kinds of beggars.
Towards the end of the survey, each participant is asked to distribute INR 100 between two randomly selected beggars (or one, if they choose to allocate the entire amount to only one type), which is implemented accordingly.
The survey will be conducted with adults in their homes, excluding those who have not been outside at least once in the past week to ensure similarity to passers-by and potential donors. Surveying homes ensures respondents' attention for the required half hour and provides privacy from other respondents, both of which are desirable but challenging on the street.
For the belief-elicitation exercise, photo collages of real beggars will be used, as approved by the Institutional Review Board at New York University. Each collage features photos of four beggars (one man, one woman, one girl, and one boy), each photographed twice – once with an item and once without – to create identical collages and identify the causal impact of items on beliefs about the beggars' preferences. There are two photo collages of beggars with items and two corresponding collages of the same beggars without items, enabling both between-subjects and within-subjects designs, with randomization conducted at the individual level.
4. Composition and meritocracy survey to measure types of beggars and the rate of their successful interactions by meritocracy beliefs:
In this survey, we document the total number of beggars and their composition by type (with and without items) observed within a 3-hour window across 80-90 streets in Delhi. Enumerators work in pairs with one documenting the observable demographic and vulnerability details (such as gender, approximate age, whether in a group, whether with a child) about the beggars with items and the other documenting them about the beggars without items. Further, each enumerator observes up to six beggars of their assigned type for 20 minutes each and documents the observable details of each passer-by that the beggar approaches to solicit charity and whether it is a successful interaction for the beggar (results in a positive donation). They conduct less than six observable surveys of beggars if there are less than six beggars of their assigned type on the street in that three-hour window.
After the three-hour window, enumerators will survey six randomly selected passers-by on the same streets (three men and three women) on their beliefs about whether the reason for begging is misfortune and inability to find paid work or it is a choice and unwillingness to work for money. We will use a Likert scale, taking values from 1 to 10, adapted from the fairness belief question on the World Values Survey. The short meritocracy survey is designed to study the relation between average meritocracy in different streets and whether that relates to the composition of beggars by type. Moreover, it allows us to examine the relation between meritocracy and charitable behavior on the extensive margin of the rate of successful charitable interactions by type.
Our primary hypothesis is that begging with an item has a signaling value, i.e., people's perception of the beggar's deservedness of charity improves due to the offering of an item while begging. Specifically, the perceived proportion of beggars who choose to do the sorting task is higher for the beggars with items than the beggars without items in the between-subjects comparison. We will test this hypothesis by regressing the type of beggar (with or without an item) on the reported belief about the number of beggars who chose the sorting task. We will also consider specifications involving controls for respondents' education, income, gender, age, migration status, and neighbourhood income group. We will also test for heterogenous treatment effects by respondents' fairness preferences and general self-reported attitude towards beggary. Moreover, we will also test for within-subject differences in beliefs about the beggars with and without items and test for order effects in this exercise. We will do a similar analysis for beliefs on free-riding and honesty preferences. Moreover, we will also test the hypothesis that people consider beggars with items to be more deserving of charitable help by testing if they allocate a higher share of INR 100 to a randomly selected beggar with items than a randomly selected beggar without items and whether this difference is influenced by the respondent's fairness considerations and general self-reported attitude towards beggars.
While we cannot fully identify the impact of offering items on the actual charitable donations towards beggars, we will compare the transfers towards beggars with and without items. This metric will be assessed through the Interactions Survey and represented by the INR amount transferred, where the reported amount by the beggar aligns with that of the donor. While providing financial incentives for truthful reporting is not feasible, respondents will be informed that their reported amount will be compared with the reported amount of the person they interacted with, providing a nudge for truth-telling. When analyzing charitable transfers directed towards beggars with an item, we only include interactions where the donor specifies charity as the motive for the transfer. This measure compares charitable transfers toward beggars with and without items. Additionally, we distinguish between charitable giving and charitable receipt. These coincide with charitable transfers in the case of transfers to beggars without items. However, for beggars with items, we elicit the donor's valuation of the product by offering to buy it from them. We will also conduct additional exercises to test if the difference in charitable donations to beggars with and without items might be driven by selection issues on both the givers and the beggars' sides.