Family Decisions on Insuring Children: What the Chip Experiment Tells Us about Reauthorization
*Names in bold indicate Presenter
Study Design: We use variation in states’ CHIP expansion policies from 1999-2012 to estimate their effects on the probability that: 1) children are insured; 2) both the parent and child within a family unit are insured; and 3) parent and child are insured publicly or privately. We create parent-child ‘dyads’ by selecting a random parent for each child in the Current Population Survey (CPS) 2000-2013. Effects of child eligibility expansions ≥25 percentage points of the Federal Poverty Level (FPL) are estimated using within-state difference-in-difference models with higher income dyads as controls. Effects of premiums are estimated using cross-state models with state/year variables indicating an expansion with: 1) existing premium; 2) new premium; and 3) no premium using child/parent dyads in state/years without child expansions as controls. Child public insurance premiums are derived from state data; private premiums are estimated from the MEPS-IC as the difference between family and employee premiums for married parents (employee and employee+1 for single parents) using work status/firm size adjusted for offer rates by firm size. We use logistic and multinomial logistic regression models to estimate effects. We use controls for health status and demographic variables including state and year fixed effects and clustered (state-year) standard errors.
Population Studied: Children (n=69,927 within-state models; n=343,624 cross-state models) and child/parent dyads (n=58,062, varies slightly with specification) in families with incomes up to 400% FPL.
Findings: CHIP eligibility expansions from 1999-2012 increase the likelihood of child insurance by about 1.8 percentage points (pp). Consistent with prior research, we find crowd-out of private insurance offsetting a large portion of the increase in public insurance. Important to state CHIP experimentation, cross-state analysis indicates expansions in states with an existing or new premium increase child coverage similarly (1.1 -1.2 pp), while expansions without premiums do not increase the likelihood of coverage. Notably, 17 of 18 expansions with premiums were stand-alone CHIP programs. We find no effects of child expansions on parental coverage. A $1,000 increase in annual child public insurance premiums decreases the probability of public coverage by 6.3 pp, increases private coverage by 3.6 pp, and increases the likelihood of being uninsured by 2.7 pp.
Policy Implications: Due in part to crowd-out, further expansions of CHIP coverage levels are not likely to be effective. CHIP programs with premiums do not discourage take-up, perhaps due to lack of ‘stigma’ in stand-alone CHIP programs, however increases in public premiums significantly reduce child enrollment. The finding that stand-alone CHIP programs with new or existing premiums are more effective than traditional Medicaid expansions suggests that the flexibility of CHIP program design should be retained under the upcoming reauthorizations of CHIP. If states choose to cover child/parent using the Basic Health Plan or a waiver that allows states to set subsides for family coverage, they will need to set premiums at lower levels.