Transcript Optimal Taxation of Beer - Univerzita Karlova v Praze

Alcohol in young adults
- recent surveys
Jakub Mikolášek
18.3.2013
Contents
• Motivation
– My research path and data pitfalls
• Prev. results – Drivers of Heavy Episodic Drinking
(as a proxy to Alcohol Abuse)
• The Data
– by National Institute of Public Health
• Key statistics – Figures – striking and misleading...
• The model
– ...until we put them together
• Conclusion
• Next Steps
– Verify, verify, verify
Motivation
To cook an economicpolicy model on alcohol externalities one
needs a lot of ingredients
• Elasticities
–
–
–
–
Varying price of few homogenous commodities
Data on individual beverage‘s consumption
Individual income data
All of it packed in a nice as-long-as-possible pannel.
• Costs & Benefits to the pubic
– Can be found in literature (extrapolated for CZ case)
• Identify Abusers
– Individual dat on consumption patterns (yearly is not
data is generally enough)
– Set of identifiers (e.g. Sociodemographic features)
Motivation
What data is available
• Household Bdget Survey (CSO)
–
–
–
–
Prices & Volumes
Aggregation per household
Yearly statistics
Curse of Averages
Socio-demograpic info limited to 1 household member
Motivation
What data is available
• Household Bdget Survey (CSO)
–
–
–
–
Prices & Volumes
Aggregation per household
Yearly statistics
Curse of Averages
Socio-demograpic info limited to 1 household member
• Czech National Monitoring Centre for Drugs and Drug
Addiction
– Individual data
– Consumption frequencies  Proxy to Abuse patterns
– Non-panel data
Motivation
What data is available
• Household Bdget Survey (CSO)
–
–
–
–
Prices & Volumes
Aggregation per household
Yearly statistics
Curse of Averages
Socio-demograpic info limited to 1 household member
• Czech National Monitoring Centre for Drugs and Drug
Addiction
– Individual data
– Consumption frequencies  Proxy to Abuse patterns
– Non-panel data
• National Institute of Public Health
– Individual data, limited to adults only
– Abuse patterns
Motivation
What data is available?
Household
Budget
Survey
National
Monitoring
Institute
National
Institute of
Public Health
Panel
YES
NO
NO
Individual Beverages
YES
YES
YES
Prices
YES
NO
NO
Income
YES
YES
PARTIAL
Individual Consumption
NO
NO
YES
Frequency
NO
YES
YES
Abusive Patterns
NO
PARTIAL
YES
Smoking
YES
YES
YES
Employment/Soc. status
PARTIAL
YES
YES
Full Age Scale
PARTIAL
YES
PARTIAL
Previous Results
Drivers of heavy episodic drinking
•
•
•
•
•
•
•
•
Being Male
Smoking
Age
Student
Retired
Disabled
Municipality
Income
 +++
 +++
+
 +++
+
(concave shape)
(but adds up with age)
(again adds up with)
- (+ in villages, insigniicant oterwise)
 (generally + but overtall low signif.)
But knowing only frequencies, we weren‘t able to assess
the level of abuse, especially for women.
The Data
National Institute of Public Health
Individual data for 2200+ respondents, aged 18-39
• Sample drawn to reflect key population statisctics
(correspondingly distributed across sex, education and
regions)
• Specifically targetted for alcohol, asking also on tobacco and
other, illicit drugs.
• Includes large set of behavioral and impact questions
• Anables direct assessment of abusive behavior - we can
reliably estimate average daily ethanol consumption of an
individual which is the best indicator of Abuse.
Dependent Variable
„Abuser“ definition
Risky Consumption
WHO defines Risky Alcohol drinking as consumption of
>20g of ethanol/day for Women
>40g of ethanol/day for Men
In our sample of 2190 eligible respondents, such a definition
would net us 526 (23,7%) „abusers“. In men, it is 30,7%, in
women only 16,6%.
Harmful Consumption
NIPH defines 10-factor composite score to assess level of
alcohol addiction and health impact. Abuser: score >=16.
In our sample, this concerns 7,6% of respondents (12,4% in
Men, 2,6% in Women)
Intermezzo
Abuse
in practice
Key Statistics
Male Female Total
Risky
Harmful
30,7%
12,1%
Risky Harmful
16,6% 23,8%
2,6% 7,5%
Risky
Employed
Unemployed -seeking
Student
Retired
Disabled
Maternity
Housewife
Unemployed -not seeking
Harmful
24,6%
34,1%
17,5%
0,0%
6,7%
18,4%
3,5%
0,0%
10,5%
3,1%
10,5%
1,6%
17,2%
66,7%
3,4%
48,1%
Primary
Secondary (w/o grad.)
Secondary (gradua.)
Higher
University
38,2%
24,7%
17,1%
28,6%
16,7%
16,3%
7,6%
4,2%
70%
2,9%
60%
3,0%
First time drunk alcohol Age
50%
40%
30%
20%
10%
0%
13
14
15
16
Risky
17
18
19
20
21
22
Harmful
Abuse x Cigarettes daily
Abuse x Income
80%
30%
70%
25%
60%
20%
50%
15%
40%
30%
10%
20%
5%
10%
0%
0
High
2
Average
4
6
Risky
Harmful
Low
8
0%
10
0
5
10
Risky
15
HARMFUL
20
25
Emipirical analysis - Variables
Children?
Risky average daily ethanol doses
(>40g for Men, >20g for women)
Experiencing alcohol addiction or repeated
Boolean
consequences of its consumption
Boolean Male, Female
Primary, Secondary (w/o graduation),
Categ.
Secondary (graduation), Higher, University
1-499, 500 -1 999, 2000 - 4 999, 5 000 - 19 999,
Categ.
20 000 - 99 999, 100 000+
Boolean YES,NO
Job Type
Categ.
Employed, Unemployed-seeking, Student, Retired,
Disabled, Maternity, Housewife, Unempl.-not seeking
Smoking Status
Categ.
Daily smoker, Occasional smoker,
Former smoker, Non-smoker
Cigarettes daily
Measure No. Of cigarettes smoked a day
Risky
Harmful
Sex
Education
Municipality Size
Boolean
First alcohol drunk - age Measure Age of first experience with alcohol
Method
Logit model (binary logistic regression)
𝑒𝑡
𝜋(𝑥) = 𝑡
𝑒 +1
Where t is a linear expression
𝑡 = 𝛽0 +𝛽1 𝑥1 +𝛽2 𝑥2 +𝛽3 𝑥3 ...+ε
We are looking for the probability of alcohol abuse 𝝅(𝒙)
given set of socio-demographical characteristics 𝒙
An individual is treated as a likely abuser if 𝝅(𝒙) >0,5.
Emipirical analysis – Results
Risky
Harmful
Men
Coef.
Education
Primary
Secondary (w/o grad.)
Secondary (gradua.)
Municipality Size
1-499
500 -1 999
2000 - 4 999
5 000 - 19 999
20 000 - 99 999
Children?
Job Type
Employed
Unemployed -seeking
Student
Disabled
Maternity
Housewife
Smoking Status
Daily smoker
Occasional smoker
Former smoker
Cigarettes daily
First alcohol drunk - age
Constant
Women
S.E.
p-Value
,667
-,040
-,196
,309
,287
,299
,212
,321
-,055
,165
-,142
-,246
,302
,226
,282
,220
,216
,154
-1,082
-1,063
-1,675
-2,972
,596
,641
,676
1,314
,001
,031
,890
,512
,407
,484
,155
,844
,453
,511
,111
,166
,070
,097
,013
,024
,284
,262
,243
,013
,033
,902
,000
,003
,000
,011
,007
,000
,019
,832
1,040
,617
,035
-,162
2,115
Coef.
S.E.
-,223
-,242
-,445
,368
,323
,321
-,452
,108
,006
-,056
-,394
,072
,402
,280
,333
,276
,277
,199
-1,744 ,811
-1,484 ,875
-1,815 ,882
-1,785 1,386
-3,554 ,961
-1,916 ,972
,115 ,378
,880 ,305
,549 ,281
,081 ,022
-,186 ,038
3,156 1,070
Men
p-Value
,265
,544
,454
,165
,487
,261
,699
,986
,840
,155
,718
,017
,031
,090
,039
,198
,000
,049
,019
,761
,004
,051
,000
,000
,003
Coef.
S.E.
Women
p-Value
,683
,084
,013
,490
,480
,499
-,464
,511
,137
-,082
-,258
-,444
,467
,292
,371
,309
,304
,215
-1,865
-1,343
-2,342
-2,140
,528
,590
,716
1,253
,071
,164
,861
,979
,123
,321
,080
,712
,791
,397
,039
,015
,000
,023
,001
,088
,385
,406
,402
,015
,039
1,027
,001
,000
,008
,080
,503
,040
,970
1,477
1,076
,703
,010
-,080
,039
Coef.
S.E.
p-Value
-,018
-,040
,004
1,178
1,138
1,130
,770
,119
-,268
-1,695
-,233
1,077
,770
,656
,911
,946
,677
,542
-3,660
-1,250
-19,658
-,928
-3,453
-3,300
1,381
1,415
4426,103
1,859
1,532
1,791
1,052
2,612
,477
,113
-,046
-1,836
,911
,854
1,063
,035
,064
2,044
1,000
,988
,972
,997
,276
,317
,856
,769
,073
,731
,047
,003
,008
,377
,996
,618
,024
,065
,017
,248
,002
,654
,001
,470
,369
Emipirical analysis – Results
Risky
Predicted
Abuser
Men
Percentage Abuser
Correct
NonAbuser
Observed
Percentage
Correct
NonAbuser
Abuser
687
77
90%
966
8
99%
Non-Abuser
209
130
38%
127
9
7%
74%
Overall Percentage
Women
Harmful
88%
Abuser
865
12
99%
1028
1
100%
Non-Abuser
151
22
13%
23
3
12%
Overall Percentage
84%
98%
Interpretation
Mother
F, SŠ, Non-smok., Maternity, Prague, First drink in 18, child
Pensioner
M, SŠ, Former-smok., Retired,Small town,First drink in 16,child
Good Student
F, VŠ, Non-smoking, Student, Brno, First drink in 20
Wild Student
M, VŠ, Smoking 20 a day, Student, Kladno, First drink in 13
Enterpreneur
M, SŠ, Smoking 40 a day, Employed, Village, First in 15, child
Lost Existence M, ZŠ, Smoking 10 a day, Unempl, Prague, First in 12
Conclusions
• The statistically important positive correlation between
Smoking is proven yet once more
• Sex confirmed to be a strong driver of alcohol abuse (M+/F-)
• Whereas Income is not proven to be a strong driver, Job Type
is. Especially for Students, Disabled, Long-term Unemployed
• Individuals who achieved only Basic Education seem to have
higher propensity to alcohol abuse, differences between
higher education levels are neglgible.
• Alcohol abuse and Age of first alcoholic encounter are highly
correlated. Unfortnately, this does not automatically imply that
increasing the minimum legal age would be efficient since the
majority of respondents had their first drink before age of 18.
Literature
•
•
•
•
•
•
•
•
CZECH NATIONAL MONITORING CENTRE FOR DRUGS
AND DRUG ADDICTION, Whole-population study on
addictive substances, Prague, 2012
AMPHORA (Alcohol Public Health Research Alliance), 2012,
Socio-cultural, economiic & demographic determinants of
unplanned alcohol consumption changes and preventive
alcohol policies.
ANDERSON, P., BAUMBERG, B., 2006 Alcohol in Europe.
London: Institute of Alcohol Studies.
CBMA (Czech Beer and Malt Association), 2007, Report on
the Czech Brewing and Malting Industries, Praha: Enigma s.
r. o.
CSÉMY, L., SOVINOVÁ, H., 2003, Spotřeba alkoholu v
České Republice, Praha: Státní zdravotní ústav.
ERNST&YOUNG, 2006, The contribution made by beer to
the Eropean economy, Amstergam: Ernst&Young
Netherlands.
LINDSAY et al., 2009, What a great night’: The cultural
rs of drinking practices among 14-24 year-old Australians, A
Monash University and Deakin University Consortium,
Australia
•
•
•
•
PŘIBYL, O., 2005, Mikroekonomická analýza spotřeby
alkoholu v ČR, Praha: Charles University in Prague, Faculty
of Social Sciences – Institute of Economic Science.
PYŠNÝ, T., POŠVÁŘ, Z., GURSKÁ, S., 2007, Analysis of
selected demand factors of wine market of the Czech
Republic, Brno: Mendel University of Agriculture and
Forestry
RAJEEV K.G., MOREY, M.J., 1995 The Interdependence of
Cigarette and Liquor Demand, Southern Economic Journal,
Vol. 62,
YAKOVLEV, E., 2011, Peers and Alcohol: Evidence from
Russia, CERGE EI working paper
Thank you for your attention
Contacts:
Jakub Mikolášek
[email protected]