Performance Evaluation of the Seoul Safety Income Project (SSIP)

Households in the treatment group will receive a monthly payment corresponding to half of the difference between their income and 85% of the standard median income. Households will be subject to an income and means test on a monthly basis during the three-year period to verify their eligibility and ensure that the amounts of SSIP income they receive reflect the changes in their income and assets. Each eligible household will receive a debit card, from which they can withdraw their monthly SSIP payment.

2022-07-11

2025-06-20

Primary outcomes can be categorized into three domains:
1. Economic outcomes: Labor market (employment status, work hours); Household finance (earned income, income from public transfer, income from private transfer, total expenditure, assets, debts); Housing (housing condition, housing satisfaction);
2. Health outcomes: Health (general health status, healthy eating, health behaviors, sleeping); Mental health (self-esteem, depression, stress, happiness, life satisfaction);
3. Family and Society outcomes: Family/education (family satisfaction, family conflict, attitude to marriage, fertility preference, amount of money spent on education/training and children’s education); Social behavior (social relationship, self-community interest); Receipt of other social benefits;

Similar to primary outcomes, we categorize secondary outcomes into three domains:
1. Economics outcomes: Labor market (job search effort, determination to work, job satisfaction, willingness to work extra hours, side jobs, leisure activities); Household finance (expenditures and assets by category, financial security, charity donation); Housing (residential unit type, ownership status);
2. Health outcomes: Health (general health status by category);
3. Family and Soceity outcomes: Family/education (reeducation training program, childcare); Social behavior (donations, degree of social isolation, trust in others, types of hobbies, volunteering activities)

SSIP recruits study households in two phases. Phase 1 started in July 2022 and Phase 2 will start in July 2023. In the first phase, 1500 households with less than 50 percent of the standard median income and assets less than USD 272,803 (≈KRW 326 million) were selected for the study sample. Among eligible applicants, 1,500 households were randomly selected, and became the study sample for Phase 1 experiment. Out of 1500 households, 500 and 1000 households were randomly assigned to the treatment group and the control group, respectively. In the second phase, 3300 additional households with less than 85 percent of the standard median income and assets less than USD 272,803 will be selected. Among eligible candidates, 3,300 households will be randomly selected, and 1,100 and 2,200 households will be randomly assigned to treatment and control groups, respectively.

We will collect information through surveys and administrative data. Study participants (of Phase 1) have already agreed to take a total of eight rounds of survey: a baseline survey before the first income payment, six rounds of follow-up surveys conducted once every six months for three years, and an endline survey one year after the income payment ends. Our research team developed survey instruments in collaboration with SMG. Household heads will be the main respondent of the surveys. We also plan to exploit administrative data on participants’ monthly labor/business income and assets level provided once a year by the Korea Social Security Information Service (SSIS). All project participants have allowed us to access their information via SSIS. Furthermore, we are able to access debit card usage records for households in the treatment group from which we may be able to infer the usage of SSIP income benefits.

Not available

We employ a stratified random sampling procedure that follows statistical criteria designed to ensure representation of the household sizes and ages that are potentially eligible for the SSIP. Stratification was based on two characteristics; household head’s age (divided into three categories – -39, 40-64, and 65+) and household size (divided into four categories – 1, 2, 3, and 4+). Randomization was conducted by SMG using a computer.

Household

No

4,800 households (1st phase - 1500; 2nd phase - 3300)

4,800 households (1st phase - 1500; 2nd phase - 3300)

3,200 households in control group and 1,600 households in treatment group.

If we set the treatment effect size at 0.1 standard deviations, holding the significance level at 0.05 and treatment to control group sampling ratio at 0.5, our power calculations suggest that the sample size for achieving statistical power of 0.8 is 2,550 and for statistical power of 0.9 it is 3,414. As our estimated sample size from pooling both phases is 4,800, the power calculations suggest that the project is capable of detecting economically meaningful treatment effects on households’ outcomes.