|
|
|
![]() ![]()
Papers and Publications
“The Impact of Telecommuting on the Journey to Work: A Two-Sample Instrumental
Variables Approach” (Job Market Paper). |
|
|
|
Telecommuting is
viewed as a public policy tool for reducing congestion
and air pollution, but it may not be as effective as
people think if workers who can telecommute choose to
live (work) farther from their workplaces (homes) than
non-telecommuters. Existing studies have not been able
to fully address this issue because they all assume
telecommuting is exogenous to commuting behavior. In
this study, I assemble information on telecommuting from
the work schedule supplement to the May 2001 Current
Population Survey (CPS) and information on commuting
behavior from the 2000 Census 5% Public Use Micro-data
Series (PUMS) and apply a two-sample instrumental
variables technique to examining the impact of
telecommuting on commute length and mode. I use the
percent of workers who work at home using the Internet
in a person’s two-digit occupation and MSA of the same
size to instrument for telecommuting. The results show
that telecommuting increases a married woman’s one-way
commute time by 9-12 minutes.
For a woman, who used to commute 24
minutes one way, five days a week, her total commute
time is lowered by 17% if she telecommutes two days
every week.
Evidence on the housing
consumption suggests that instead of changing
residential location, married women may take jobs
farther from homes if they can work at home for part of
a week.
[full paper] |
|
|
| “Do
People Drive Less on Code Red Days?" (Dissertation Essay II), 2007. |
|
|
|
This
study examines whether a voluntary information program
--- the Air Quality Action Days program in Baltimore
metropolitan area, is effective in getting cars off the
road on high ozone days. The program forecasts daily
ozone level in the summer and announces a code red day
when ozone is forecast to exceed the EPA standard, i.e.
125 ppb. People are urged not to drive on Code Red days.
First, I argue with a
simple discrete choice model that giving up driving is
not necessarily optimal even for people who take into
account environmental externality of driving.
I look at
traffic volumes on highways in the Baltimore area to see
whether they are lower on code red days. The regression
discontinuity design is used to measure the impact of
the announcement on driving patterns. I find that the
program has generally little effect except that it
reduces morning inbound traffic volumes by 4-5 percent.
To achieve sufficient traffic reduction to indeed
mitigate ozone level, a permit program that restricts
driving episodically deserves consideration.
[full paper] |
|
|
|
"What Drives Telecommuting? The Relative Impact of Worker Demographics,
Employer Characteristics, and Job Types" (with
Margaret Walls
and
Elena Safirova),
Transportation Research Record, 2007, No. 2010, 111-120. |
|
|
|
We analyze a 2002
survey of Southern California residents to evaluate the
relative importance of factors that determine a worker’s
propensity to telecommute and telecommuting frequency.
The survey collected a wealth of individual demographic
information as well as job type, industry, and employer
characteristics from about 5,000 residents. In agreement
with previous studies, we find that the propensity to
telecommute is increasing with worker age and
educational attainment. At the same time, we conclude
that the propensity to telecommute depends to a large
extent on a worker’s job characteristics and that the
quantitative effects of job characteristics are at least
as important as demographic factors. We also study what
factors affect telecommuting frequency based on a
one-week commuting diary of the telecommuters in the
survey. The industry and occupation categories that play
a significant role in affecting propensity to
telecommute do not have similar effects on telecommuting
frequency. On the contrary, some other job-related
factors show substantial influences.
[full paper] |
|
|
|
“A
Note on the Sampling Properties of the Vincentizing (Quantile
Averaging) Procedure”
(with Jeff Rouder and Paul Speckman), Journal of Mathematical
Psychology, 2004, 48(3), 186-195. |
|
|
|
(This work is
based on my Master thesis) To assess the
effect of a manipulation on a response time
distribution, psychologists often use Vincentizing or
quantile averaging to construct group or ‘‘average’’
distributions. We provide a theorem characterizing the
large sample properties of the averaged quantiles when
the individual RT distributions all belong to the same
location-scale family. We then apply the theorem to
estimating parameters for the quantile-averaged
distributions. From the theorem, it is shown that
parameters of the group distribution can be estimated by
generalized least squares. This method provides accurate
estimates of standard errors of parameters and can
therefore be used in formal inference. The method is
benchmarked in a small simulation study against both a
maximum likelihood method and an ordinary least-squares
method. Generalized least squares essentially is the
only method based on the averaged quantiles that is both
unbiased and provides accurate estimates of parameter
standard errors. It is also proved that for
location-scale families, performing generalized least
squares on quantile averages is formally equivalent to
averaging parameter estimates from generalized least
squares performed on individuals. A limitation on the
method is that individual RT distributions must be
members of the same location-scale family.
[full paper] |
|
|
Works in Progress
|
“Episodic
Control of Ozone Precursor Emissions from Mobile Sources” (Dissertation
Essay III). |
|
|
|
This study
examines the cost and feasibility of a
hypothetical scheme in the Washington metropolitan area
that requires people to purchase permits in order to
drive on high ozone days. This scheme should be more
effective than the voluntary programs and cost-effective
than other regulations that require emissions reduction
all year long. The cost of this scheme depends on the
demand for permits, equal to the area under the demand
curve for permits, to the right of the quantity of
permits issued. With EPA grant for dissertation, I am
conducting a stated preference study in the Washington
metropolitan area. The survey collects information on
household demographic characteristics and vehicle
ownership and usage as well as willingness to pay for a
permit. The survey is in field now and I expect to have
data from 1,200 households by the end of February. With
the estimated demand function for permits, we can
determine the cost associated with a given emissions
reduction requirement, compare it with the costs of
other control programs that use year-round controls, and
investigate the distribution of the cost of the scheme. |
|
|
|
“Working
from Home: The Effects of Commute Distance and Congestion?” (with Margaret Walls and Elena Safirova), 2007. |
|
|
|
Although
respondents in many surveys cite congestion and long
commutes as the primary reason they have a strong desire
to work from home, no empirical study has fully explored
the link between commute distance, roadway congestion
and actual telecommuting choice. Also, land use patterns
at both work location and home location may affect
individual telecommuting opportunity and decision. In
this study, we investigate these issues by combining
data from the 2002 SCAG (Southern California Association
of Governments) Telework Survey
with high-resolution land use data as
well as travel times produced by the SCAG travel demand
model at travel analysis zone (TAZ) level. Moreover, the SCAG Survey
allows us to distinguish partial-day telecommuters from
the full-day ones. The preliminary results find that
commute distance increases one's propensity to
telecommute full day but has no impact on partial-day
working from home. Congestion regardless of how it was
measured has little effect on either choice. Work county
is more important in determining telecommuting status
than home county. We try to understand this result in
depth by looking at land use patterns at finer
geographical level. |
|
|
|
"Modeling
Electricity Demand under Time-Differentiated Pricing" (with Dallas Burtraw
and Karen Palmer). |
|
|
|
The project aims
to refine the demand side of a detailed equilibrium
model of the U.S. electricity sector, named Haiku. In
the original Haiku, electricity demand was modeled in a
way that essentially assumes the cross price
elasticities
are zero between time blocks of a day. This may work
well under flat-rate pricing, but is problematic given
Time-of-Use (TOU) pricing. After an extensive survey of
the theoretical and empirical literature on demand
modeling and TOU pricing, we decided to model
electricity demand as constant elasticity of
substitution (CES), which should balance model
flexibility and computational burden reasonably well.
Currently, I am working to develop a routine to
translate empirical estimates in the literature into CES
parameters for Haiku that ideally vary by region, season
and customer types. The difficulties lie in the fact
that the estimates were obtained from studies differing
in numerous aspects and the data used for estimation are
generally not available. We also investigate methods
that set boundaries for the estimates or extrapolate the
estimates in economically sound way. When the CES
structure is integrated into Haiku,
the model opens door to a variety of
research and policy questions. For
instance, we plan to study the impacts and welfare
consequences of a carbon emissions tax on electricity
markets with time-varying pricing. |
|
|
|
|
 |
