Politicians Have Championed Various Changes in Cash Aid 1994 legislation on SSI & SSDI –3 year limit on benefits for addictive disorders –“representative payee” system Welfare Reform in 1996 –Ended SSI benefits for people categorized as disabled due to substance abuse –Gramm Amendment 1996 – lifetime ban on food stamps and TANF aid to drug offenders Proposition N (San Francisco) 2002 – “Care Not Cash” –Converts General Relief in San Francisco from cash to in kind aid
Literature on Cycles in Drug Consumption Monthly pattern in deaths (Phillips 1999) –14% more substance abuse deaths in first vs. last week of month Monthly pattern in psychiatric admissions (Halpern & Mechem 2001) –Psychiatric admissions for substance abuse 14% higher first week (vs. 6% for non-substance abuse) Cocaine use among disabled vets (Shaner 1995) –105 male vets on disability with history of schizophrenia & cocaine use –Highest cocaine concentration in body during first 3 days of month
Contributions of This Paper 1)Document the monthly cycle in hospital admissions by substance 2)Determine which government programs are driving the monthly cycle in admissions 3)Assess whether alternate disbursal regimes can smooth the monthly cycle 4)Test if cash aid affects the level of admissions or just the timing of admissions
Data California Hospital Discharge Data 1994-2000 –Census of hospitalizations –Includes patient demographics, cause of hospitalization (ICD-9 codes) and treatment provided Medi-Cal Eligibility Data 1994-2000 –Linked to hospital data –Includes eligibility for welfare and Supplemental Security Income (SSI) for Aged, Blind, or Disabled –Does not include General Relief –Proxy for Social Security Disability Insurance (DI)
Possible Causes of the Cycle Monthly Paychecks Supplemental Security Income (SSI/SSP) –For low income aged, blind or disabled – 5 month waiting period for disabled –Approximately 1 Million recipients state-wide, two-thirds disabled –Benefits average about $600/month for individuals ($1100/month couples) –Checks arrive on the 1 st (or last previous business day if on weekend) Social Security Disability Insurance (DI) –Replacement rate varies with income –Eligible for Medicare after 2 years –Aid arrives 3 rd of month Welfare –Largest of the programs with 2 million recipients statewide –Benefits about $550/month in 1997 –Checks typically arrive on the first but there is a little variation by county General Relief –County program for indigent adults (varies by county typically ~ $250 per month) Unemployment Insurance –Paid bi-weekly Workers’ Compensation –Some benefits paid bi-weekly Supply side factors?
“Against Medical Advice” Exit Rates for Welfare, SSI Drug Admits
Summary of Cycles by Programs Welfare – only weak cycle SSI – very strong cycle DI – indirect evidence of a cycle The overall cycle appears to be due largely to SSI and DI Exit patterns (particularly AMA) consistent with patients going home to pick up their checks
Can Alternative Disbursement Schemes Reduce the Cycle? Peaks in drug consumption costly –Stress fixed resources at Emergency Rooms –Hospital admissions may be indicator for other drug- related activity (e.g. crime, policing) Alternative disbursement regimes 1.Stagger the day on which recipients receives monthly aid 2.Disburse individual’s aide more frequently in smaller batches –Low cost due to states switching to electronic benefit transfer
Empirical Evidence on Alternative Disbursement Regimes Los Angeles County disbursement of welfare –Pre June 1997: Day 1 –Post June 1997: Staggered Days 1-10 based on case number (recipients could pick up at issuance outlet after designated day) –Effect: Individuals still have “fat wallets”, but everybody doesn’t have them at the same time Question –Does aggregate cycle change? –Keep in mind Welfare recipients show relatively small cycle Welfare reform may create other changes pre and post
Simulated Effect on Aggregate Drug Cycle of Disbursing SSI, SSDI and Welfare with a Day1-10 Scheme Assume –Cycles for certain subpopulations entirely driven by timing of aide disbursement –Consumption pattern independent across groups (no “agglomeration economies” to consumption) –No supply side response to change in timing of demand Simulate Aggregate cycle –Take observed cycle (from single day disbursement) & simulate the cycle if 1/10 th of recipients = day 1, 1/10 th =day 2, …, 1/10 th = day 10 –SSI: from Day 1 to Days 1-10 –SSDI (proxied by Medicare): from Day 3 to Days 1-10 –Welfare outside LA post 97: Day 1 to Days 1-10 –Welfare in LA post 97: keep the same observed cycle –All others (employed, UI, others) keep the same
Do the Programs Increase Drug Use? Anecdotal evidence of a drug cycle from clinicians is taken to suggest that aid increases level of drug use Problem with simple comparisons: unobserved determinants of drug use correlated with cash aid eligibility Approach: focus on “SSI transitioners” –5 month waiting period –Analyze 1 month window around eligibility –Our data has about 2500 admissions for patients transitioning from other Medi-Cal to SSI Large increase in cash aide No systematic difference in unobservables (hopefully)
Conclusions Monthly cycles in admissions and exits is evidence that aid disbursement affects timing of drug consumption –“fat wallets” aggravate impulse control problems SSI and SSDI are primary contributors to cycle –Targeted changes in disbursement can smooth –Staggered vs. individual smoothing –Appears it should be low cost intervention (Limited) evidence that aid affects timing but not levels
Causes of Monthly Drug Cycles Are Unknown “Fat wallets” early in the month could have a number of causes –Monthly pay checks (but not bi-weekly) –Federal transfers (SSI, SSDI) –State transfers (AFDC/TANF, Food Stamps, UI) –Local transfers (General Relief) Specific early month holidays? –New Year’s or July 4 th Supply side causes?