Nudging towards quality self-employment jobs

How to cite

Heller, Lorena. 2023. "Nudging towards quality self-employment jobs." AEA RCT Registry. AEARCTR-0010941. https://doi.org/10.1257/rct.10941-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Lorena Heller*

* Primary investigator

Registry ID
AEARCTR-0010941
DOI
https://doi.org/10.1257/rct.10941-1.0
First registered
Last updated
Status
Completed
Trial dates
2022-11-28 – 2023-07-11
Countries
Bolivia (Plurinational State of) (La Paz), Bolivia (Plurinational State of) (Cochabamba)
Registry record
http://www.socialscienceregistry.org/trials/10941

Abstract

In the last ten years, the percentage of self-employed in Latin America had doubled Europe. But self-employment in Latin America has different characteristics compared to other regions. Usually, it is a means of subsistence to get out of unemployment rather than a decision to start a business (Caceres & Caceres, 2017). Proof of this is that due to the COVID-19 crisis, a large proportion of workers lost their jobs and decided to work as self-employers ().

Moreover, given the gender roles and gender stereotypes present in Latin America, women have been more affected in comparison to men. A % of female workers in Latin America are self-employed workers in the informal labor market. This type of job led women that could not have gotten a formal job due to skills gaps and gender discriminatory barriers to enter the labor market. (Fapohunda, 2012: Biles, 2009; Chakraborty, 2020). In addition, some women choose this type of work for the labor flexibility that allows them to continue with their housework (Babbit, Brwon & Mazaheri, 2015). Although being self-employed can help women to enter the labor market, job satisfaction is low among these workers (Aguilar et al., 2013). Structurally, self-employed workers have a lower job quality; they do not have a fixed income (Hamilton, 2000; van Praag & Versloot, 2007), experience precarious working conditions (European Commission, 2008; Shore, 2011; van Praag & Versloot, 2007), and they usually do not contribute to the pension system or have health insurance. Although in Latin America in recent years, the percentage of workers enrolled in a health insurance plan had increased, still by 2021, 30.9% of workers don’t have a health insurance plan (CEPALSTAT, 2021). Similarly, most of the countries in Latin America had a smaller percentage of affiliates in the pension system (e.g., in Bolivia in 2021, only 40.1% of employees are affiliated with the pension system, CEPALSTAT, 2021).

Through responses from different questionnaires and focus groups, we found that most women wish to become formalized workers. Specifically, when they mention that they want to have a health insurance plan and be affiliated with the pension system. We might think that one of the possible reasons why women do not have these benefits is because of low income. However, after covering all their expenses, most women have additional resources; on average, they have monthly savings between 30% to 35% of their income. Therefore, if most women want to improve their job quality by affiliating to a health insurance plan and they have the resources to do it, what is the reason that holds them to obtain these benefits?

To answer this question and in search of possible solutions, we conducted different focus groups and a laboratory experiment that evaluates the implementation of nudges that modify the behavior of self-employed women. The focal groups help us understand the limitations women face in contributing to a pension system as to a health insurance plan. In the experiment, we randomly divide participants into six groups. Each group received a different treatment that followed a 2*3 factorial design. The first dimension includes an informative message about the benefits of contributing to a long-term pension system and the advantages of having health insurance. The second dimension is related to a nudge that reduces the costs of enrolling in a savings or retirement plan, as proposed by Thaler & Sunstein (2008).

Following the design of the decision space of Train & Weeks (2005) and the decision scheme of Douglas & Shepherd (2002), each participant chooses a work scheme where the decision variable is the work intensity (easy, medium, and difficult). To reflect a true employment setting, we used an effort elicitation task with different difficulty levels and amount of exercises. The higher the intensity (task difficulty) and the larger number of exercises completed the greater the reward they can obtain.

Normally, in self-employment, the income is not only determined by the effort and working hours but also by external factors that make the income unstable (for example, facing low demand and health problems). In the experiment, to simulate this situation, we presented two states of nature that randomly determined a healthy or unhealthy state. In the “Healthy” state participants received the total reward, while in the “Unhealthy” state participants lose 80% of their reward, regardless of the intensity and number of hours worked. However, each participant has the option to enroll in a health insurance plan. This insurance would be delivered to the participant if the state of nature determines that they are unhealthy. Given their chosen plan, they could keep their reward or lose a smaller amount. In addition, we gave participants the option to contribute to long-term insurance. If participants decide to contribute to this system at the end of the experiment they will receive their reward, but additionally, four months after they will receive their social security contribution plus its interest rate earnings.

We found that a message that incentivizes pensions increases its contribution by 6%. Similarly, the health insurance message increases its contribution by 42%. The nudge that decreases the effort participants need to make to contribute has a positive significant effect of 5% only on the contribution of the pension system. But for the percentage of contribution to health insurance, we don’t see a significant effect. When the informative message is combined with the active decision nudge, we are still able to see a positive effect however it decreases from the effect of only delivering the informative message for only providing the active decision treatment.

Intervention

Implement a laboratory experiment to examine the effectiveness of various nudges that encouraged voluntary enrollment in health insurance and Social Security among self-employed women. The intervention consisted of recruiting participants for different experimental sessions and randomly assigning them to one of six treatments. These treatments included informative messages highlighting the benefits of contributing to a long-term pension system and having health insurance, as well as nudges designed to reduce the effort and time in the process of enrolling in the pension system or health insurance. Participants engaged in an effort elicitation task with varying difficulty levels and exercise quantities, simulating real-world employment conditions. They had the option to choose the most challenging task intensity and complete exercises to earn a monetary reward. Participants repeated this process for nine rounds. In every round, they could experience a healthy or unhealthy state randomly. Additionally, participants could enroll in a health insurance plan to mitigate potential losses in the unhealthy state, and those contributing to long-term insurance received their rewards along with interest rates paid over the past four months.

Experimental design

We designed a laboratory experiment to evaluate the implementation of nudges that modify the behavior of self-employed women to increase their job quality.
To reflect the actual employment decisions, we used an effort elicitation task. We followed the decision space design of Train & Weeks (2005) and the decision scheme of Douglas & Shepherd (2002), where each participant chose a work scheme. First, women could choose the task difficulty (i.e., easy, medium, and hard). The monetary reward they could obtain depended on the number of correct exercises and the level of the task. In addition, we ask participants if they wish to contribute to the pension system and a health insurance plan. In every round, the participant decided on the effort task difficulty. At the end of the exercise, the facilitators notify them of the amount of their reward. With this information, participants decided if they wanted to contribute to health insurance and the pension system.
Effort tasks are commonly used to assess individual investment and performance in an experimental setting. We used the “Pairs to 100” task, which consists of finding in a set of numbers a pair that sum up to one hundred. The size of the number sets on which they must find the pairs defined the difficulty. Regardless of the level, we asked participants to find four pairs. In each round, they had three minutes to solve a maximum of 6 exercises.
In our experiment, we aimed to simulate the instability and unpredictable nature of self-employment income by introducing two randomly determined states of nature: (1) healthy and (2) unhealthy. In the healthy state, participants received the full reward, while in the unhealthy state, they faced the possibility of losing 80% of the reward. However, each participant had the opportunity to mitigate this risk by enrolling in a health insurance plan. If participants decided to contribute, they would have to stand up from their seats and deposit a form in an amphora outside the room. Depending on the plan they chose, participants would either incur a smaller loss or maintain the full reward.
Moreover, participants had the option to contribute to social security benefits. Similarly, if participants decided to contribute to this system, they would have to stand up from their seats and deposit a form in an amphora outside the room. At the end of the game, they received their reward, and four months after the experiment implementation, they collected their social security contribution plus their interest rate earnings

Design details

Randomization method
Each treatment consist of a nudge that encourage workers to make decisions focused on improving their quality of employment. We will assign treatments on a session-by-session basis. The treatment assignment will be done for all the sessions prior its implementation.
Randomization unit
Experimental sessions
Clusters
2 cities
Observations
420 participants, 210 in every city.
Arms
70 participants by treatment
Primary outcomes
We have two main variables of interest, the proportion of contribution to a pension system and health insurance from their total reward.

Cited by

No citation data available for this registration yet.

Keywords