The Effects of Government Transfers on Monthly Cycles in Drug Abuse, Crime and Mortality Carlos...

Post on 21-Dec-2015

214 views 1 download

Tags:

Transcript of The Effects of Government Transfers on Monthly Cycles in Drug Abuse, Crime and Mortality Carlos...

The Effects of Government Transfers on Monthly Cycles in Drug Abuse, Crime and

Mortality

Carlos Dobkin (UC Santa Cruz) and Steve Puller (Texas A&M)

Cycles in Drug Consumption: Literature

• Monthly pattern in deaths (Phillips et al., NEJM 1999)– 1% more deaths in first vs. last week of month– 14% more substance abuse deaths in first vs. last week of month

• Monthly pattern in psychiatric admissions (Halpern & Mechem, Am J Med, 2001)– Psychiatric admissions for substance abuse 14% higher first week

(vs. 6% for non-substance abuse)

• Cocaine use by disabled vets (Shaner, NEJM, 1995)– 105 male vets on disability with history of schizophrenia & cocaine

use– Highest cocaine concentration in body during first 3 days of month,

followed by highest number of hospital admissions 3-5 days later

• Economic literature on consumption smoothing – Food stamp recipients do not smooth caloric intake over the month

(Shapiro, 2005)– Social Security recipients do not smooth consumption over the

month (Stephens, 2003)

Contributions of This Paper

• Determine which programs drive the cycle in hospital admissions

• Understand the unintended consequences of government transfer programs on:

– Health (Hospitalization & Death)– Crime

• Discuss implications for alternative means of disbursing aid

– Cycles vs. Levels

Major Cash Aid Programs

• Welfare (AFDC, TANF, CalWORKs)– Averaged about $550/month for family with no income– Received: 1st of month (few exceptions including LA post-97)

• Supplemental Security Income (SSI)– Two-thirds of CA’s 1 million recipients are disabled– About $600/month for individual ($1100 for couples)– Received: 1st of month

• Social Security Disability Insurance (DI)– Partial income replacement if sufficient work history– Eligible for Medicare after 2 years– Received: 3rd of month for most recipients in our sample

• General Relief (GR)– Indigent population not qualifying for welfare– About $200/month for individual– Received: Varies

Data

• California Hospital Discharge Data 1994-2000– Census of hospitalizations– Includes patient demographics, cause of hospitalization and

treatment provided• Medi-Cal Eligibility Data 1994-2000

– Linked to hospital data– Includes individuals receiving welfare and Supplemental Security

Income (SSI) for Aged, Blind, or Disabled– Does not include General Relief – Social Security Disability Insurance (DI) (proxied by Medicare &

under 65)• California Mortality Data 1994-2000

– Census of deaths in California (CA Dept of Health Services)• California Arrest Records 1993-1999

– Census of arrests in California (CA Dept of Justice)

SSI RECIPIENTSAverage Daily Hospital Admission with mention of Cocaine, Amphetamines or Heroin

0

10

20

30

40

50

60

-13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Day 0 is the First Day of the Month

Adm

issi

ons

Per

Day

Pre Jan 97

Post Jan 97

SSI Pre and Post Welfare Reform Act

Possible Causes Other Than Cash Aid Programs

• Weekends, Certain Holidays?– Weekends evenly distributed over 7 yrs– Figure 1 similar if “regression adjust”

• Supply-side factor?• Paycheck effect?

– Negligible cycle for patients w/ private insurance (employed?)

– Paychecks commonly received around 1st and 15th

• Other personal income at beginning of month?

Source: payday loan data from Skiba and Tobacman

Surge in Admissions on the 15th?

Possible Causes Other Than Cash Aid Programs

• Weekends, Certain Holidays?– Weekends evenly distributed over 7 yrs– Figure 1 similar if “regression adjust”

• Supply-side factor?• Paycheck effect?

– Negligible cycle for patients w/ private insurance (employed?)

– Paychecks commonly received around 1st and 15th

• Other personal income at beginning of month?

Welfare SSI MedicareEarnings $1,900 $1,489 $3,437SSI $266 $5,842 $1,113Social Security (including DI) $418 $2,415 $4,373Unemployment and Workers Comp $153 $160 $194Welfare/Public Assistance $5,479 $480 $598Disability (excluding Social Security) $32 $330 $642All other $407 $722 $1,559

Table 2: Personal Income by Program in 1997 for Californians Aged 20-47

Program Individual Covered Under

Alcohol Admissions

Monthly Cycles in Exit Rates

Corroborating Evidence: Welfare Check Receipt in LA County

• June 1997: Los Angeles County changed timing of welfare check disbursement– Pre-June 1997: All checks 1st of month

– Post-June 1997: Staggered over first 10 days of month

• Test for causal effect of welfare checks– Did peak of LA county welfare cycle shift to later in

month?

– Likely a low power test

– Diff-in-diff to allow for other trend

• Comparison group: 10 other largest counties

• Regression Specification:

Simulated Post June 1997 Using the Pattern from Day 1 Disbursement

2

2.5

3

3.5

4

4.5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

LA Pre June 1997 (Day 1 Disbursement)

LA Day 1-10 Simulated

Choosing “Early” and “Middle” Days

Results – LA “Experiment”

Welfare and SSI Recipients in California

0

500,000

1,000,000

1,500,000

2,000,000

2,500,000

3,000,000

1996

-01

1996

-02

1996

-03

1996

-04

1996

-05

1996

-06

1996

-07

1996

-08

1996

-09

1996

-10

1996

-11

1996

-12

1997

-01

1997

-02

1997

-03

1997

-04

1997

-05

1997

-06

1997

-07

1997

-08

1997

-09

1997

-10

1997

-11

1997

-12

Month

Wel

fare

Rec

ipie

nts

0

200,000

400,000

600,000

800,000

1,000,000

1,200,000

SS

I R

ecip

ien

ts

Welfare

SSI

Caseloads Over Time

Mortality

• Increased drug use can increase risk of dying– Drug/alcohol overdoses – Motor Vehicle Accidents– Violence

• Look at deaths in hospital – Know program coverage– Know timing of onset of fatal injury

• Look at overall deaths– Larger sample makes it possible to examine particular

causes

Within Hospital Mortality

Regression-Based Estimate of Change in Outcome at 1st of Month

• Second order polynomial in days from the first of the month fully interacted with a dummy for after the first of the month

• Post: Dummy for event after the 1st of the month

• Days: Days from the first of the month

• We interpret β1<>0 as a “change”

iiiiiiiiii XPostDaysPostDaysDaysDaysPosty ** 254

23210

22% increase

Total Mortality

4% increase

15-16% increase 8-11% increase

Crime

• Arrival of checks may change probability of committing crimes directly or indirectly

• Drug possession and drug sale– Likely to increase as there are more

transactions to interdict

• Revenue generating crime– May decrease or increase

Drug Crimes

Revenue Generating Crimes

Arress in 1993

0

1000

2000

3000

4000

5000

6000

7000

8000

1/1/

93

1/31

/93

3/2/

93

4/1/

93

5/1/

93

5/31

/93

6/30

/93

7/30

/93

8/29

/93

9/28

/93

10/2

8/93

11/2

7/93

12/2

7/93

Day of Year

Cou

nts

of A

rres

ts

Summary of Results on Crime

• Arrests for drug possession and sale population wide increase by 11-28%

• 16% drop in arrests for prostitution

• Modest drop in burglary arrests

Policies to Consider

• Distribute cash aid in smaller, more frequent batches– Reduces prevalence of “full wallets”– May reduce bingeing and number of adverse events– Low cost due to EBT

• Target small subpopulation with repeat drug admissions & substitute in-kind aid– Among 45 thousand SSI recipients ever admitted

from 1994-2000, 1004 recipients (2%) are admitted more than 11 times and represent 16% of SSI admissions

– Expensive

Conclusions

• Cash aid to certain individuals results in an immediate increase in drug use

• Increased drug use facilitated by government transfers can result in considerable harm to the individual receiving the check

• SSI and SSDI cause the cycle in hospitalizations documented in public health literature

The End

All Admission Alcohol Heroin Cocaine Amphetamines External causes (Injury) 0.110 0.181 0.132 0.140 0.152 Alcohol & Drug Dependence 0.007 0.120 0.162 0.122 0.119 Affective Psychosis 0.024 0.084 0.064 0.119 0.113 Alcohol or Drug Psychosis 0.005 0.073 0.134 0.102 0.077 Delivery 0.217 0.007 0.037 0.088 0.143 Circulatory Problems 0.152 0.079 0.046 0.075 0.046 Schizophrenic Disorder 0.013 0.032 0.015 0.066 0.062 Other Mental Conditions 0.013 0.044 0.032 0.050 0.082 Respiratory Problems 0.069 0.051 0.062 0.045 0.029 Infectious Disease 0.020 0.016 0.029 0.020 0.019 Nonorganic Psychosis 0.004 0.008 0.006 0.016 0.036 Neoplasm 0.057 0.017 0.007 0.005 0.004

Primary Cause of Admission (based on ICD-9)

All Admission Alcohol Heroin Cocaine Amphetamines Proportion of Patients Enrolled in Each Program Welfare 0.049 0.024 0.051 0.076 0.131 SSI Aged 0.037 0.014 0.005 0.001 0.000 SSI Blind 0.004 0.002 0.002 0.003 0.001 SSI Disabled 0.093 0.191 0.273 0.281 0.196Admissions per Month per 10K recipients Welfare Rate 48.176 0.983 0.486 0.628 0.810 SSI Aged Rate 238.799 3.816 0.294 0.065 0.019 SSI Blind Rate 323.026 8.502 2.139 1.937 0.825 SSI Disabled Rate 287.914 24.877 8.315 7.434 3.871

Program Coverage of Hospital Admissions

All Admission Alcohol Heroin Cocaine Amphetamines Primary Cause of Admission (based on ICD-9) External causes (Injury) 0.110 0.181 0.132 0.140 0.152 Infectious Disease 0.020 0.016 0.029 0.020 0.019 Respiratory Problems 0.069 0.051 0.062 0.045 0.029 Delivery 0.217 0.007 0.037 0.088 0.143 Circulatory Problems 0.152 0.079 0.046 0.075 0.046 Neoplasm 0.057 0.017 0.007 0.005 0.004 Alcohol or Drug Psychosis 0.005 0.073 0.134 0.102 0.077 Schizophrenic Disorder 0.013 0.032 0.015 0.066 0.062 Affective Psychosis 0.024 0.084 0.064 0.119 0.113 Nonorganic Psychosis 0.004 0.008 0.006 0.016 0.036 Alcohol & Drug Dependence 0.007 0.120 0.162 0.122 0.119 Other Mental Conditions 0.013 0.044 0.032 0.050 0.082

Table 1A: Demographics by Hospital Admission by Type of Drug Mentioned on Admission Record

Employees ProportionWeekly 905,562 0.21Biweekly 2,409,063 0.56Semimonthly (Typically 1st and 15th) 720,142 0.17Monthly 252,101 0.06Other 7,704 0.00Total 4,294,572

Table 4: Timing of Pay Checks

Regression Set up

• Second order polynomial in days from the first of the month fully interacted with a dummy for after the first of the month

• Post: Dummy for event after the 1st of the month

• Days: Days from the first of the month

0 1 2 3 4 5_ * _ *y B B post B days B days sq B days post B days sq post e