Aspirations, Beliefs, and Behavior: Evidence from a Randomized Religious Intervention in Western Kenya

Pentecostal Christianity is linked strongly with increased aspirations, hope, and trust, all factors which economists have identified to play a major role in economic development and poverty alleviation. (e.g., Nunn and Wantchekon, 2011, Lybbert and Wydick, 2018; Genicot and Ray, 2020; Bloem, 2021; McKenzie et al., 2022). Much of the recent economic literature has focused on the drivers of the strong global growth of Pentecostal Christianity. In Brazil, for instance, Costa et al. (2023) find that recessions lead to increases in Pentecostal affiliation, likely due to increased social insurance provided by the churches. However, an open question is how the practices and beliefs of Pentecostalism change economic behavior. Understanding this question requires entangling the direction of causality: are those who are more aspirational more likely to join Pentecostal churches, or do the practices and beliefs taught in Pentecostalism make one more aspirational?

The global rise in Neo-Pentecostal and Charismatic Christianity (hereafter NPC) in recent decades has been dramatic, particularly in the Global South. In Sub-Saharan Africa (SSA) in particular, it is estimated that there were approximately 229 million adherents to this form of Christianity as of 2020 and it is the fastest growing religious group in the region (Johnson and Grim, 2023). The continued spread of NPC may have important implications for economic growth in one of the least economically developed regions of the world. Freeman (2015) describes that particular traits of NPC can have significant economic impacts, such as a focus on material wealth and well-being as a reward for faith (i.e., the “Prosperity Gospel”), which empowers individuals to take risks and stimulates entrepreneurial behavior. Indeed, Van Dijk (2012) describes sermons in NPC churches as explicitly presenting the Christian Gospel as a roadmap to economic success.

Randomized experiments can aid in identifying the relationship between religious affiliation/beliefs and economic outcomes. This proposed research project will provide the first randomization-based evidence on economic outcomes arising from NPC messages such as the prosperity gospel. To this point, however, there have been few randomized interventions related to religion in developing countries. One notable example, however, is Bryan et al. (2021), who in the Philippines show that poor households randomly selected to participate in an Evangelical Protestant education program have higher religiosity and income, the latter of which the authors attribute to an increase in ``grit.’’ We plan to diverge from this previous work by learning the economic effects of an intensive but short-term program related to NPC messaging in an African setting. Butinda et al. (2023) highlight how economists have largely underestimated the impact of beliefs in influencing economic development in Africa. This setting and design allows us to uncover increasingly relevant and generalizable evidence into the developmental implications of the fastest growing religious doctrine across the continent.

In our proposed project, we plan to organize intensive religious workshops to occur over two-days within informal settlements of Kisumu, Kenya. These workshops will serve as the intervention exposure, providing different forms of religious messaging to attendees randomly assigned to attend that workshops. We expect these workshops to be two days long to generate significant enough exposure to generate potential differences in behavior, but not so long as to be too inconvenient to the attendees.

2024-07-19

2024-07-22

Survey responses to 1) religious attitudes and beliefs, 2) feelings of power and control, 3) grit 4) self-control 5) alcohol use in past week; results of choice experiment described below (i.e. amount given to charity, to pastor, to spouse; amount "saved"; amount given to entreprenuership fund.)

To construct the religious indices, we will follow Bryan, Choi, and Karlan (2021) and first standardize the responses to each question such that the control group has unit variance and zero mean: from each index component (i.e., survey question) we will subtract the control mean and divide by the standard deviation of those in the control group. We will then sum each of these standardized components, then standardize this summation again in reference to the control group. This leads to the creation of four indices: Intrinsic (depth and personal connection to their faith), Extrinsic (use of religion to advance their social goals), General (questions that do not fall into either the intrinsic or extrinsic category), and Fundamentalism (questions adapted from Altemeyer and Hunsberger (2004)).

Of the 1400 randomly selected individuals in this project, 600 will be a control group and 800 will be in the treatment, which will be split evenly between two different treatment arms: participants in one treatment group will participate in a two-day long workshop with an NPC pastor who will focus on tenants of the prosperity gospel, while the second treatment group will attend a workshop with an establishment church speaker (e.g., Roman Catholic) for two days. The participant will not know beforehand which group they have been assigned. The invitation to the workshops will also detail to the potential participant that a small monetary incentive will be provided for participation, which, as described below, will be used to help determine the economic outcomes.

Following the intervention, enumerators will conduct surveys and experimental games with the treatment and control individuals, which will take place within one week of the intervention to mitigate outside influences affecting the results. When the enumerator visits the household, they will introduce the experimental game by providing part of the incentive compensation to the participant (1 USD) which will be for a practice game, and the remainder of the compensation (7 USD) for the main experimental game.

For the primary experimental game, the enumerator will present eight scenarios related to economic behavior and decision-making in a random order, stressing that only one would be randomly chosen as binding.

Following this choice experiment, the enumerator will conduct a survey with the participant, asking demographic and household questions, questions related to time use and employment, questions on expenditures, and questions on alcohol consumption (alcoholism is a significant public health issue in this area). There will also be a detailed module in the survey related to religious practices and beliefs.
Given the results of the choice experiment and pre- and post-workshop surveys, we will test the hypothesis that attendance at a workshop featuring an NPC preacher focused on the prosperity gospel has meaningful effects on participant behavior and beliefs.

Given the area of the settlements, a random point generator will select 1400 households to participate in the program. From each household, one adult will be randomly selected by enumerators who visit the household to provide the invitation, collect contact information, provide informed consent, and give a short baseline survey. Of the resulting 1400 individuals, 600 will be a control group and 800 will be in the treatment, which will be split evenly between two different treatment arms: one treatment group will participate in a two-day long workshop with NPC pastors who will focus on tenants of the prosperity gospel, while the second treatment group will attend a workshop with an establishment church speaker (e.g., Roman Catholic) for two days.The participant will not know beforehand which group they have been assigned. The invitation to the workshops will also detail to the potential participant that a small monetary incentive will be provided for participation, which, as described below, will be used to help determine the economic outcomes. The workshops will be in different church locations to minimize non-speaker effects and prevent communication between the groups during the workshops, and project staff will track attendance during and over the two days. NPC workshops will be held in randomly selected NPC churches that will have agreed to host workshops (four per locality: Nyalenda A and B, and Manyatta A and B of Kisumu), while establishment church workshops will be held in Anglican or Roman Catholic churches (one per locality). Staff will provide lunch each day to all participants at the workshops.

Following the intervention, enumerators will conduct surveys and experimental games with the treatment and control individuals, which will take place within one week of the intervention to mitigate outside influences affecting the results. When the enumerator visits the household, they will introduce the experimental game by providing 1 USD to the participant which will be for a practice game, and the remainder of 7 USD for the main experimental game. For the practice game, the enumerator will read the following prompt:

“Here is X Kenyan Shillings (equivalent of 1 USD). This is your money and yours to keep and do with it as you would like. It is part of the compensation for your participation in the survey. I will now read several scenarios to you. You will be able to use this money you have received in the scenarios, although you do not need to. Think carefully about each scenario because one, and only one, scenario will be selected as binding after the activity is complete and you will be obligated to provide any money you indicated in that scenario if selected as binding. However, you do not need to indicate that you will provide any money for any of the scenarios, if you do not wish. The maximum amount that you can choose to give for each scenario is X KSh.”

The enumerator will then present the following four scenarios in a random order, stressing that only one would be randomly chosen as binding:
1. How much money would you like to give as a gift to your ward representative?

2. How much money would you like to donate to a national charity called the Kenya Children’s Fund?

3. How much money would you like to throw in the garbage?

4. How much money would you like to send to your mother (if mother no longer alive, indicate female relative)

After the participant has decided amounts for each scenario, a random number generator will select one option as binding.
Following this practice round, the enumerator will proceed to play the main experimental game with the respondent and will read them the following prompt:

“Here is X Kenyan Shillings (equivalent of 7 USD). This is your money and yours to keep and do with as you would like. It is compensation for your participation in the survey. I will now read several scenarios to you. You will be able to use this money you have received in the scenarios, although you do not need to. Think carefully about each scenario because one, and only one, scenario will be selected as binding after the activity is complete and you will be obligated to provide any money you indicated in that scenario if selected as binding. However, you do not need to indicate that you will provide any money for any of the scenarios, if you do not wish. The maximum amount that you can choose to give for each scenario is X KSh.”

The enumerator will then present the following five scenarios in a random order, stressing that only one would be randomly chosen as binding:
1. How much money would you like to donate to a national charity called "Anglican Development Services Kenya"?
2. How much money would you like to donate to a national charity called "Catholic Relief Services"?
3. How much money would you like to donate to a national charity called "Christian Aid in Kenya"?
4. How much money would you like to donate to a national charity called "Kenya Muslims Charitable Society"?
5. How much money would you like us to give to your spouse, adding X%? If no spouse, indicate your nearest adult relative.
6. How much money would you like to donate to the preacher at: (treatment group: your workshop) (control group: your church)?
7. How much money would you like to give back to us, and after one month, we will return to you (via mpesa) with R% interest?
8. How much money would you be willing to contribute to an entrepreneurship fund to be shared between you and other participants in the program to purchase supplies to start a communal business?

Following this choice experiment, the enumerator will conduct a survey with the participant, asking demographic and household questions, questions related to time use and employment, questions on expenditures, and questions on alcohol consumption (alcoholism is a significant public health issue in this area (WHO, 2018; Murphy, 2023)). There will also be a detailed module in the survey related to religious practices and beliefs.

Given the results of the choice experiment and pre- and post-workshop surveys, using linear regressions, we will test the hypothesis that attendance at a workshop featuring an NPC preacher focused on the prosperity gospel has meaningful effects on participant behavior and beliefs.

Households selected by a computer placing random points onto a map of localities. Churches selected by creating a random ranking of all Pentecostal churches and a random ranking of all Anglican/Catholic churches in each locality. We will choose the top four Penetecostal churches from that list from each locality, and the top Anglican/Catholic church from each locality to invite to host workshops.

Households. Stratified random sampling by locality.

No

1,400 households

1,400 households

600 households in control, 400 in treatment arm one (Pentecostal), 400 in treatment 2 (Mainstream).