Saving Teens: Using a Policy Discontinuity to Estimate the Effects - - PowerPoint PPT Presentation

Saving Teens: Using a Policy Discontinuity to Estimate the Effects of Medicaid Eligibility

Laura Wherry1 Bruce Meyer2

1UCLA David Geffen School of Medicine 2University of Chicago Harris School of Public Policy

AcademyHealth ARM June 26, 2017

Introduction Policy Discontinuity Data and Methods Results Conclusions

Introduction

U.S. policy has largely focused on expanding public health insurance to address disparities in child health Strong evidence that public insurance increases use of medical care, but evidence of impact on health is more limited

See comprehensive review in Howell and Kenney (2012)

Only recently have we been able to evaluate the long-term effects of public health insurance coverage for children

2 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Public Health Insurance May Affect Long-Term Health

Payoffs from certain types of medical care might not be evident until later

e.g. preventive services protect healthy children from future risks

In addition, linkages to other social services or freeing up of resources for

• ther investments may have long-term impacts

Long-term effects are an important potential program benefit

3 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Overview of Study

We use a quasi-experimental design that approximates random assignment to identify the effect of public health insurance on health We examine immediate and longer-term effects of childhood Medicaid on mortality for black and white children Findings indicate increase in public coverage decreased longer-term mortality for black children Just one study among many....

4 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Medicaid Expansions for Children

Medicaid eligibility for children historically linked to cash welfare - primarily single-mother families with very low incomes In 1984, Congress began to expand Medicaid eligibility for all children with family incomes below poverty To phase in the expansions, Congress specified that only applied to children born after September 30, 1983

5 / 16

Average Years of Childhood Public Eligibility by Birth Cohort and Family Income

Average Years of Childhood Eligibility 2 4 6 8 10 12 14 16 18 Oct−79 Jan−80 Apr−80 Jul−80 Oct−80 Jan−81 Apr−81 Jul−81 Oct−81 Jan−82 Apr−82 Jul−82 Oct−82 Jan−83 Apr−83 Jul−83 Oct−83 Jan−84 Apr−84 Jul−84 Oct−84 Jan−85 Apr−85 Jul−85 Oct−85 Jan−86 Apr−86 Jul−86 Oct−86 Jan−87 Apr−87 Jul−87 0−24% FPL 25−49% FPL 50−74% FPL 75−99% FPL 100−124% FPL 0.22 years 2.01 years 3.44 years 4.57 years 0.43 years Size of discontinuity = 0.19 years of eligibility 125−150% FPL Birth Cohort

Average Public Eligibility for Each Age of Childhood by Birth Month Cohort

0.0 0.2 0.4 0.6 0.8 1.0 Age in Years Share of Birth Cohort Eligible for Public Health Insurance 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

• Sept. 1983
• Oct. 1983

Ages 4−7 Ages 8−14 Ages 15−18

Introduction Policy Discontinuity Data and Methods Results Conclusions

Differences in Exposure by Child Race

We examine effects of expansions separately by child race

Black children more likely to be poor than white children (45.6% vs. 16.1% in 1991)

Percent Average Gain (in Years) Average Gain (in Years) Gaining for Children Gaining for Total Child Eligibility Eligibility Population Black children 17.13% 4.53 0.82 White children 8.18% 4.48 0.37

8 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Outcomes and Data

Construct rates of death for birth cohorts: Oct. 1979 - Sept. 1987

Number of deaths divided by total population at risk (per 10,000) Data sources:

Multiple Cause Mortality Files, 1979-2011 (restricted) Natality Files, 1979-1987

Distinguish between internal and external causes of death

9 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Empirical Analysis

Estimate effect of Medicaid expansions on mortality at ages 8-14 (immediate effect), and ages 15-18 and 19-23 (longer-term effects) Use a regression discontinuity approach to compare outcomes for cohorts born just before and after September 30, 1983 Graphs and regression analyses model trends in mortality unrelated to Medicaid expansions

Quadratic function used in estimates presented here, alternative specifications reported in paper

Estimated discontinuity at cutoff provides estimate of effect of Medicaid

10 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Internal Mortality for Black Children

(a) Ages 8-14 (b) Ages 15-18 (c) Ages 19-23

• Internal Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Black, Ages 8−14

• Internal Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 1.5 2.0 2.5 3.0

Black, Ages 15−18

• Internal Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 2.5 3.0 3.5 4.0 4.5

Black, Ages 19−23

ˆ β = −0.094 ˆ β = −0.448 ∗ ∗∗ ˆ β = 0.052 (0.087) (0.125) (0.125)

∗∗∗p < 0.01,∗∗ p < 0.05,∗ p < 0.1 11 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Internal Mortality for White Children

(a) Ages 8-14 (b) Ages 15-18 (c) Ages 19-23

• Internal Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 0.7 0.8 0.9 1.0 1.1 1.2 1.3

White, Ages 8−14

• Internal Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 1.0 1.2 1.4 1.6

White, Ages 15−18

• Internal Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 1.4 1.6 1.8 2.0 2.2

White, Ages 19−23

ˆ β = 0.014 ˆ β = 0.022 ˆ β = −0.007 (0.028) (0.046) (0.054)

∗∗∗p < 0.01,∗∗ p < 0.05,∗ p < 0.1 12 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

External Mortality for Black Children

(a) Ages 8-14 (b) Ages 15-18 (c) Ages 19-23

• External Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 1.0 1.5 2.0 2.5

Black, Ages 8−14

• External Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 4 5 6 7 8

Black, Ages 15−18

• External Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 9 10 11 12 13

Black, Ages 19−23

ˆ β = −0.046 ˆ β = 0.041 ˆ β = 0.071 (0.094) (0.278) (0.260)

∗∗∗p < 0.01,∗∗ p < 0.05,∗ p < 0.1 13 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

External Mortality for White Children

(a) Ages 8-14 (b) Ages 15-18 (c) Ages 19-23

• External Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 0.8 1.0 1.2 1.4 1.6

White, Ages 8−14

• External Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 4.0 4.5 5.0 5.5 6.0

White, Ages 15−18

• External Mortality

Jun−80 Feb−82 Oct−83 Jun−85 Feb−87 6.0 6.5 7.0 7.5

White, Ages 19−23

ˆ β = 0.110∗∗∗ ˆ β = 0.085 ˆ β = 0.018 (0.033) (0.101) (0.094)

∗∗∗p < 0.01,∗∗ p < 0.05,∗ p < 0.1 14 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Placebo Test

Estimate jumps at nondiscontinuity points for these outcomes and age ranges (>100 simulations)

White External Mortality, Ages 8−14

β ^ at September 30, 1983 = 0.075 Frequency 5 10 15 20 25 30 −0.15 −0.1 −0.05 0.05 0.1 0.15

Black Internal Mortality, Ages 15−18

β ^ at September 30, 1983 = −0.441 Frequency 5 10 15 20 25 30 −0.4 −0.2 0.2 0.4

Suggests unlikely estimate for blacks due to chance, more likely for whites

Estimate for whites also sensitive to covariate inclusion

15 / 16

Introduction Policy Discontinuity Data and Methods Results Conclusions

Conclusions

Compelling evidence of large decline (19%) in internal mortality for black teens (ages 15-18) as a result of increased childhood Medicaid eligibility

No evidence of similar effect for white children (rule out >4.5%) No reversal at ages 19-23, but no further declines

Limitations of current study:

Susceptible to confounders that differentially affect cohorts at cutoff Rely on variation from the 1990s that may limit generalizability Lack of information on mechanisms Not enough time passed to look at longer-term effects on adult health

16 / 16