Date post: | 12-Apr-2017 |
Category: |
Science |
Upload: | cannabissummit |
View: | 98 times |
Download: | 1 times |
Effects of Medical Marijuana Market Growth onSubstance Use and Abuse
Rosanna SmartPh.D. Candidate, Department of Economics
University of California, Los Angeles
Cannabis Science & Policy Summit
April 17, 2016
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 1 / 15
Why study medical marijuana markets?
I 1937, concerns of large externalities led to US prohibitive tax
I Today, movement away from prohibition
DecriminalizationMedical marijuana laws (MMLs)Commercial legalization
1996 2004 2014
I Limited evidence on consequences
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 2 / 15
Why study medical marijuana markets?
I Prior work largely compares outcomes pre- and post-MML enactment
- Results vary depending on years covered, specification, etc.- Implicitly assumes an equal and immediate effect of the law
I Different MML regulations → different effects
- Supply regulations will determine competition, access, price- Effects will depend on implementation
I Policy effects may not be immediate
- Lags in implementation- Changes in federal enforcement
This paper: Does growth in the size of legal markets for medical marijuanaaffect non-medical marijuana use or other health outcomes?
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 3 / 15
Outline of Approach
I Collect new data to measure the size of the legal market
- Counts of registered medical marijuana patients- Document variation between states and over time- Show legal market size responds to changes in supply costs
I Use changes in legal medical market to track effects on recreational use
- Continuous measure accounts for policy dynamics- Allows for heterogeneous effects of market “penetration”- Isolate supply-side effects using production cost shifters as instruments
I Estimate effects on traffic fatalities and substance-related mortality
- Potential age differences in substitution behavior- Separate analysis by age
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 4 / 15
Plan for this talk
1 Effects of State and Federal Policy on Market Growth
2 Data
3 ResultsMarijuana UseTraffic AccidentsAlcohol- and Opioid-Related Mortality
4 Discussion
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 5 / 15
What drives growth in the legal market?
I State variation in supply restrictions
Strictest Self-production (e.g. HI, AK)↓ Limited state-licensed dispensaries (e.g. NM, AZ)
Laxest Unrestricted production (e.g. CO, CA)
I Time variation in federal enforcement
- Pre-2009: federal threats to MML states↓Risk Oct, 2009 (Ogden Memo): de-prioritized prosecution of state-law compliers↑Risk June, 2011 (Cole Memo): reversed Ogden → raids on producers
I If federal enforcement risk represents costs to legal users and producers:
- Ogden Memo should increase market size- Cole Memo should decrease market size* Supply response should be largest in states with lax producer restrictions
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 6 / 15
Pre-Ogden: Registration rates reflect federal barrier to entry
Hawaii: Self-Production
Colorado: Unrestricted CG/Dispensaries
New Mexico: Licensed Dispensaries
Montana: Unrestricted CG/Dispensaries
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 7 / 15
Ogden Memo: States with lax supply restrictions see most growth
Hawaii: Self-Production
Colorado: Unrestricted CG/Dispensaries
New Mexico: Licensed Dispensaries
Montana: Unrestricted CG/Dispensaries
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 7 / 15
Cole Memo: Large-scale producers shut down
Hawaii: Self-Production
Colorado: Unrestricted CG/Dispensaries
New Mexico: Licensed Dispensaries
Montana: Unrestricted CG/Dispensaries
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 7 / 15
Data: Registration rates as a measure of legal market
I Collected registered patient data for 16 of the 18 MML states
- Limitations: Statistics not maintained similarly across states- Strengths: continuous, measures market penetration (intensity of “treatment”)
I Supply shifters
- State supply restrictions- Timing of federal memos
I State-level outcomes- Marijuana consumption: NSDUH (2002-2012)
- Past-month use, past-year initiation by age group (12-17, 18-25, 26+)
- Externalities and substitution
- Traffic fatalities: FARS (1990-2013)- Alcohol- and opioid-poisoning deaths: CDC (1999-2013)
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 8 / 15
Results: Effects of Legal Market Growth on Recent Marijuana Use
Federal Memo Interactions as Instruments for Registration RatesEffects on Prevalence of Population Reporting Past-Month Marijuana Use
Age 12-17 Age 18-25 Age 26+IV OLS IV OLS IV OLS
Registration Rate 0.428* 0.442** 1.65*** 1.21*** 0.937*** 0.865***(0.223) (0.163) (0.405) (0.232) (0.283) (0.207)
[5.9] [6.1] [9.5] [7.0] [21.2] [19.6]
Endogeneity p-val 0.90 0.21 0.86Hansen J p-val 0.35 0.89 0.28
Mean Outcome 7.2 17.3 4.4
N=539. Regressions include state/year FE and state-level covariates. Robust SE clustered at state leveland implied percent change in use in square brackets.
Yjt = β0 + β1ˆRRjt +X ′jtδ2 + uj + vt + εjt
RRjt = γ0 + γ1[m−Ogden]Loosejt + γ2[m− Cole]Loosejt + γ3Loosejt +X ′jtδ1 + uj + vt + µjt
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 9 / 15
Compare Market Growth to MML Enactment: Past-Month Use
Effects of Law (MML) vs. Legal Market Size (Registration Rate) onShare of Population Reporting Past-Month Use
Ages 12-17 Ages 18-25 Ages 26+
My specification: Registration Rate as Policy Variable of Interest
Registration Rate 0.442*** 0.387*** 1.212*** 0.588*** 0.865*** 0.463***(0.163) (0.129) (0.232) (0.209) (0.207) (0.143)
[6.1] [5.3] [7.0] [3.4] [19.6] [10.5]
Past work specification: MML Enactment as Policy Variable of Interest
MML=1 0.450** -0.118 0.853 0.822 0.852*** 0.056(0.220) (0.388) (0.551) (0.549) (0.172) (0.162)
[6.2] [-1.4] [4.9] [4.7] [19.4] [1.1]
Mean of outcome 7.2 7.2 17.3 17.3 4.4 4.4
State-specific trends N Y N Y N Y
N=539. WA, ME excluded. Registration rate and MML included separately. Regressions include S/Y FE andstate covariates. Robust SE clustered at state level and implied percent change in square brackets. Meanregistration rate for MML states is 0.24 (SD=0.56).
Yjt = β0 + β1RRjt +X ′jtδ2 + uj + vt + εjt
Yjt = γ0 + γ1MMLjt +X ′jt θ2 + uj + vt + εjt
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 10 / 15
Mortality From Motor Vehicle Accidents (1990-2013)
Effects on Traffic Fatalities (Single Vehicle), by Age of Driver InvolvedPredicted % Change in Fatalities from 1pp Increase in Registration Rate (95% CI)
Age 15-20 Age 21-24 Age 25-44 Age 45-64
Mean Total Fatalities 145.1 120.7 348.4 176.3
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 11 / 15
Mortality From Motor Vehicle Accidents (1990-2013)
Effects on Traffic Fatalities (Single Vehicle), by Age of Driver InvolvedPredicted % Change in Fatalities from 1pp Increase in Registration Rate (95% CI)
Age 15-20 Age 21-24 Age 25-44 Age 45-64
Mean Alcohol-Related Fatalities 46.0 60.2 153.5 50.5
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 11 / 15
Mortality From Motor Vehicle Accidents (1990-2013)
Effects on Traffic Fatalities (Single Vehicle), by Age of Driver InvolvedPredicted % Change in Fatalities from 1pp Increase in Registration Rate (95% CI)
Age 15-20 Age 21-24 Age 25-44 Age 45-64
Mean Weekend Fatalities 74.4 63.9 169.8 71.9
Mean Nighttime Fatalities 96.8 87.7 224.1 88.4
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 11 / 15
Mortality From Motor Vehicle Accidents (1990-2013)
Effects on Traffic Fatalities (Single Vehicle), by Age of Driver InvolvedPredicted % Change in Fatalities from 1pp Increase in Registration Rate (95% CI)
Age 15-20 Age 21-24 Age 25-44 Age 45-64
Mean BAC=0 Fatalities 44.9 24.8 73.0 52.5
Mean BAC>0 Fatalities 36.7 50.5 129.1 44.1
Mean Cannabis-involved Fatalities 8.3 7.4 13.1 3.9
Mean Both-involved Fatalities 4.1 4.9 8.5 2.2
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 12 / 15
Substance-Related Poisoning Mortality (1999-2013)
Registration rate effects on deaths by Alcohol, Prescription Opioids, and HeroinPredicted % Change in Deaths from 1pp Increase in Registration Rate (95% CI)
O: State, Year FE +Covariates, ∆: +State-specific Linear Trends
Age 15-24 Age 25-44 Age 45-64
Mean Alcohol-Related Poisonings 25.4 131.1 143.1
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 13 / 15
Substance-Related Poisoning Mortality (1999-2013)
Registration rate effects on deaths by Alcohol, Prescription Opioids, and HeroinPredicted % Change in Deaths from 1pp Increase in Registration Rate (95% CI)
O: State, Year FE +Covariates, ∆: +State-specific Linear Trends
Age 15-24 Age 25-44 Age 45-64
Mean Opioid-Related Poisonings 44.7 207.4 216.2
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 13 / 15
Substance-Related Poisoning Mortality (1999-2013)
Registration rate effects on deaths by Alcohol, Prescription Opioids, and HeroinPredicted % Change in Deaths from 1pp Increase in Registration Rate (95% CI)
O: State, Year FE +Covariates, ∆: +State-specific Linear Trends
Age 15-24 Age 25-44 Age 45-64
Mean Heroin-Related Poisonings 19.9 71.0 47.3
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 13 / 15
Summary of Findings
I Laws reducing supply costs have large effects on legal market size
- MML passage alone has little effect- Federal legalization may have larger effect than suggested by state analyses
I Growth in medical marijuana market size increases adolescent use
- 100 more adult legal users leads to 6 more adolescent users- MML measure misses differences across states and changes over time
I Age differences in substitution behavior generate welfare trade-off- Youths: jointly use alcohol and marijuana
- 6-9% ↑ in traffic fatalities caused by young drivers
- Older Adults: marijuana is a substitute for heavy alcohol and opioid use
- 6% and 11% ↓ in alcohol- and opioid-related deaths
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 14 / 15
Limitations
I Only have a rough measure of marijuana consumption
- Increased casual use vs. daily use have different health implications- Use is self-reported
I Cannot directly determine mechanisms for spillovers to adolescents
- Evidence suggests ↑ access and ↓ prices important channels- Does access change through formal market, black-market, secondary markets?
I By no means a complete cost-benefit analysis
- No discussion of revenues or cost-savings to the state- Measure contemporaneous and “severe” effects- Other outcomes could include dependence, tobacco use, productivity,
educational attainment, crime
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 15 / 15
Thank you!
DRAFT: PLEASE DO NOT CITE WITHOUT AUTHOR’S PERMISSION 15 / 15