The Impact of Emergency Cash Assistance in a Pandemic: Experimental Evidence from Colombia

We evaluate the impact of cash relief for vulnerable households during a pandemic. On March 31, 2020, Colombia rolled out an unconditional cash transfer program to help low-income households deal with the economic consequences of the coronavirus pandemic. The transfer is worth 75,000 pesos (8% of the monthly minimum wage) and paid in April and May to one million households. A subsample of 3,400 households was randomly assigned to receive this cash transfer (treatment) or not (control). We survey these households via phone in June 2020, after households received the second COVID-19 cash transfer.

2020-06-03

2020-06-23

The primary outcomes in this study are consumption and food security, children's education, behaviors to mitigate the spread of COVID-19, physical and mental health, financial risk, intimate partner violence, trust and attitudes towards the government’s response to the pandemic. We measure these outcomes 0 to 4 weeks following the second cash transfer. We also use composite measures of these outcomes following the approach outlined in Anderson (2008).

- Consumption: Total weekly spending on groceries and temptation goods (e.g., alcohol, cigarettes). Total monthly spending on rent, utilities, clothing, medication, transportation, debt, phone or internet access, and education.
- Food security: In May 2020, the (negatively coded) number of days the adults / children in the household reduced food consumption / went a full day without food; in May 2020, (negatively coded) indicators for whether adults / children in the household had to rely on others for food, or had to beg for food.
- Education: The number of hours children spent on education last week; indicators for the main challenges associated with children's education while schools are closed; indicators for the main actions taken to support children's education while schools are closed.
- Behaviors to mitigate the spread of coronavirus: in the last week, how often did he/she i) leave their home at any moment; ii) leave their home for work; iii) go to a market or food store; and iv) receive a visit at home. Also in the last week, how many people did he/she interact with excluding household members. Lastly, in the last week, whether he/she washed hands with soap and water more or less often than before the national quarantine, and whether he/she wore a face mask when going out in public. Total weekly spending on cleaning items to reduce the spread of COVID-19 (e.g., alcohol, soap, face masks, antibacterial gel hand sanitizer).
- Physical health: Indicators for whether any household member developed symptoms related to COVID-19 (e.g., cough, fever, difficulty breathing) since the national quarantine began.
- Mental health: Indicators for whether any household member felt anxiety or sadness, behaved more aggressively or had difficulty sleeping since the national quarantine began. Also, what their main source of concern is regarding the coronavirus pandemic (e.g., physical health, mental health, economic needs, other household members, not being able to educate children).
- Financial risk: Indicators for whether the household sold assets, used their savings, borrowed money, was unable to pay back a loan, or lent money to relatives since the national quarantine.
- Intimate partner violence: Indicators for whether an individual experienced physical, emotional, economic, or sexual violence (this information is available only for those living with their significant other).
- Trust and attitudes towards the government: whether he/she trusts the Colombian government to promote the well-being of its citizens; whether he/she supports the national quarantine; whether he/she supports emergency aid programs rolled out by the government in response to the coronavirus pandemic; whether he/she supports evading the VAT; and whether he/she would voluntarily donate money and/or work to support his/her community.

The salience of the COVID-19 transfers, labor supply, migration.

- The salience of the COVID-19 transfers: Whether recipients know they are beneficiaries of the COVID-19 transfer. For those who confirm having received a COVID-19 transfer, i) how much they believe they received; ii) whether they experienced difficulties receiving the transfer (if so, what type of difficulties); and iii) what he/she spent the COVID-19 transfer on.
- Labor supply: whether the head of household worked for pay in the last week.
- Migration: whether he/she lives in a different home since the national quarantine.

The coronavirus pandemic forced the Colombian government to expedite the COVID-19 cash transfer to assist people living in extreme poverty. A subset of 3,400 households eligible to receive the COVID-19 cash transfer was selected to be part of an RCT. The sample selection for the RCT proceeded in five steps:

Step 1: Start with the universe of 2.7 million households (roughly 17% of all households in Colombia) who receive "Familias en Acción." Define eligibility for the COVID-19 transfer as receiving "Familias en Acción" and living in extreme poverty, defined as having (a) a SISBEN IV "Level A" score or (b) a SISBEN III score below 10. This results in some 733,000 households eligible to receive a COVID-19 transfer in Colombia.

Step 2: Order Colombia's 1,105 municipalities by quartiles of MPI. Remove the poorest quartile from the sample.

Step 3: For the remaining municipalities, generate quartiles based on the number of households eligible to receive a COVID-19 transfer in each municipality.

Step 4: Randomly select approximately municipalities in each quartile to obtain 53 (randomly-selected) municipalities. These 53 municipalities have a total population of 17 million individuals or 6 million households, of which around 725,000 receive "Familias en Acción" and around 156,000 are eligible to receive a COVID-19 transfer.

Step 5: From these 156,000 households, randomly assign 1,500 (with a 20% oversample) to treatment and 1,500 (with a 20% oversample) to control. This results in roughly 3,400 households, of which 1,700 households are assigned to treatment and 1,700 to control.

Our main estimates of the impacts of the COVID-19 cash transfer will be obtained by the following OLS regression specification:
y_i = a+b T_i + X_i'G + e_i
where y is the outcome for household i, T is the treatment dummy that equals 1 if the household was (randomly) selected to receive a COVID-19 transfer, X is a vector of baseline controls (e.g., household size, municipality fixed effects), and e is an idiosyncratic error term. We use heteroskedasticity-robust standard errors.

The randomization was done in office by a computer

Household

No

Around 3,400 households.

We aim to contact all 3,400 households by phone. We expect to get a response rate of roughly 40%, for a final sample of around 1,360 observations.

We will aim to survey the 1,700 treatment households and the 1,700 control households.