Experimental evaluation of an online counseling program expanding neighborhood information among rental voucher recipients in Chile

We developed an online platform, Aprendo y Arriendo, to deliver customized information to voucher holders. The experiment’s goal is to provide different information treatments in order to identify which information has a higher effect in the lease up rate and in producing mobility to neighborhoods with better characteristics.

Voucher use information treatment (VUIT):
*Beneficiaries are contacted via email and encouraged to use the platform.
*After entering into the platform, they will be shown information about the voucher program, including an explanation of the program and frequently asked questions.
*Before leaving the platform, voucher holders get access to a blog where they can leave their questions. Also, they are provided with a whatsapp number and email where they reach out to ask directly.
*All beneficiaries receive weekly email reminders to use their vouchers and access the platform

Neighborhood effect information treatment (NEIT):
*neficiaries will have access to an animated video that explains the relevance of neighborhood characteristics in the development and well-being of her families. The video language is simple and approachable, but it is founded on scientific research.

Neighborhood’s characteristics browser treatment (NCBT):
*Beneficiaries will have access to a neighborhood browser in the platform where they can enter an address, street intersections or specific place. They will then learn characteristics of the respective neighborhood in the following key indicators:
1. Age distribution within the neighborhood.
2. The proportion of residents in the neighborhood with a higher education degree or a higher technical education degree.
3. The Luminosity Index measures the proportion of the public sector that has public lighting provision.
4. Schooling education index: Based on the SIMCE test, this index displays the average test score in language and math for the schools closest to the neighborhood.

Each treatment arm is a combination of the treatments mentioned previously. The design will allow us to determine which barriers are more important in terms of mobility preferences and which strategies to increase lease-up rate and opportunity moves are more cost-effective. The various treatment arms are defined as follows:

Control: Does not receive any of the previously mentioned treatments. They experience the policy’s business as usual.

Treatment 1 - Voucher Information: Beneficiaries receive VUIT, which means they are encouraged to use the voucher and are given information on how to do so.

Treatment 2 - Voucher Information + Neighborhood effect: Beneficiaries are given VUIT and NEIT, which means they are encouraged to use the voucher, are given information on how to do so, and are informed why moving to a better neighborhood is important for their future.

Treatment 3 - Voucher Information + Neighborhood effect + Neighborhood Browser: Beneficiaries receive VUIT, NEIT, and NCBT, which means they are encouraged to use the voucher, are given information on how to do so, are informed about why choosing a better neighborhood is important for their future, and have information about neighborhood’s key characteristics.

2022-03-09

2023-03-09

Residential mobility (=1 if the beneficiary changed her neighborhood)
Previous and current residency geocoded location to link to neighborhood level data
Use of the voucher (=1 if the beneficiary used their voucher)
Time span between voucher assignment public announcement and lease-up.

Beliefs on the relevance of their neighborhood on well-being
Beliefs on neighborhood quality.
Preferences for immobility.
Self reported searching efforts (Number of houses taken into account to lease-up, frequency of contact with landlords, time spent searching, among others).
Interaction with mails (opened, clicked)
Access to platform
Information seen
Time spent on each information section
Addresses searched in the browser
Other information about navigation (clicks in key buttons, answers to quick surveys, among others).

Our experimental sample consists of 11,417 recipients of rental housing vouchers in the 2021 Regular round of the program, divided into 4 cohorts. The first cohort received the voucher in July 2021, the second cohort received it in October 2021, the third in December 2021 and the fourth in January 2022.

We will conduct an stratified randomization considering the following categorical variables:
• Region of residency: The geographical stratification seeks to ensure balance in the sample in types of rental markets that have significant heterogeneity across the national territory.
• Beneficiary cohort: The randomization will take into account the cohort to which the beneficiary was assigned.
• Tenancy: Tenant is a dummy variable that indicates whether or not the family is renting the unit where they lived at the time of the application. This dummy variable stratifies based on the different costs and benefits of using the subsidy between a beneficiary who doubles up with other families and another who rents a household at the time of application.
• Family with children in school age (5-18 years): The stratification by this variable is based on the various mobility constraints that these families face.
• Family who responded the baseline survey: Is a dummy variable that indicates if the family responded the online baseline survey sent before they know they are voucher recipients.

Randomization done in office by a computer

Randomization at the individual level (beneficiaries of the housing choice voucher)

No

0

11,417 voucher recipients

Treatment 1: 2.855
Treatment 2: 2.839
Treatment 3: 2.865
Control: 2.858

between 0.085 and 0.15 standard deviations (see pre analysis plan)