Survey-based values for the trade-off between poverty and longevity

Last registered on June 24, 2024

Pre-Trial

Trial Information

General Information

Title
Survey-based values for the trade-off between poverty and longevity
RCT ID
AEARCTR-0013756
Initial registration date
June 06, 2024

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
June 24, 2024, 12:19 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Affiliation
University of Oxford

Other Primary Investigator(s)

PI Affiliation
The World Bank
PI Affiliation
The World Bank

Additional Trial Information

Status
In development
Start date
2024-06-17
End date
2025-08-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Mortality is one of the main sources of well-being losses across the world. Cross-country well-being comparisons are drastically affected when mortality is included, as highlighted by Becker et al. (2005) or Jones and Klenow (2016). In a similar vein, Murphy and Topel (2006) estimate that the life-expectancy gains realized in the US over the period 1970-2000 were worth about half of US GDP at the end of that period. In spite of its importance, mortality is often overlooked when monitoring well-being. This omission is in fact the norm for poverty measures. Poverty measures that ignore mortality make “biased” well-being comparisons for two reasons. First, they simply do not account for the high intrinsic value of longevity, i.e., being alive. Second, they are in fact perversely reduced by the death of poor individuals, which constitutes a “mortality paradox” (Kanbur and Mukherjee, 2007). As a result, their “biased” comparisons prevent from properly monitoring low well-being. This issue leads to sub-optimal policy-making because poverty measures are regularly used for guiding and evaluating policies. Clearly, budget allocations that affect mortality should account for their mortality consequences. The primary objective of this research project is to propose/calibrate welfare measures that integrate longevity in a way that is consistent with how most people would make trade-offs between their own incomes and longevity. For this purpose, we will carry out an online survey with participants in a sample of countries from different income categories and continents (e.g. USA, South Africa, Ethiopia, Indonesia, France, Brazil, Egypt, Philippines, Nigeria, Mexico and India). The survey probes the attitudes of respondents, i.e., their stated preferences. The surveys will allow us to estimate the functional form and parameters of the Social Welfare Function that respondents implicitly use when making trade-offs involving income or lifespan.
External Link(s)

Registration Citation

Citation
Decerf, Benoit, Christopher Hoy and Olivier Sterck. 2024. "Survey-based values for the trade-off between poverty and longevity." AEA RCT Registry. June 24. https://doi.org/10.1257/rct.13756-1.0
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details

Interventions

Intervention(s)
Intervention Start Date
2024-06-17
Intervention End Date
2025-08-31

Primary Outcomes

Primary Outcomes (end points)
The main outcome variable is a binary variable identifying whether respondents are willing to pay for a costly medical treatment that would increase their lifespan.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Our main contribution is to elicit people’s preference on alternative lives, which differ with respect to longevity and incomes. We will use a survey experiment to know how study participants trade-off income and lifespan. In particular, we are interested in the way people trade-off these aspects when confronted with the perspective of large income losses, which would drastically affect their standards of living and push them into poverty.

The survey will target online participants in yougov.co.uk surveys targeting about 20,000 participants from 10 countries, including high-income, middle-income, and low-income countries. In each country, we will target 2000 participants.
Experimental Design Details
Not available
Randomization Method
The randomization is done on Alchemer, the survey platform used for this research.
Randomization Unit
Randomization is done at the individual level.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
No clustering in this survey experiment.
Sample size: planned number of observations
The study will reach 20,000 respondents from 10 countries, including high-income, middle-income, and low-income countries. In each country, we will target 2000 participants.
Sample size (or number of clusters) by treatment arms
One-third of respondents in each country will be allocated to the treatment group.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The sample size of 2000 respondents in each country (670 in the treatment and 1330 in the control group) was determined by examining the sample size of other cross-country randomized survey experiments in this field as well as conducting statistical power calculations. In the case of the seminal cross-country randomized survey experiments by Alesina et al. (2018) and Alesina et al. (2022) they also included around 600 to 1000 respondents in each treatment group in each country. Furthermore, a Journal of Economic Literature article summarizing best practices in online randomized survey experiments providing information interventions suggests that having at least 600 respondents per treatment/control group should provide adequate power to detect an effect that is commonly observed in the literature (Haaland et al., forthcoming). In terms of the specific statistical power calculations, we considered two broad scenarios. The first is to consider the minimum effect we would be able to detect (with the standard probability of type one and type two errors of 0.05 and 0.2) from a single treatment on a single binary outcome (using the example of collapsing a Likert scale into a dummy variable that takes the value of 1 for strongly agree or agree and 0 otherwise). We could detect an effect in the order of six percentage points based on the conservative assumption of 50 percent of the control group agreeing with the statement in the question. The second is to consider the minimum effect we would be able to detect on a subset of respondents from a single treatment on a single binary outcome. In this case, the underlying sample size in each of the treatment and control groups will shrink based on the priors of respondents (e.g., whether they believe their household is currently poor). We consider three examples, the first covering 75 percent of all respondents, the second covering 50 percent of all respondents and the third covering 25 percent of all respondents. We present the effect size we could detect with these smaller sample sizes below, once again based on the conservative assumption of 50 percent of the control group agreeing with the statement in the question. Table 1 – How the minimum detectable effect in a single country varies based on sample size Number of respondents per treatment/control group with a particular prior belief/preference 1000 (100%) 750 (75%) 500 (50%) 250 (25%) Minimum detectable effect (Percentage points) 6 7 9 13 This table shows how the minimum detectable effect in a single country varies based on the sample size. There are multiple reasons to believe that we will potentially have much higher statistical power to detect effects than what is presented in Table 1. Firstly, in the instance whereby the mean in the control group for a binary outcome is higher or lower than 0.5, the minimum detectable effect will be smaller. Secondly, by pooling data across countries (with country fixed effects) we will have substantially more power to detect an impact from the treatment, however it is hard to quantify this ex-ante as it will depend on variation between countries.
IRB

Institutional Review Boards (IRBs)

IRB Name
Ethics Committee for the Social Sciences and Humanities
IRB Approval Date
2024-03-26
IRB Approval Number
SHW_2024_34