Medicaid Insurance in Old Age. p. 1/31 MEDICAID INSURANCE IN OLD AGE Mariacristina De Nardi Eric French John Bailey Jones
Medicaid Insurance in Old Age. p. 2/31 Our previous findings Retirees with higher permanent income live longer and hold lots of assets even when very old. Out-of-pocket medical expenses that rise with permanent income and age go a long way towards explaining these saving patterns. Means-tested health insurance (such as Medicaid) has a large impact on the savings of the retirees across the income distribution.
Medicaid Insurance in Old Age. p. 3/31 Public Health Insurance for the elderly Medicare: Virtually everyone age 65+ is eligible No income or asset tests Pays for most medical medical services, but not all (e.g., nursing homes) Medicaid: Means-tested health insurance which covers nursing home costs if poor or impoverished Medicaid is currently paying for 70% of all nights spend in nursing homes. Nursing homes are very expensive.
Medicaid Insurance in Old Age. p. 4/31 This paper: questions Medicaid was designed to insure the poorest retirees against medical expenses. We ask: Do people with higher income end up on Medicaid? How big are Medicaid payments for high-income versus low-income people? What is the degree of Medicaid redistribution? How much do people across the income distribution value Medicaid insurance? Is Medicaid of about the right size?
Medicaid Insurance in Old Age. p. 5/31 AHEAD cohort of HRS Household heads aged 70 or older in 1993/4 Consider only the retired singles Follow-up interviews in 1995/6, 1998, 2000, 2002, 2004, 2006, 2008, 2010 Asset, medical expense data begins in 1996 (1994 data faulty), uses 2,673 individuals Use full, unbalanced panel Permanent income = annuity income from Social Security + defined benefit pension plans
Medicaid Insurance in Old Age. p. 6/31 Net Worth Net Worth by age, birth cohort and permanent income quintile.
Medicaid Insurance in Old Age. p. 7/31 Share receiving Medicaid Fraction receiving Medicaid by age, birth cohort and permanent income quintile.
Medicaid Insurance in Old Age. p. 8/31 Forces working against redistribution Heterogeneity life expectancy High income live longer than low income medical expenses High income spend more on medical care than low income Two pathways to qualify for Medicaid (lifelong poor or impoverished by medical expenses)
Medicaid Insurance in Old Age. p. 9/31 Literature on Medicaid/Medicare insurance Lots of work on health effects of Medicaid/Medicare. Less on Medicaid/Medicare s effect on savings/medical expenses. We focus on this.
Medicaid Insurance in Old Age. p. 10/31 Key model features Single people aged 70 and older Consumption of medical and non-medical goods, and savings decision Medical expenditure increases one s well being but does not affect health Consistent with many papers. Much of medical spending, especially late in life, is on long-term care. Spending improves quality of life, not length of life.
Nursing home quality varies a lot Medicaid Insurance in Old Age. p. 11/31
Medicaid Insurance in Old Age. p. 12/31 Model Single people aged 70 and older Flow utility from medical and non-medical consumption u(c t,m t,µ t ) = 1 1 ν c1 ν t +µ t 1 1 ω m1 ω t, where: t = age; c t = non-medical consumption; m t = consumption of medical goods and services, includes nursing home, drugs, doctor visits; items paid out of pocket as well as by insurers µ t = stochastic medical needs shifter.
Medicaid Insurance in Old Age. p. 13/31 Medical needs shocks log(µ t ( )) = f(age, health status, ψ t ), ψ t = ζ t +ξ t, ξ t N(0,σξ 2 ), ζ t = ρ m ζ t 1 +ǫ t, ǫ t N(0,σǫ). 2 Health takes on 3 states: good, bad, nursing home.
Medicaid Insurance in Old Age. p. 14/31 Health and lifespan Women, rich, healthy, live longer than men, poor and sick. Health transition probabilities and mortality rates are functions of: gender permanent income age past health
Medicaid Insurance in Old Age. p. 15/31 Two key features of the insurance system Private and Medicare (plus VA) health insurance pay a share of total medical expenditurem t (1 q(h t )) q(nursing home t ) =.90 (from MCBS) q(good or bad t ) =.29 (from MCBS) Social insurance programs (Medicaid and SSI) Medicaid utility floors
Medicaid Insurance in Old Age. p. 16/31 Medicaid as providing utility floor Medicaid transfers vary with medical needs. But we (the econometricians) do not fully observe medical needs directly. So we need a model that satisfies the criteria: Medicaid transfers vary with medical needs. Model matches distribution of Medicaid payments.
Medicaid Insurance in Old Age. p. 17/31 Medicaid as providing utility floor Calculating transfers for the medically needy... Government computes minimum expenditure x mt = c t +q(h t )m t to achieve 1 1 1 ν c1 ν t +µ t 1 ω m1 ω t = u, Government makes transferb mt given expenditurex mt b mt = max { 0, x mt resources t }, Given exogenous transfer, the person makes optimal decisions.
Medicaid Insurance in Old Age. p. 18/31 Modeling Medicaid tests and transfers Categorically needy: assets and income test Medically needy: total resources low compared to medical expenses Two eligibility criteria
Medicaid Insurance in Old Age. p. 19/31 V t (a t,g,h t,i,ζ t,ξ t ) = Recursive formulation max c t,m t,a t+1,i M { 1 1 ν c1 ν t + µ(h t,ζ t,ξ t,t) 1 ω m 1 ω t ) + βs g,h,i,t E t (V t+1 (a t+1,g,h t+1,i,ζ t+1,ξ t+1 ) + β(1 s g,h,i,t )θ (e(a t+1)+k) 1 ν (1 ν) }, s.t. a t+1 = a t +y n (ra t +y t )+b t I M c t m t q(h t ) 0, b t = b(t,a t,g,h t,i,ζ t,ξ t,u) a t+1 A d, if I M > 0.
Medicaid Insurance in Old Age. p. 20/31 Two-step estimation strategy First step: estimate parameters of income, health, mortality, and co-pay profiles. Second step: taking as given the estimated first-step parameters, choose preference parameters, utility floor, and medical needs shocks to match Median assets Medicaid recipiency rate Median and 90th percentile of out-of-pocket medical expenditures First and second autocorrelations of medical expenditures by PI quintile, cohort and age, using the method of simulated moments (MSM).
Medicaid Insurance in Old Age. p. 21/31 Results from first step estimation Look at Income quintiles Life expectancy Nursing home risk
Income quintiles by age Medicaid Insurance in Old Age. p. 22/31
Medicaid Insurance in Old Age. p. 23/31 Males Females Income Nursing Bad Good Nursing Bad Good Percentile Home Health Health Home Health Health All 10 2.2 5.9 7.2 3.0 9.7 11.8 10.2 50 2.2 7.1 9.1 3.3 11.7 14.1 12.3 90 2.2 8.6 11.0 3.6 13.7 16.0 14.2 Men 9.4 Women 13.3 Bad Health 10.5 Good Health 13.7 Life expectancy at age 70.
Males Females Income Bad Good Bad Good Percentile Health Health Health Health All 10 26.7 28.1 37.3 39.9 36.7 50 26.1 27.9 38.9 42.5 38.5 90 24.8 28.2 39.9 43.8 39.5 Men 27.3 Women 40.9 Bad Health 36.0 Good Health 39.6 Percent ever entering a nursing home, people alive at age 70. Medicaid Insurance in Old Age. p. 24/31
Medicaid Insurance in Old Age. p. 25/31 Second step estimates flow utility = bequest utility Estimated parameters: 1 1 1 ν c1 ν t +µ t 1 ω m1 ω t = θ (e(a t+1)+k) 1 ν (1 ν) β =.995, ν = 3.0, ω = 3.4 model-predicted price elasticity of medical expenditure = 0.27. Categorically needy income limit = $6,420. Utility floor (consumption equivalent) = $5,260. Bequest motives parameters imply MPC out of terminal wealth of 17% and bequest threshold of $3,600.
Medicaid Insurance in Old Age. p. 26/31 Medicaid payments Out-of-pocket expenses Income MCBS Model MCBS AHEAD Model Quintile Data Data Data Bottom 6,170 4,860 3,850 2,380 1,800 Fourth 4,220 3,610 4,740 3,810 3,510 Third 2,030 2,450 5,190 4,480 5,450 Second 1,080 1,460 5,880 5,520 6,470 Top 900 810 6,820 6,390 7,620 Men 2,040 1,820 4,540 4,090 4,780 Women 3,080 2,760 5,480 4,330 6,230 Average Medicaid payments and out-of-pocket expenses.
Medicaid Insurance in Old Age. p. 27/31 Present Income Discounted Annuity Quintile Value Value Bottom 25,200 3,540 Fourth 19,200 2,470 Third 12,600 1,500 Second 8,800 1,010 Top 4,300 490 Men 6,400 970 Women 14,200 1,650 Good Health 11,000 1,210 Bad Health 15,900 2,170 Medicaid Payments at age 74.
Medicaid Insurance in Old Age. p. 28/31 Changing Medicaid Fix preference parameters at baseline estimates and Reduce consumption value of categorically needy floor by 25% Reduce consumption value of both floors by 25% Increase consumption value of both floors by 10%
CAT floor down 25% Both floors down 25% Change in Change in Income Discounted Individual Discounted Individual Quintile Payments Valuation Payments Valuation Bottom 7,200 9,600 8,300 10,900 Fourth 1,200 2,300 6,500 6,600 Third 0 0 4,400 6,100 Second 0 0 3,100 7,200 Top 0 0 1,600 9,000 Men 400 500 2,200 4,500 Women 1,700 2,500 4,800 8,900 Good Health 1,000 1,600 3,800 8,500 Bad Health 2,300 3,100 5,400 7,700 Cutting Medicaid at age 74. Medicaid Insurance in Old Age. p. 29/31
Both floors up 10% Change in Income Discounted Individual Quintile Payments Valuation Bottom 3,900 1,700 Fourth 3,000 1,400 Third 1,800 800 Second 1,300 800 Top 700 1,300 Men 1,000 200 Women 2,200 1,400 Good Health 1,600 1,200 Bad Health 2,400 1,200 Increasing Medicaid at age 74. Medicaid Insurance in Old Age. p. 30/31
Medicaid Insurance in Old Age. p. 31/31 Key Findings High income people... end up on Medicaid if live long enough can receive large Medicaid transfers value these transfers a lot Medicaid provides valuable insurance and, for the current retirees, its size is about right.