Nola Ogunro , Ellen Hahn , Eric Thompson and

advertisement
Nola Ogunro1, Ellen Hahn1, Eric Thompson2 and
Aaron Yelowitz1
University of Kentucky
2 University of Nebraska-Lincoln
1
Presented at the 49th Annual Meeting of the Western Regional Science Association (02-23-10)
The authors’ wish to thank the Robert Wood Johnson Foundation for financial support
`
`
During the 1990s/early 2000s, selected Arizona
municipalities passed local ordinances eliminating
smoking in bars and restaurants
Among the implications of these ordinances, a change in
h
hourly
l wages could
ld result
lt since
i
workers
k
no llonger need
d tto
be compensated for working in a smoky environment
◦ Long-term health risk (cancer, heart disease, others)
g
coughing,
g
g smoky clothes and
◦ Current “nuisance” (eye irrigation,
h )
hair)
`
This paper estimates such a compensating differential for
“front-of-the-house”
front of the house workers in smoking-allowed
smoking allowed
restaurants and bars in Arizona during the 1999 to 2004
period
`
`
`
`
Selective adoption of smoke-free laws suggests
quasi-experiment
quasi experiment
There also may be differences in exposure between
“f
“front-of-house”
t f h
” and
d “back-of-house”
“b k f h
” workers.
k
Access to payroll records from Arizona stores of
national franchisee for a mid-priced restaurant (with
bar) chain operating in many Arizona municipalities
Payroll records avoid potential bias with self-reported
wage data from surveys
`
`
`
`
200,000 pay periods from 9,000 workers at all of
the
Arizona
restaurants over the
h franchisees
f
hi
A i
h April
A il
1999 to November 2004 period
Quarterly “bonuses”
bonuses were allocated to two-week
two week
pay-periods during the quarter based on hours
worked in each pay-period. Tips are included.
Payroll records included data such as age,
gender, race, and occupation. Data on tenure
also was available
available.
No information on labor supply variables such as
education, marital status, and whether a student
`
Pre- and post-treatment data are available for restaurants
that were smoke
free and open before adoption of
smoke-free
ordinance (use these cases):
‚
‚
‚
`
Tucson (3),
Tempe (2),
Chandler (1)
Only post-treatment data available for some restaurants
(don’t use):
◦ Opened during sample period, but after adoption of ordinance
‚
‚
Tucson (1)
Gilbert (1)
◦ Opened before the sample period
‚
`
Mesa (4)
Eleven control restaurants (no smoke-free ordinance
during sample period (use)
lnWAGESijt=b0+b1TIMEt+b2CITYj+b3Xi+b4NONSMOKINGijt+eijt
where components of X are:
age and age square
j b tenure and
job
d job
j b tenure square, and
d
dummy variables indicating race,
gender
gender, job occupation
lnWAGESijtk=b0+b1TIMEt+b2CITYj+b3FRONTk+b4CITYj
×TIMEt+ b5CITYj×FRONTk+b6TIMEt×FRONT+
b7NONSMOKINGijtk+b8Xi+eijtk
where “Front”
Front refers to front of the house
workers such as bartenders, servers and
hosts
`
Wages rise (at a decreasing rate) with age and
tenure
`
F
Females
l earn 4% lless than
h males
l
`
Non whites earn 5% less
Clustering on
Employee ID
Fixed
Effects
Random
Effect
DD
-0.003
(0.009)
-0.033** -0.029**
(0.003)
(0.003)
DDD
-0.049*
(0
022)
(0.022)
-0.018** -0.018**
(0
006)
(0
006)
(0.006)
(0.006)
*=significant at 5%, ** signification at 1%
`
`
We find a 2 to 5% drop in wages
This represents a compensating differential
off $0.17
$0 17 to $0.45
$0 45
`
Repace and Lowrey model (lung cancer risk)– an average excess lung
cancer risk of 0.0001
0 0001 assuming a 2 year workplace exposure.
exposure
`
($0.31 X 1500hr/yr X 2 yrs)/.0001 = $9.3 million
`
Similar to $7 million estimate of Viscusi and Aldy (2003)
`
There are several important difficulties with the $9.3
$9 3 million figure
that suggest that it may be an over-estimate
◦ Only focuses on lung cancer risk and excludes other risks such as
heart disease
◦ Estimate likely capture “nuisance” effects of working in a smokefree environment such as eye irritation, coughing, and smoke
smell on close and hair
`
`
`
`
Selective adoption of municipal smoking
ordinances in Arizona in the 1990s created a
quasi-experiment to study labor market impacts
Our access to payroll data allows us to avoid
difficulties with self
self-reported
reported wages
Results from a difference-in-difference-indifference model estimates suggest a
compensating differential of between $0
$0.17/hr
17/hr to
$0.45/hr to work in a smoking-allowed
restaurant
These rates suggest a value of life estimate “in
in
the millions” though several caveats lead to
imprecision in the estimates.
Download