The Labor Market Impacts of the 2010 Deepwater Horizon Oil Spill and Offshore Oil Drilling Moratorium (NBER)
Summary
NBER Working Paper 20409, August 2014, by Joseph E. Aldy of the Harvard Kennedy School, estimates the net effects of the 2010 Deepwater Horizon oil spill, the spill response and the offshore drilling moratorium on Gulf Coast employment and wages. It recounts that the Transocean Deepwater Horizon suffered a blowout on April 20, 2010 in a BP lease in the Gulf of Mexico, and that the Department of the Interior suspended offshore deep water drilling on May 27, 2010. Despite predictions of major job losses in Louisiana, Aldy finds that Louisiana coastal parishes, and oil-intensive parishes in particular, saw a net increase in employment and wages. Gulf Coast Florida counties, especially those south of the Panhandle, saw a decline in employment.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
- Issuer
- Source documents
- Document type
- The Labor Market Impacts Of The 2010 Deepwater Horizon Oil Spill And Off Shore Oil Drillin
- Case
- The Labor Market Impacts Of The 2010 Deepwater Horizon Oil Spill And Off Shore Oil Drilling Moratorium
Cite this document
“The Labor Market Impacts of the 2010 Deepwater Horizon Oil Spill and Offshore Oil Drilling Moratorium (NBER).”
Original at www.nber.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/the-labor-market-impacts-of-the-2010-deepwater-horizon-oil-spill-and-off-shore-oil-drilling-moratorium/
, SHA-256 e16d9e410c9e0a2bf6467d83c7772fc37add35f271b56da520f795f7e3539464.Full text
NBER WORKING PAPER SERIES
THE LABOR MARKET IMPACTS OF THE 2010 DEEPWATER HORIZON OIL
SPILL AND OFFSHORE OIL DRILLING MORATORIUM
Joseph E. Aldy
Working Paper 20409
http://www.nber.org/papers/w20409
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
August 2014
Susie Chung, Napat Jatusripitak, Brett Long, and Carlos Paez provided excellent research assistance
for this project. Jonathan Abramson, Julia Hein, Juliette Kayyem, Mary Landry, and Pete Neffenger
assisted in the provision of data on the spill and spill response. Ed Glaeser, Josh Goodman, Bill Hogan,
Dick Morgenstern, Erich Muehlegger, Danny Shoag, Rob Stavins and seminar participants at Harvard,
the AERE 2013 Summer Conference Sponsored Sessions, and the IZA 2013 Workshop on Labor Market
Effects of Environmental Policies provided useful comments on an earlier draft. Research support
was provided by the Taubman Center for State and Local Government at the Harvard Kennedy School.
The views expressed herein are those of the author and do not necessarily reflect the views of the National
Bureau of Economic Research.
NBER working papers are circulated for discussion and comment purposes. They have not been peer-
reviewed or been subject to the review by the NBER Board of Directors that accompanies official
NBER publications.
© 2014 by Joseph E. Aldy. All rights reserved. Short sections of text, not to exceed two paragraphs,
may be quoted without explicit permission provided that full credit, including © notice, is given to
the source.
The Labor Market Impacts of the 2010 Deepwater Horizon Oil Spill and Offshore Oil Drilling
Moratorium
Joseph E. Aldy
NBER Working Paper No. 20409
August 2014
JEL No. J30,J64,Q40
ABSTRACT
In 2010, the Gulf Coast experienced the largest oil spill, the greatest mobilization of spill response
resources, and the first Gulf-wide deepwater drilling moratorium in U.S. history. Taking advantage
of the unexpected nature of the spill and drilling moratorium, I estimate the net effects of these events
on Gulf Coast employment and wages. Despite predictions of major job losses in Louisiana — resulting
from the spill and the drilling moratorium — I find that Louisiana coastal parishes, and oil-intensive
parishes in particular, experienced a net increase in employment and wages. In contrast, Gulf Coast
Florida counties, especially those south of the Panhandle, experienced a decline in employment. Analysis
of accommodation industry employment and wage, business establishment count, sales tax, and commercial
air arrival data likewise show positive economic activity impacts in the oil-intensive coastal parishes
of Louisiana and reduced economic activity along the Non-Panhandle Florida Gulf Coast.
Joseph E. Aldy
Harvard Kennedy School
Taubman 382, Mailbox 57
79 JFK Street
Cambridge, MA 02138
and NBER
joseph_aldy@hks.harvard.edu
1 Introduction
On April 20, 2010, the Transocean Deepwater Horizon suffered a catastrophic
blowout while drilling in a BP lease in the Gulf of Mexico’s Macondo Prospect
that resulted in the largest oil spill in U.S. history. Due to the ongoing spill and
concerns about the safety of offshore oil drilling, the U.S. Department of the
Interior suspended offshore deep water oil and gas drilling operations on May
27, 2010, in what became known as the offshore drilling moratorium. The media
portrayed the impacts of these events on local employment, with closed fisheries,
idle rigs, as well as boats skimming oil and workers cleaning oiled beaches.
This paper examines the net impact of the oil spill, spill response, and the
drilling moratorium on employment and wages in the Gulf Coast. The spill and
moratorium represented unexpected events in the Gulf Coast region. Coastal
counties and parishes in this region were expected to bear the vast majority of
the effects of these two events, while inland areas were expected to be largely
unaffected. The moratorium was expected to affect Louisiana — with significant
support of the offshore drilling industry — but not, for example, Florida, which
had no active drilling off of its coastline. The timing and magnitude of the spill
response varied across the states over the course of the spill as well.
These characteristics of the spill, spill response, and moratorium motivate
an event study difference-in-differences strategy to estimate the impacts of these
events on the local labor markets. In this framework, the spill, spill response,
and moratorium are considered exogenous events that “treat” the coastal coun-
ties, as defined by their hydrologic characteristics by the National Oceanic and
Atmospheric Administration, in the Gulf region. Given the surprise nature of
these events, they can credibly be characterized as exogenous and hence the
pre-event period should not include anticipatory behavior in the Gulf economy.
I categorized five Louisiana “oil parishes” identified by the U.S. government
as most active in support of offshore drilling activities for “treatment” by the
drilling moratorium.
To estimate the net effects of these events on employment and wages, I use
monthly county /parish-level data (quarterly data for wages) from the Quarterly
Census of Employment and Wages (QCEW). Figure [ijillustrates graphically the
employment levels in the Louisiana oil parishes, the non-oil coastal parishes of
Louisiana, counties on the Florida Gulf Coast, and the control (inland) counties
in the Gulf States and identifies the dates of the spill and the drilling mora-
torium. To facilitate comparisons over 2010, I have indexed total employment
for these four groups of counties such that they each equal 1.0 in January 2010.
Employment growth among these four groups follows a nearly identical trajec-
tory through April 2010 and then a significant divergence occurs. Florida Gulf
Coast counties experience a significant drop in employment from May through
July and then begin to experience employment growth immediately after the
capping of the well (July 15). The non-oil parishes on the Louisiana coast track
quite closely the employment path of the inland counties throughout the spill. In
contrast, the Louisiana oil parishes had effectively flat employment throughout
the spill and moratorium and avoided the decline in employment that affected
these other Gulf Coast regions during May, June, and July of 2010.
In statistical analysis based on 2010 data, I find that the net employment
effect of the spill, spill response, and moratorium is a fairly precise zero for
most parts of the Gulf Coast during 2010. In particular, the coastal counties
of Texas, Mississippi, and the Florida Panhandle all experienced net job im-
pacts that cannot be statistically distinguished from zero. Three Gulf Coast
sub-regions experienced statistically significant changes in employment. I find
that the most oil-intensive parishes in Louisiana witnessed a 1.2% increase in
employment (95% confidence interval, 0.4 to 2.0%), and the Alabama coastal
counties experienced a 1.3% increase in employment as well. In contrast, the
Non-Panhandle Florida counties on the Gulf Coast experienced a 2.7% decrease
in employment (95% confidence interval, -1.9 to -3.5%).
Complementing these employment impacts, I find that the Louisiana oil
parishes experienced a statistically significant increase in the average wage of
about 2%. Likewise, the Alabama coastal counties enjoyed higher wages of
about 4 to 6% during the spill. The rest of the Gulf Coast counties and parishes
experienced economically small and statistically insignificant impacts on wages.
I undertook an array of robustness checks of the base econometric model.
First, I accounted for seasonality in labor markets by extending the panels to
cover the 2008-2010 period and I permitted seasonality (month fixed effects)
to vary by coastal regions and by states in some specifications. Second, I em-
ployed a modified definition of coastal counties/parishes. Third, I substituted
U.S. non-Gulf State counties as controls for the inland Gulf State counties and
parishes. Finally, I omitted Texas observations, since the state was effectively
“up current” from the spill and largely unaffected by the spill events. The ad-
verse employment impacts in the Non-Panhandle Florida coastal counties hold
across nearly all robustness checks. The results for Louisiana parishes are mixed
across the robustness checks, with non-oil parishes experiencing statistically sig-
nificant positive employment impacts when accounting for seasonality and with
some variations in geographic controls. The Louisiana oil parishes typically have
either statistically significant 1-2% employment increases or smaller estimates
that cannot be statistically distinguished from zero. The positive and statis-
tically significant wage impacts in the Louisiana oil parishes and the Alabama
hold across virtually all robustness checks.
I conducted a variety of external validity checks. First, I implemented the
employment and wage models for the mining support industry — expected to
be impacted by the drilling moratorium — and the accommodation industry —
expected to be impacted by the spill and spill response. There is some mixed
evidence of statistical declines in mining support employment in the Louisiana
oil parishes and Alabama coastal counties, but no evidence of adverse impacts
on wages in this industry. The Non-Panhandle Florida coast experienced statis-
tically significant declines in accommodation industry employment and wages.
Second, I estimated the impacts of the spill on the number of business estab-
lishments, and found statistically significant increases of 1-5% for Louisiana oil
parishes and non-oil Louisiana coastal parishes across an array of specifications,
while Alabama, Mississippi, and all of Gulf Coast Florida experienced statisti-
cally significant declines in the count of establishments. Third, I investigated
worker migration using IRS tax statistics, but found no statistically significant
impacts of these events on net migration. Fourth, I analyzed parish-specific
sales tax data for Louisiana and found a statistically significant increase of 11%
in sales tax revenues in the Louisiana oil parishes. Fifth, I investigated quarterly
air travel passenger arrivals by airport in the Gulf Coast states, which provides
some evidence of lower air passenger arrivals in Florida Gulf Coast airports than
other airports during this time. I also provide evidence of the magnitude of the
spill response, the positive impact of initial compensation claims on employ-
ment, and the very small take-up of benefits available through the rig worker
assistance fund to further illustrate the net labor market impacts of these events.
The next section synthesizes the relevant literature on local labor market
shocks and describes the predicted labor market impacts reported in the media
during the spill and moratorium in 2010. The third section outlines the em-
pirical strategy. Section four presents the empirical results for the base models
for employment and wages, followed by an extensive array of robustness and
external validity checks. The final section concludes.
2 Predicted Labor Market Impacts of the Spill
and Drilling Moratorium
2.1 Local Labor Market Shocks
The local labor markets literature highlights the expected outcomes of an ad-
verse shock to labor demand — such as the shutting down of a fishery due to
an oil spill or a drilling moratorium. A negative demand shock should decrease
wages and increase unemployment and, with time, result in out-migration of
workers, who seek better wages elsewhere (Blanchard et al.|/1992). The impact
on firms could be mixed, with the initial demand shock causing some firm exit,
but the wage readjustment process leading to in-migration of firms that seek
out low-cost labor and eventually mitigate some of the impact of the negative
shock. Due to the dynamics of migrating workers and firms, finds
that a transitory shock is likely to have a more pronounced impact on wages
than a permanent shock.
A positive demand shock - such as a large spill clean-up effort — should in-
crease wages, reduce unemployment, and result in net in-migration of workers.
The positive shock to economic activity could result in short-run firm entry,
although higher wages could discourage long-run entry. In light of worker mi-
gration, the increase in employment under a positive labor demand shock could
reflect more jobs for residents and/or more jobs for migrants|"|/Enrico| (2011)’s
assessment of the local labor market literature notes that empirical evidence on
this question is mixed.
While the theoretical and empirical literature have been motivated by and
focused on a wide array of local labor market shocks, of particular relevance
to the analysis in this paper is the impact of Hurricane Katrina, one of the
most destructive and deadly hurricanes to strike the Gulf Coast, on the region’s
labor markets. In the months following the storm, payroll employment fell by
more than one-third in New Orleans (Groen and Polivka} |2008). The number of
business establishments in Orleans Parish fell nearly 20% in the two years after
the hurricane 2008). Yet, in contrast to the oil spill, which primarily
shocked demand, Katrina adversely impacted labor supply and demand.
(2008) notes that the higher wages and relatively low unemployment a year after
the storm suggests that the reduction in labor supply dominated the decline
‘During the 2010 oil spill, some Gulf Coast political leaders complained that spill response
jobs went to non-residents.
in labor demand. In analysis of individual income tax return data,
find that the gap in wage earnings between Katrina victims and the
control group had closed by 2007, and non-employment differences had likewise
closed by 2009. In light of these findings, in the regression models that control
for seasonality presented below, I focus on panels over the 2008-2010 period (as
opposed to longer panels) to minimize “Katrina effects” in my statistical models.
Since theory cannot unambiguously resolve the net effect of simultaneous
positive and negative demand shocks in a local labor market, I focus on an
empirical analysis of the impact of the spill, spill response, and moratorium
on the Gulf Coast labor markets. If the adverse shocks (spill and moratorium)
dominate the positive shock (spill response), then I would expect a decline in em-
ployment and wages, as well as out-migration and, in the short run, fewer busi-
ness establishments. Before turning to the empirical framework and analysis, I
present some evidence of the potential scope of adverse labor market impacts
predicted in spring and summer 2010 during the spill and drilling moratorium.
2.2 Predicted Labor Market Impacts of the Spill
In the weeks after the Deepwater Horizon sank to the bottom of the Gulf of Mex-
ico, analysts and politicians began to predict the potential employment impacts
of the spill. In May 2010, the Atlanta Federal Reserve Bank identified about
130,000 jobs at risk from the spill in the forestry /fishing, arts /entertainment /recreation,
and accommodation and food services industries in the Gulf States
(2010). An economist at the University of Central Florida estimated
that 39,000 (195,000) jobs could be lost in Florida if the spill caused a 10%
(50%) decline in tourism (Harrington| |2010). The spill adversely impacted lo-
cal employment through the closing of state and federal Gulf fisheries and by
discouraging some tourists from vacationing on the Gulf Coast that summer
(Slay 2071).
Some early evidence suggested that regional tourism, especially in Florida,
would bear adverse impacts from the spill. |Oxford Economics] (2010) presented
information reflecting tourist travel intentions to the Gulf Coast in summer
2010. First, they reported on consumer webpage views for TripAdvisor, which
is the world’s most popular travel website. Gulf coast destinations of Clearwater,
Destin, Gulf Shores, Fort Myers Beach, Key Largo, Panama City Beach, and
Pensacola each had changes in TripAdvisor page views in June 2010 relative to
June 2009 ranging between -25% and -52%, while Atlantic coast destinations,
such as Daytona Beach, Hilton Head, Miami, Myrtle Beach, and West Palm
Beach experienced changes in page views ranging between -4 and +17%. Oxford
Economics also reported that a June 2010 survey found that 10% of households
intending to travel to the Gulf Coast for vacation had changed their plans as a
result of the oil spill. (Oxford Economics] (2010) estimated a reduction in tourism
revenues by 12% in the first. year after the beginning of the spill.
2.3. Predicted Labor Market Impacts of the Drilling Mora-
torium
Within a week of the U.S. government’s announcement of the May 27 drilling
moratorium, representatives of the oil and gas industry highlighted potentially
large employment losses (see Table {1). The Louisiana Mid-Continent Oil and
Gas Association identified potential job losses in excess of 30,000
[Workforce Commission] |2010). In June, several economists at Louisiana State
University independently estimated moratorium-related job losses in the range
of 10,000 — 20,000 for the state of Louisiana (Dismukes| |2010} |Mason| |2010}
[Richardson| |2010). The U.S. government also employed regional multiplier
models in two analyses: an internal Department of Interior assessment in July
reportedly estimated job losses in excess of 23,000 (Power and Eaton} |2010)
while an interagency working group report published in September estimated
job losses in the 8,000 — 12,000 range (U.S. Department of Commerce} |2010).
Senator Mary Landrieu of Louisiana stated that the moratorium “could cost
more jobs than the spill itself” 2010). John Hofmeister, the former
CEO of Shell, stated that “50,000 people could lose their jobs” (2010).
The head of one Florida-based investment firm wrote in his oil spill blog that
“an extended moratorium... will cost up to 200,000 higher-paying jobs in the
oil drilling and service business and that the employment multiplier of 4.7 will
put the total job loss at nearly 1 million permanent employment shrinkage over
the next few years” (2010).
The pre-spill employment data can put these moratorium job loss estimates
in context. First, about 9,000 rig workers worked on projects in the Gulf of
Mexico covered by the moratorium (U.S. Department of Commerce} |2010). The
total number of workers — onshore and offshore — in the oil and gas industry and
in support services to oil and gas extraction in the Gulf States in April 2010
numbered about 110,000 (BLS| |nd).
3 Empirical Strategy
The Deepwater Horizon oil spill and the offshore drilling moratorium were two
unprecedented and unexpected events. With an estimated release of about 5
million barrels of oil over nearly three months, the Deepwater Horizon spill was
some 50 times larger than the second biggest spill in U.S. history, the 1969
Santa Barbara spill (National Commission on the BP Deepwater Horizon Oil)
Spill and Offshore Drilling) |2011). The scope of the Deepwater Horizon spill
&
likewise triggered an unprecedented spill response.
The May 27, 2010 offshore drilling moratorium was also an unexpected event.
In 2009, a bipartisan energy bill passed the Senate Energy and Natural Re-
sources Committee that would have opened up the eastern Gulf of Mexico to
drilling (American Clean Energy Leadership Act, S. 1462). In March 2010,
President Obama announced a new offshore leasing plan that would also make
available more of the Gulf to offshore drilling. Prior to the spill, political mo-
mentum pointed toward more Gulf of Mexico offshore drilling, not the potential
for restricted access and drilling activities. A search of GoogleNews, shows
no media articles calling for a moratorium on drilling throughout the Gulf of
Mexico before May 27, 2010.
The economic impacts of the spill, spill response, and moratorium varied
within and among the Gulf States. Figure 2 illustrates the coastal and inland
counties for each of the five Gulf States as well as the location of the Deepwater
Horizon oil spill. Coastal counties and parishes in this region bore the vast ma-
jority of the effects of these events, while inland areas were largely unaffected.
The moratorium was expected to affect select Louisiana parishes — with signif-
icant support of the offshore drilling industry — but not Florida, which had no
active drilling off of its coastline.
The timing and magnitude of the spill response varied across the states over
the course of the spill as well. The spill began with the Deepwater Horizon
explosion on April 20 and the sinking of the rig on April 22. Soon after the rig
collapsed to the bottom of the Gulf of Mexico, remote robots captured video of
oil leaking from the well. Throughout May, BP undertook an array of efforts
to contain the well without immediate success, and by the end of the month
it appeared very likely that the spill would not be stopped until a relief well —
requiring at least three months — intercepted the leaking well some 17,000 feet
below sea level. On May 27, the Department of the Interior issued a six-month
suspension order for deepwater drilling, commonly referred to as the offshore
drilling moratorium. On July 15, the leak was capped. On September 19, the
relief well officially killed the well. Nearly a month later on October 12, the
Department lifted the drilling moratorium.
These spatial and temporal characteristics of the spill and moratorium mo-
tivate the empirical framework. I use a difference-in-differences strategy to
estimate the impacts of the spill, spill response, and moratorium on employ-
ment and wages. In this framework, the spill, spill response, and moratorium
are considered exogenous events that “treat” coastal counties in the Gulf region
(see Table (2). Given the surprise nature of these events, they can credibly be
characterized as exogenous and hence the pre-event period should not include
anticipatory behavior in the Gulf economy (i.e., relocating fishing vessels to the
Atlantic coast in anticipation of the spill or relocating drilling rigs to another
region in anticipation of the moratorium). Expectations that oil could impact
any part of the Gulf Coast — as evident with the discussion above on online
Florida tourism searches and clearly indicated in contemporaneous media cov-
erage of the spill — suggests that all Gulf state coastal counties and parishes
should be considered “treated” by the spill. Coastline incidence of oil ranged
from Cameron Parish, the western-most coastal parish in Louisiana to Wakulla
County, on the eastern edge of the Florida Panhandle Spill response activities
focused on these coastlines and nearby waters (as well in the immediate vicinity
of the Deepwater Horizon), although some spill response efforts originated in
Texas as well. Very little spill response activities occurred in the Non-Panhandle
Florida coastal counties.
I employ the National Oceanic and Atmospheric Administration’s definition
of Gulf of Mexico coastal counties and parishes for Alabama, Florida, Louisiana,
Mississippi, and Texas (National Oceanic and Atmospheric Administration] |nd).
These counties are assumed to be “treated” by the spill for May through July
2010. In addition, I isolate the five Louisiana parishes identified by the U.S.
government as most active in support of offshore drilling activities — Iberia,
Lafayette, Lafourche, St. Mary’s, and Terrebonne (Interagency Economic Re-
port 2010) | These are assumed to be treated by the moratorium for June
2 Refer to the Environmental Response Management Applica-
tion (ERMA) Deepwater Gulf Response website for geographic
data on the oil spill: http://gomex.erma.noaa.gov/erma.html# /x=-
89.37870&y=29.14486&z=6 &layers=23036+5723+23566+244024272544155+2174, last
accessed Augsut 3, 2014.
’Dismukes (2010) estimates that nearly three-quarters of the economic impacts of the
moratorium would be borne by these five parishes in his multiplier analysis for Louisiana.
through October 20104] For the “control” group, I employ all non-coastal coun-
ties and parishes in these five states. Table [3] presents summary statistics on
the number of coastal and inland counties by Gulf State as well as average
employment, weekly wage, and establishment counts.
Formally, I specify the following regression equations:
In(yit) = a; + 6, + C1 [spill], [coastal]; + ex (1)
In(yit) = a; + 64 + B1[spill],1[nonoil — coast]; + y1[mor]:1 [oil]; + ei% (2)
6
In(yit) = O% +5:+)— E1[spill],1[nonoil — coast]; 1[region]; +y1[mor];1 [oil]; +e:
j=l
(3)
in which y represents one of two labor market outcomes: (a) total employ-
ment for county i in month t, and (b) average weekly wage in 2010 dollars for
county 7 in quarter tp a@ represents county fixed effects, 6 represents month-
year (quarter-year) fixed effects in the employment (wage) specifications; the
indicator functions 1[spill] and 1[mor] take the value of 1 for the months May
through July (oil spill duration) and for the months June through October
(moratorium duration), respectively {| the indicator functions 1[coastal], 1[oil],
1[nonoil —coast] take the value of 1 for NOAA-classified Gulf Coast counties and
parishes, the five Louisiana “oil” parishes of Iberia, Lafayette, Lafourche, Saint
Marys, and Terrebonne, and all coastal counties and parishes except for the five
oil parishes, respectively; and the indicator function 1[region] takes the value
of 1 for each of these six regions: Alabama, Panhandle Florida, Non-Panhandle
Florida, Louisiana, Mississippi, and Texas. I estimate the base models with
data for January through October 2010 (I test for sensitivity to panel length
and seasonal controls in the robustness checks). The standard errors are clus-
4The appendix table lists Gulf Coast counties within each treatment category.
5T also employ the quarterly-based specification for establishment count analyses in the
robustness checks below.
®For the wage and establishment count analyses, these indicators take the value of 1 for
2010 quarter 3. The spill began in week four of the second quarter, and the moratorium
began in week nine of the second quarter. Given this timing and the likely lag for wage
adjustment, I specify the wage and establishment count models such that the net treatment
of spill, response, and moratorium are assumed to occur in 2010q3.
10
tered by county/parish to account for potential serial correlation in the data
(Bertrand et. al.| |2004).
To implement this model, I employ QCEW monthly county/parish-level em-
ployment and quarterly county /parish-level wage data (BLS}{nd). The BLS and
state employment security agencies compile monthly employment and quarterly
wage and establishment data for workers covered by various unemployment in-
surance programs, and hence is effectively a measure of employment provided
by employers. It does not include data on the self-employed, proprietors, armed
forces, domestic workers, and railroad workers. This dataset only permits an
investigation of employment levels; it does not include data on labor force par-
ticipation, unemployment, or unemployment rates "| In addition, the QCEW
provides employment and wage data, for some counties and parishes in this re-
gion, by sector and industry, and I use these data to investigate industry-specific
labor market impacts in the robustness checks.
4 Results
4.1 2010 Panel Analyses of Employment and Wages
Table [4] presents the employment results for estimating equations Equa-
tion |1| which permits an examination of a common treatment of the spill, spill
response, and moratorium on all coastal counties and parishes, shows fairly pre-
cise zero impacts for the Gulf Coast counties. The estimated coefficient cannot
be distinguished from zero and the 95% confidence interval ranges from about
a 7/10 of 1% decline to a 1/10 of 1% increase in employment.
The model in column 2 allows for differential impacts for the five oil-intensive
parishes of Louisiana and for the rest of the Gulf Coast counties. This model
effectively focuses on the treatment of spill, spill response, and moratorium on
the oil-intensive parishes and the treatment of spill and spill response on the
rest of the Gulf Coast region, which had little economic activity that could
be impacted by the offshore drilling moratorium. Not surprisingly, the non-oil
Gulf Coast counties show similar impacts as the entire Gulf Coast in column 1
results. The oil parishes exhibit a statistically significant employment increase
of 1.2%.
“While the Current Population Survey’s Local Area Unemployment Statistics (LAUS)
provides such information at the county-by-month level, the LAUS data are imputed for most
counties and months, in contrast to the QCEW count data.
11
The model in column 3 provides even greater flexibility in estimating the im-
pacts of treatment by the spill, spill response, and moratorium by allowing for
effects to vary by state and sub-state region, including the Louisiana oil parishes.
The results in column 3 illustrate a statistically significant 1.2% employment
gain in the Louisiana oil parishes, a statistically significant 1.3% employment
gain in Alabama coastal counties, and a statistically significant 2.7% employ-
ment decline in Non-Panhandle Florida Gulf Coast counties. Based on average
2009 employment levels in these three regions, these statistically significant es-
timated impacts translate into about 3,000 more jobs in the oil parishes, 3,000
more jobs in the Alabama coastal counties, and 50,000 fewer jobs in the Florida
coastal counties reflecting the net effects of the spill, spill response, and mora-
torium.
Table [4] columns 4 — 6, present the results for impacts of these events on the
average weekly wage. As evident above in the discussion of employment, the
average treatment effect for the Gulf Coast counties as a whole is not statistically
different from zero (the 95% confidence interval is about +/- 1 %) (column
4). The Louisiana oil parishes experience a statistically significant increase in
wages of about 2% (columns 5 and 6). Likewise, the Alabama coastal counties
appear to have higher wages, with a statistically significant estimate of about
4% (column 6). The non-oil Louisiana coastal parishes experience a statistically
significant 2.5% decline in wages. Both Florida regions, Texas, and Mississippi
experience small and statistically insignificant wage impacts during the oil spill.
Given the evident variation in employment and wage outcomes by state and
sub-state region, I focus on model (3) in the subsequent robustness checks.
4.2 Robustness: Seasonality
To address the concern that the results presented above reflect regular seasonal
variations in labor markets, I expanded the analyses to include a longer panel
(2008-2010) and allowed for month (quarter) fixed effects to vary between coastal
and non-coastal regions as well as by state. Table [5] presents the employment
impacts for these seasonality robustness checks and includes the model from
column 3 in Table 4 for reference.
While the oil parishes exhibit a statistically significant employment increase
of 1 — 2% in the 2010 panels (columns 1, 3, and 5), the longer panels show
fairly precise zeroes for the oil parishes (columns 2, 4, and 6). These results
highlight a question for this kind of analysis (and is common to all reduced-
12
form difference-in-differences empirical models) — when should the panel start?
The panel could start as early as January 1990 (given the construction of the
public domain QCEW dataset) or as late as April 2010. A longer panel pro-
vides more power to estimate parameters, but if omitted time-varying factors
affect a subset of counties or parishes, then lengthening the panel could affect
the estimation of county/parish fixed effects and the treatment effects. For ex-
ample, Hurricane Katrina and post-hurricane rebuilding differentially impacted
counties and parishes in 2005 and subsequent years. In addition, the increase
in oil prices over 2003-2008, including the sharp run-up in spring and summer
of 2008, could have spurred greater drilling activity and associated support ac-
tivity employment in the oil parishes than in 2010. A longer panel, however,
permits the estimation of seasonal trends in employment that may explain some
of the variation evident in a 2010-only panel. I report panel results for 2008-
2010 — long enough to control for seasonality but post-Katrina and after the
local labor market had begun to converge to a new normal, at least as described
in the literature reviewed above in section 2B]
Alabama coastal counties enjoy a statistically significant employment in-
crease of 1.3 to 2.0% in the 2010 panel (columns 1, 3, and 5), but small, positive,
statistically insignificant effects in the 2008-2010 panel. Non-Panhandle Florida
Gulf Coast counties experience a statistically significant employment decline of
1.7 — 2.7% in four of the six specifications presented in Table |5| (columns 1,
3, 5, and 6). With longer panels, Texas coastal counties appear to experience
statistically significant employment gains ranging between 1.5 and 2.0% for the
2008-2010 panel.
Table [6] presents the seasonality robustness checks for the wage impacts. In
five of the six specifications, Louisiana oil parish wages are statistically signif-
icant and higher than control wages by 2 — 3%. Likewise, in five of the six
specifications, Alabama coastal county wages are about 4 — 6% higher and sta-
tistically significant. In 2008-2010 panels, Texas coastal county wages are 1.4
— 1.9% higher. The statistically significant decline in non-oil Louisiana coastal
parish wages only holds for two models with the 2010 panel. The two Florida
regions and Mississippi experience small and statistically insignificant wage im-
pacts during the oil spill.
81 have also estimated these models with 2007-2010 and 2009-2010 panels, which yield very
similar results to the 2008-2010 panel models.
13
4.3. Robustness: National Controls
The primary empirical strategy in this paper rests on the assumption that the
non-coastal counties and parishes in the Gulf Coast region are not impacted by
the spill, spill response, or moratorium and thus can serve as controls in the
regressions. Given the close proximity many of these counties and parishes have
to the treated coastal region, there may be a risk that the controls are affected by
changes in economic activity in the treatment region. For example, if a worker
lost her job in a treatment county and relocated to a control county where she
took on a new job, then we would have a case in which the employment status of
the individual is unchanged but this empirical framework would estimate a delta
of -2 for employment (for treatment employment minus control employment
during the treatment period). In a similar way, if a worker quits a job in a
control county so he can move to a treatment county and participate in spill
response, then this approach would again show a change in employment (in this
case a delta of +2) despite the fact that the employment status for the worker
in question is unchanged.
To address this possible concern, I run the base regressions with a modified
sample. I exclude all control counties and parishes in the Gulf States and I
add all non-Gulf State counties in the United States. Thus, I use the non-Gulf
Coast region as the control group for the treatment of the spill and moratorium
on the Gulf Coast counties and parishes. This also permits an illustration of
how employment trends in the Gulf Coast region compared with the rest of the
nation during these events.
Table [7| columns 1 and 2, shows the results for employing national controls
for employment outcomes. In both panels, the Florida coastal regions have a
statistically significant lower employment of 2 to 4%. Neither the Louisiana oil
parishes nor the Alabama coastal counties show employment impacts statisti-
cally different from zero in each of the panels. While the non-oil Louisiana oil
parishes and coastal counties of Mississippi and Texas appear to have statisti-
cally lower employment in the 2010 panel on the order of 1 to 1.5%, these results
do not hold up in the 2008-2010 panel.
Table |8| columns 1 and 2, presents the results for wage impacts using the
national sample. The Louisiana oil parishes have statistically significant higher
wages ranging between 2.5 — 3.2%. Likewise, Alabama has statistically higher
wages in the 5 — 6% range. There is some evidence of statistically higher wages
in Texas — up to 2% — in the 2008-2010 panel. None of the other regions have
14
wage impacts statistically different from zero.
4.4 Robustness: Omit Texas
Texas counties represent about 48% of the sample in the Gulf State statistical
analyses. Texas may not have experienced much of an impact from the oil spill,
since the spill occurred southeast of the boot of Louisiana and the vast. majority
of the oil moved to the east and north from the leaking well. To address the
concern that the Texas counties — coastal and inland — may not be appropriate
for inclusion in these statistical analyses, I have run the base regressions with a
four-state Gulf sample that omits coastal and inland counties of Texas.
Table |7| columns 3 and 4, shows the results for the model runs with this
modified sample. As before, the Louisiana oil parishes and the Alabama coastal
counties have statistically higher employment in the 2010 panel, but not in
the 2008-2010 panel. The non-oil Louisiana parishes have statistically higher
employment, ranging from about 0.7% to 2.2% in the 2010 and 2008-2010 pan-
els, respectively. The Non-Panhandle Florida coastal counties have statistically
lower employment of about 2.3% in the 2010 panel, although this effect falls and
is not significant in the 2008-2010 panel. Overall, these results are fairly similar
to those specifications that include Texas counties in the regression model.
Excluding the Texas observations does not qualitatively alter the estimated
wage impacts of the spill, response, and moratorium (Table|g} columns 3 and 4).
Louisiana oil parish wages experience a statistically significant 2.2% increase,
and the Alabama counties enjoy statistically significant increases of 4 — 6%
across these panels as well. There is limited evidence of a statistically significant
decline in the non-oil Louisiana coastal parish wages, as illustrated in the 2010
panel. None of the other state-specific coastal wage impacts are statistically
different from zero in this “Omit Texas” framework.
4.5 Robustness: Definition of Coastal
The base regressions employ the NOAA definition of a coastal county or parish
in the Gulf States. I have also employed a more narrow definition that requires
a county or parish to meaningfully border the Gulf of Mexico. In these analyses,
I have dropped what I now define as “buffer” counties and parishes — those that
NOAA identifies as coastal but do not have meaningful coastline. This reduces
the sample from 534 to 444 counties and it also eliminates those counties that
may have experienced relatively weak “treatment,” when compared to those
15
counties with significant coastline, and relatively stronger “treatment,” when
compared to those counties further inland.
The Louisiana parishes — oil-intensive and non-oil — have statistically higher
employment in the 2010 panel but not in the 2008-2010 panel (Table{7| columns
5 and 6). Alabama coastal counties (in this case, Baldwin and Mobile) have
statistically higher employment of 1.9% to 2.7% across the panels, while Non-
Panhandle Florida coastal counties have statistically lower employment of more
than 3% in the two panels. These results illustrate the robustness of the adverse
employment impacts to Florida and potentially highlight the positive employ-
ment impacts to the two Alabama counties sitting on the Gulf Coast.
For wage effects under this alternative definition of coastal counties and
parishes (Table |8| columns 5 and 6), the oil-intensive Louisiana parishes have
statistically significant higher wages of about 2% in the panels. Coastal counties
in Non-Panhandle Florida appear to experience a statistically significant decline
of about 1.4 to 2.3% in these panels. In addition, Alabama appears to benefit on
the order of 1.5 — 2.0% in the average wage. There is no statistically significant
impact of the spill on non-oil Louisiana coastal parishes in this framework.
4.6 External Validity: Industry-Specific Impacts
To complement the analyses of total employment and wages, I have also es-
timated equation |3} with industry-specific data for the support activities for
mining (NAICS 213, which includes drilling oil and gas wells NAICS 213111)
and accommodation (NAICS 721) industries. These industry-specific analy-
ses permit further investigation of the impacts of the spill, spill response, and
moratorium on directly affected industries. While additional industries, such as
fishing (NAICS 1141), water transportation (NAICS 483), oil and gas drilling
(NAICS 211) would certainly be of interest for this kind of analysis, the cen-
soring of the public domain county-level employment data renders these panels
much too small (3, 20, and 62 counties, respectively, out of 534 in the region).
Table [9] presents the estimated employment and wage impacts for the support
activities for mining and accommodation industries. As a result of data cen-
soring, the models for support activities for mining industry exclude Florida
counties.
For the mining support industry, the Louisiana oil parishes, Alabama coastal
counties, and Mississippi coastal counties appear to experience a statistically
significant decline in employment based on the 2010 panel, but this is robust
16
to extending the panel to 2008-2010 only for Alabama, which suffers an 11%
decline. In contrast, non-oil Louisiana parishes experience a 14% increase in
mining support employment in the longer panel. Louisiana oil parishes and
Alabama coastal counties also experience a decline in the wage in the mining
support industry based on the 2010 panel, although this effect falls to a statis-
tical zero for each region in the longer panel. Mississippi coastal counties have
a statistically significant 1.6-3.3% increase in mining support wages in the two
panels.
For the accommodation industry, the Louisiana oil parishes have a statisti-
cally significant increase of nearly 5% in employment, while the non-Panhandle
Florida counties (-4.7%), Mississippi coastal counties (-4.8%), and Texas coastal
counties (-2.7%) experience statistically significant declines in employment in
the 2010 panel. These results hold only for the non-Panhandle Florida counties
in the 2008-2010 panel, which shows a 3.7% decline in accommodation employ-
ment in these counties. For this industry, again, only the non-Panhandle Florida
counties show a statistically significant decline in wages, ranging from -3.6 to
-6.2% in the two panels, although there is weak evidence that the Louisiana oil
parishes enjoyed a 4.2 to 4.7% increase in the accommodation industry wage
(statistically significant at the 10% level in each panel).
4.7 External Validity: Establishment Count
The QCEW provides quarterly counts on the number of business establishments
in each county. I employ equation (3) and replicate the specifications as reported
in Table|6]on wages with the natural logarithm of the establishment count as the
dependent variable. Table [10] presents the results for these specifications with
quarter, quarter-by-coastal, and quarter-by-state fixed effects (in addition to
county fixed effects) for 2010 and 2008-2010 panels. As in the wage regressions,
the third quarter of 2010 is considered the “treated” quarter in these statistical
analyses.
In all specifications, non-oil Louisiana coastal parishes experience statisti-
cally significant increases in the establishment count, ranging from about 0.5 to
4%. In all but one specification, the Louisiana oil parishes likewise experience
statistically significant increases of about 1 to 4%. In all specifications, Pan-
handle Florida and Non-Panhandle Florida experience statistically significant
declines in establishment count of about 1 to 2% in the former and about 2
to 4% in the latter. In a majority of specifications, Alabama and Mississippi
17
coastal counties also experience statistically significant declines in establishment
count, 3 and 1%, respectively.
The net positive impacts of these events on Louisiana business establish-
ments is consistent with the zero to positive impacts on employment and wages
for the two Louisiana regions in nearly all specifications. Likewise, the adverse
impact on Non-Panhandle Florida coastal business establishments squares with
the finding of a decline in employment across virtually all statistical models.
4.8 External Validity: Migration
The standard models of local labor markets suggest that a negative (posi-
tive) shock that decreases (increases) wages will result in out-migration (in-
migration). To investigate the potential impacts of the spill, spill response, and
drilling moratorium on migration, I use annual migration data provided by the
|U.S. Internal Revenue Service| to estimate a version of equation [3] In par-
ticular, I estimate this regression model with an array of migration dependent
variables: net migration (scaled by the non-migration population), the natural
logarithm of in-migration, and the natural logarithm of out-migration. Net mi-
gration reflects the net effect of total in-county migration and total out-county
migration for a given year. Given the annual nature of the data, I consider
the 2010 data for the various coastal counties and parishes as treated, and run
the models with 2009-2010, 2008-2010, and 2007-2010 panels. I also run these
models with various assumptions about fixed effects: using year, year-by-state,
and year-by-coastal fixed effects (and all with county fixed effects) in various
model runs.
The net migration models consistently show — across panel lengths and var-
ious assumptions over fixed effects — no statistically significant impact of the
spill, spill response, and moratorium on net migration flows in coastal counties
and parishes in the Gulf States. For the in-migration and out-migration models,
there is no evidence that any region, in any specification, experienced a statisti-
cally significant increase in either in-migration or out-migration. There is some
evidence that Louisiana oil parishes and non-oil Louisiana parishes experienced
statistically significant declines in in-migration and out-migration. In almost
all specifications, the Florida Panhandle and Non-Panhandle Florida counties
experienced statistically significant declines in in-migration and out-migration,
with the Non-Panhandle counties having the largest declines (as much as -9%)
of any region. The reductions in both in- and out-migration may reflect un-
18
certainty over the duration and persistence of these events on the local labor
markets.
4.9 External Validity: Sales Tax Revenue
To further corroborate the labor market findings for Louisiana and Florida,
I separately analyze these states’ parish-/county-level sales tax data? The
State of Louisiana reports sales tax revenues by parish for the state general
sales tax on an annual basis for fiscal years that run from July to June. The
state sales tax rate was 4% over the sample period and across all parishes |]
The revenue data are for fiscal years ending in June of a given calendar year
(Louisiana Department of Revenue} |2010) |2011). For this analysis, I conduct
a standard difference-in-differences analysis that allows for comparisons across
two time periods: (1) July 2009 — June 2010, and (2) July 2010 — June 2011; and
comparisons across three groups: (1) non-oil Louisiana coastal parishes, (2) oil
parishes, and (3) inland (control) parishes. Specifically, I estimate the following
regression:
In(reviz) = a; +6¢+21[2010/2011],1[nonoil — coast]; +71[2010/2011], 1 [oil]; +e,
(4)
in which rev represents total sales tax revenue for parish 7 in year t; a rep-
resents county fixed effects; d represents the fixed effect for the 2010/2011 year
(the 2009/2010 year fixed effect is omitted); the indicator function 1[2010/2011]
takes the value of 1 for the 2010/2011 tax revenue year; and the other indicator
functions were defined above.
The oil parishes experienced an increase in sales tax revenues for the July
2010 — June 2011 period relative to the previous twelve months. This increase
in tax revenues differs statistically from the non-oil coastal parishes and the
inland parishes, each of which experienced no meaningful change in the level
of revenues|! | The difference-in-differences estimator using parish-level data to
®This analysis focuses on Louisiana and Florida because of the distinctive impacts of the
spill, response, and moratorium on these two states and the availability of public use sales tax
data.
10 Technically, the 4% state general sales tax consists of a 3.97% state sales tax and a 0.03%
Louisiana Tourism Promotion District sales tax. The revenue data presented covers only the
state sales tax; it does not include parish or city-established sales tax revenues.
114 Wald test of the hypothesis that @ = 4 is rejected at the 1% level, and 4 is statistically
significant at the 1% level.
19
compare the before and after time periods yields a statistically significant 11%
increase in sales tax revenue for the oil parishes relative to the inland parishes.
The State of Florida reports county sales tax data on a monthly basis|"?] I
use gross sales subject to the state sales tax in a specification similar to (4):
In(salesiz) = a; + 6: + B1[spill],1[pan]; + y1[spill],l[nonpan]; + eit (5)
in which sales represents gross sales subject to the state sales tax for county
i in month t; @ represents county fixed effects; 6 represents month and year
fixed effects; and the other indicator functions were defined above. Given the
seasonal nature of Florida tourism, I estimate equation (5) with a 2010 sample
(January to July) and January 2009 — July 2010 and January 2008 — July
2010 samples. The Panhandle counties experienced a 12-15% increase in gross
sales relative to the non-Gulf Florida counties during the spill months, and this
impact is statistically different from zero at the 1% level in all three sample
periods. In contrast, Non-Panhandle Gulf coast counties experienced relatively
small and statistically insignificant sales increases (1.5-2.6%) during the spill.
Wald tests of the hypothesis that the coefficient estimates on the Panhandle and
Non-Panhandle indicator variables are equal are rejected at the 1% significance
level in all three sample periods.
4.10 External Validity: Commercial Air Travel
The labor market impacts presented above suggest that tourism may have been
adversely impacted by the oil spill, especially in Florida. To further investigate
these impacts, I employ data compiled by the |U.S. Bureau of Transportation|
Statistics] (nd), which tracks commercial air travel, including number of pas-
senger arrivals by airport by quarter. According to the BTS, there were 115
airports in the five Gulf States receiving commercial air passengers in the sec-
ond quarter of 2010. For this analysis, I limit the sample of airports to those
with at least 50,000 passenger arrivals per quarter, which corresponds to about
four daily arrivals of a Boeing 737-sized aircraft. This restriction drops a num-
ber of military installations that occasionally receive civilian arrivals and very
small airports. The resulting sample includes 32 airports that received more
127 accessed Florida county-by-month sales tax data from
http://dor.myflorida.com/dor/taxes/colls from 7 2003.html on August 12, 2012.
20
than 97% of all commercial air passengers in these five states in the second
quarter of 2010. Coastal counties and parishes host 15 of these airports.
To evaluate the potential impacts of the oil spill on commercial air travel, I
estimate the following regressions:
In(passiz) + a4 + 6, + B1[spill],1 [coastal]; + ei, (6)
In(pass;z) +a; + 6+ 61 [spill], 1[F Lcoast]; + y1[spill],1[nonF Lcoast]; + i (7)
in which pass represents the total number of arrivals for airport 7 in quarter
t; a represents airport fixed effects; 6 represents quarter-year fixed effects; the
indicator function 1[spill] takes the value of 1 for the third quarter of 2010; the
indicator functions 1[{coastal], 1[F Lcoast], and 1{nonF'Lcoast] take the value
of 1 for airports in NOAA-classified Gulf Coast counties and parishes, airports
in Florida coastal counties, and airports in non-Florida Gulf Coast counties,
respectively. I estimate the models with varying lengths of panels, ranging from
2008:Q1 through 2010:Q3 to only the first three quarters of the 2010 calendar
year. The standard errors are clustered by airport.
For the specifications of equation there is a modest but statistically
insignificant reduction (-6%) in commercial air travel passengers for the oil
spill quarter. By estimating differential impacts for Florida versus non-Florida
coastal airports (equation (6), there is some weak evidence that Florida airports
are adversely impacted by the oil spill. With the 2010 panel, I estimate a
17% decline in Florida coastal airport passenger arrivals, although this estimate
cannot be statistically distinguished from zero. Using the 2007-2010, panel I
estimate a similar 15% decline that is statistically different from zero at the
10% level.
4.11 External Validity: Spill Response Labor Mobiliza-
tion
The unprecedented mobilization of spill response resources — including more
than 800 specialized skimmers, 120 aircraft, 8,000 vessels, and nearly 50,000 re-
sponders (Aldy\|2011) — provided employment opportunities that could counter
the potential adverse effects of the spill and the moratorium. While many
of these responders represented workers relocating temporarily to address the
21
spill, some were local displaced workers. For example, fishermen who faced
closed state and federal fisheries during the spill could participate in the Ves-
sels of Opportunity program. Through this program BP paid $1,200 - $3,000
per day per vessel for skimming, booming, and related response operations
{tional Commission on the BP Deepwater Horizon Oil Spill and Offshore Drilling}
2011). Between April and July 2010, the U.S. Coast Guard expended nearly
$600 million on spill response, with more than $100 million each for personnel
and cutters as well as more than $250 million for other federal, state, and local
government clean-up efforts (National Pollution Funds Center| 2013) [*9]
4.12 External Validity: BP Clean-up Expenditures and
Economic Damage Compensation
The clean-up activities and compensation for economic damages provided by BP
could counter at least. some of the impacts of lost income on economic activity.
BP’s expenditures in the Gulf States for damage compensation and clean-up
were quite significant. By June 1, 2010, BP reported spending nearly a billion
dollars for clean-up, and the clean-up tab increased to more than $3 billion by
July 5. On September 17, BP reported clean-up spending of nearly $10 billion.
BP reported compensation for damage claims of $40 million through June 1,
$147 million through July 5, and nearly $400 million through August 23, when
BP turned over the claims process to the independent Gulf Coast Claims Facility
(GCCF) created as a part of the June 16 agreement with the White House
(Aldy| |2011). On September 21, 2010, the GCCF reported paying out nearly
$300 million in its first month in operation, with 36% and 29% of the funds to
Louisiana and Florida claimants, respectively (2010). Fishing industry
claims represented about $120 million followed by nearly $75 million in claims
from food, beverage, and lodging industries. By May 2012, the GCCF had paid
out about $6.5 billion in claims, with the food, beverage, and lodging industry
claims in excess of $1 billion representing the largest industry category. By
then, Florida’s claims amounted to a 38% share of all fund payments, followed
by Louisiana’s share of 28% (GCCF| (2012).
The compensation payments for oil-spill related damages are similar to some
forms of targeted fiscal stimulus. Drawing from |Feyrer and Sacerdote} (2011),
I estimate the relationship between compensation claims and the change in
13The Coast Guard sought and received compensation for these clean-up and spill response
expenditures from BP, the responsible party as specified under the Oil Pollution Act.
22
employment using county-specific claims data from the September 2010 GCCF
report:
A (ean -~atB (erat) 40; Fe; (8)
population / , population /,
in which the dependent variable is the difference in the ratio of employment
to total county population over September 2009 — September 2010 (using a
12-month difference to control for seasonality), the key variable of interest is
the ratio of compensation claims to total county population in September 2010,
and @ represents state fixed effects to control for state-specific drivers of em-
ployment. With compensation per capita measured in $100,000 per person, my
estimated § of 2.93 (with a robust standard error of 0.42) suggests an implicit
fiscal multiplier of about $34,000 per job, which is relatively small given the
fiscal multiplier literature (see |Council of Economic Advisers} |2014). This ap-
proach suffers from an obvious endogeneity problem: compensation claims are
larger in counties suffering greater economic damages from the spill |] Such an
endogeneity problem, however, should bias the coefficient estimate down and
increase the dollars per job implicit multiplier. My model likely suffers from
another source of endogeneity: counties suffering significant economic damages
from the spill also experienced a significant influx of spill response resources and
activities, as detailed above. While this model is not statistically identified, the
apparent bias is consistent with various factors increasing employment in some
of the worst hit parts of the spill, especially in Louisiana and Alabamal[!*|
4.13 External Validity: Rig Worker Assistance Fund
On June 16, 2010, as a part of a larger agreement, the White House and BP
agreed that BP would set aside $100 million for a Rig Workers Assistance Fund
(RWAF) to benefit any rig workers laid off as a result of the drilling moratorium.
BP allocated the $100 million to the Baton Rouge Area Foundation, which was
responsible for designing and implementing the assistance fund. The RWAF
made grants, ranging from $3,000 to $30,000, available to individuals who had
14Feyrer and Sacerdote employ political instruments based on the seniority of Congressional
delegations in their stimulus analyses. There is no obvious analog to the oil spill.
15The average county-level per capita compensation payments are comparable across the
Alabama coast, Florida panhandle, and Louisiana coast (oil and non-oil parishes). Payments
are about a factor of three lower in the Mississippi coast, and an order of magnitude lower
in the Non-Panhandle Florida gulf coast counties and the Texas coast. Inland compensation
payments are, on average, three orders of magnitude smaller.
23
been working on deepwater drilling rigs as of May 6, 2010 and had lost their
jobs. The RWAF took a first round of applications in September 2010 and paid
out grants totaling $5.6 million to 347 workers. To put these claims for lost jobs
in context, 347 rig workers would staff less than two deepwater drilling rigs like
the Deepwater Horizon.
The RWAF opened a second round in the Spring of 2011 to provide grants
to those individuals who lost their jobs in a position that supported deepwater
drilling rigs, such as shipyard workers, caterers, drilling support operations, etc.
In this second round, the RWAF paid out about $5.8 million to 408 workers.
These relatively modest claims for displaced rig support workers are consistent
with the evidence that few rigs left the Gulf of Mexico: only five of the 46
rigs operating on April 20, 2010 had left the region as of September 10, 2010
(U.S. Department. of Commerce} |2010). Anecdotal evidence indicates that “oil
companies used the enforced suspension to service and upgrade their drilling
equipment, keeping shipyards and service companies busy” (Broder and Krauss}
Boro).
5 Conclusions
The April 20, 2010 explosion on the Deepwater Horizon drilling rig precipi-
tated several economic shocks to the Gulf Coast region: an unprecedented U.S.
oil spill, an unprecedented mobilization of spill response resources, and an un-
precedented moratorium on deepwater drilling. This paper has investigated the
impact of these shocks on employment and wages in the Gulf Coast region.
Non-Panhandle Florida gulf coast employment fell during the oil spill, as
evident across panels and various robustness checks. The analysis for the ac-
commodations industry provides additional evidence of the adverse impact of
the spill on Non-Panhandle Florida coastal employment in a tourism-oriented
industry. The evaluation of commercial air passenger arrivals also suggests evi-
dence of a decline in tourism in the Florida coast as a result of the spill. A casual
comparison of the Florida Panhandle — treated by the spill and spill response —
to the Non-Panhandle Florida coast — treated by only the spill — could imply
a positive employment impact of 1 to 2% of total employment associated with
spill response.
The analysis of employment shows little adverse impact in Louisiana, the
state closest to the leaking well in the Macondo Prospect of the Gulf of Mexico.
24
The various statistical models illustrate either statistically significant increases
in employment in the oil parishes (2010 panel) and the non-oil coastal parishes
(2008-2010 panel) or small effects (point estimates < |0.5%]|) that are statisti-
cally indistinguishable from zero. Figure [3] presents the estimated employment,
impacts for each coastal region based on the 2010 panel model (Table[5} column
1; left graph) and the 2008-2010 panel model (Table [5] column 2; right graph).
These graphs show the significant adverse impacts in the Non-Panhandle Florida
coastal counties in contrast with the rest of the Gulf Coast region. The Louisiana
oil parishes also enjoyed a statistically significant increase in the average wage
across all panels and virtually all robustness checks, and the oil parishes and
non-oil Louisiana parishes alike experienced a statistically significant increase
in the number of business establishments. Complementary analyses of sales tax
revenues in Louisiana indicate that the oil parishes enjoyed greater levels of
economic activity during the spill, spill response, and moratorium than non-oil
coastal parishes and inland parishes in Louisiana. In aggregate, this evidence
suggests a net positive labor market shock to the oil parishes and a net zero to
positive shock to the non-oil Louisiana coastal parishes during the period of the
spill and moratorium !!>|
The results of these statistical analyses for the Louisiana parishes differ sig-
nificantly from the predictions made with various state and regional multiplier
models employed to assess the impacts of the drilling moratorium. A number
of analysts quickly undertook multiplier analysis of the moratorium after its
announcement. None of these analysts employed their modeling tools to evalu-
ate the employment and economic activity impacts of the spill itself or the spill
response. Figure [4] presents the estimated combined employment impacts for
all Louisiana coastal parishes (oil parishes and the non-oil parishes) based on
the estimated 2010 and 2008-2010 panel models (Table |5} columns 1 and 2).
The net effect of the spill, spill response, and drilling moratorium resulted in
a statistically significant increase of about 6,400 — 20,000 in coastal Louisiana
employment relative to the counterfactual. All of the predicted negative im-
16 4 casual comparison of the oil parishes (treated by spill, spill response, and moratorium)
to the non-oil Louisiana coastal parishes (treated by spill and spill response) would result in
an inconclusive assessment of the incremental impact of the moratorium. In some statistical
models, the comparison would suggest that the drilling moratorium increased jobs and other
models would yield the opposite conclusion. In virtually all wage models, the comparison
would suggest a net positive impact of the moratorium on labor compensation. The more likely
outcome is that the intensity of spill response activity centered more on the oil parishes than
non-oil parishes and the moratorium, as evident in the limited take-up of benefits available
through the Rig Worker Assistance Fund, had little economic impact.
25
pacts on employment from the published 2010 analyses fall outside the 95%
confidence intervals of these models. The estimated employment losses in the
ex ante multiplier models, in comparison to employment gains in the ex post
statistical analysis — and for that matter, simple graphical analysis in Figure [T]-
suggests several shortcomings of the multiplier tools. First, the ceteris paribus
assumption made in the drilling moratorium multiplier analyses did not appro-
priately represent the economic environment in the Gulf Coast region during
the summer of 2010. Everything else was not equal; a significant influx of spill
response resources provided a source of income and employment for at least
some of those displaced by the spill and the moratorium. Second, these multi-
plier models operated as if a significant number of drilling rigs would relocate
to other regions and layoff a significant number of workers. This did not pan
out, perhaps in part resulting from the uncertainty about future government
regulation — including the length of the moratorium — in the Gulf of Mexico.
These results yield several policy implications. First, a significant pulse of
resources in spill response appears to offset much of the adverse impacts of the
spill. This is not a determination that the optimal level of spill response was
pursued, but to simply note that the spill response delivers an array of imme-
diate and longer-term economic and environmental benefits. In other words,
spill response represents a kind of economic stimulus that creates employment
opportunities, not unlike conventional fiscal stimulus. Second, the ambiguity
about the length of the drilling moratorium may have mitigated some of the
adverse impacts of the drilling moratorium. Throughout what was originally
billed as a 6-month moratorium, Department of the Interior officials noted that
it could end early (and in fact, the moratorium ended more than one month
early). This uncertainty may have created an incentive for rig owners to wait,
not unlike how uncertainty associated with an irreversible investment can cre-
ate value in waiting for new information. Third, multiplier analyses that do
not characterize the complexity and temporal attributes of an economic shock
may be uninformative and potentially biased for policy deliberations. To be
fair, multiplier models provide analysts with a tool to conduct ex ante analysis
premised on a few assumptions about the economic environment that is, by
definition, not available through ex post statistical analysis of employment and
wage data. Just as scientific models of the fate of oil spilled from the Macondo
Prospect would deliver misleading predictions if they failed to account for skim-
ming, dispersant applications, deepwater containment, boom deployment, and
other means of mitigating the effects of the spill, economic models that, by as-
26
sumption, fail to account for the economic and employment impacts of response
activities would also produce misleading predictions. Finally, the net positive
labor market impacts in the regions exposed to the most substantial clean-up
activity and the net negative labor market impacts in those regions with the
least clean-up activity illustrate how a rapid, signicant infusion of resources in
response to a labor market shock, such as a natural or human-caused disaster,
can mitigate the shock's adverse impacts to labor income and participation.
27
References
Aldy, J. E. (2011). Real-time economic analysis and policy development during
the bp deepwater horizon oil spill. Vand. L. Rev. 64, 1795-1817.
Bertrand, M., E. Duflo, and §. Mullainathan (2004). How much should we trust
differences-in-differences estimates? Quarterly Journal of Economics 119,
249-275.
Blanchard, O. J., L. F. Katz, R. E. Hall, and B. Eichengreen (1992). Regional
evolutions. Brookings papers on economic activity, 1-75.
BLS (n.d.). Quarterly census of employment and wages. Data, Bureau of Labor
Statistics, Washington, DC. ftp://ftp.bls.gov/pub/special.requests/|
Broder, J. and C. Krauss (2010). Job losses over drilling ban fail to ma-
terialize. The New York Times. [Online; posted 24-August-2010]
\//waw.nytimes .com/2010/08/25/us/25drill.html?_r=0,
Chriszt, M. and M. Hammill (2010). Estimating the oil spill’s impact in the
gulf. Technical report, Atlanta, GA. |http://macroblog.typepad.com/
macroblog/2010/05/estimating-the-oil-spills-impact-in-the-gulf.
html,
Condon, S. (2010). Mary landrieu: Drilling moratorium could cost more jobs
than oil spill. CBS News. [Online; posted 14-July-2014] http://tinyurl.
Council of Economic Advisers (2014, February). The economic impact of the
american recovery and reinvestment act: five years later. Report to congress,
Executive Office of the President.
Deryugina, T., L. Kawano, and S. Levitt (2014, July). The economic impact
of hurricane katrina on its victims: evidence from individual tax returns.
Working paper, University of Illinois.
Desel, J. (2010). Ex-shell ceo: Job loss from gulf oil spill could top 50,000.
KENS5 News. [Online; posted 2-June-2010] http: //www.kens5. com/news/
Ex-Shell-CEO- Job-1loss-from-Gulf-011-spill-could-top- 50000-95400
html,
Dismukes, D. E. (2010). Deepwater moratorium: Overview of impacts
for louisiana. Technical report, Baton Rouge, LA.
jrallyforeconomicsurvival . com/PDF/DISMUKES . pdf}
Enrico, M. (2011). Local labor markets. Handbook of labor economics 4, 1237-
1313.
28
Feyrer, J. and B. Sacerdote (2011). Did the stimulus stimulate? real time esti-
mates of the effects of the american recovery and reinvestment act. Technical
report, National Bureau of Economic Research.
GCCF (2010). Gccf program statistics — overall summary. Techni-
cal report. https: //web. archive. org/web/20100923042450/http:
//gulfcoastclaimsfacility.com/GCCF_Overall_Status_Report_
09-21-2010. pdf,
GCCF (2012). Overall program statistics. Technical report. [nttps://web.
archive.org/web/20120515193036/http://gulfcoastclaimsfacility.
com/GCCF_Overall_Status_Report . pdf
Groen, J. A. and A. E. Polivka (2008). The effect of hurricane katrina on the
labor market outcomes of evacuees. The American Economic Review, 43-48.
Harrington, J. (2010). Tracking oil spill job losses a slip-
pery task. The Tampa Bay Times. [Online; posted 17-June-
2010] http://www. tampabay.com/news/business/workinglife/
|tracking-oil-spill-job-losses-a-slippery-task/1103427,
Kotok, D. (2010). Oil slickonomics: Part 8, chemotherapy in the gulf of mex-
ico. [Online; accessed 17-June-2010] http://www. cumber.com/commentary.
jaspx?file=062010.asp)
Louisiana Department of Revenue (2010). Annual tax collection report
09-10. Technical report, Louisiana Department of Revenue, Baton
Rouge, LA. http://www.revenue. louisiana. gov/forms/publications/
|TCReport (09.10)Final . pdf,
Louisiana Department of Revenue (2011). Annual tax collection report 10-
11. Technical report, Baton Rouge, LA. http: //revenue. louisiana. gov/
\forms/publications/TCReport10.11(Final)i.pdf,
Louisiana Workforce Commission (2010). Crisis caused by the
moratorium. Technical report, Baton Rouge, LA. http:
//www.rallyforeconomicsurvival .com/PDF/Eysink/20Louisianas
20Moratorium/20PPT/%2025june/%20920. ppt%20 [Read-Only] . pdf,
Mason, J. R. (2010). The economic cost of a moratorium on offshore oil
and gas exploration to the gulf region. Technical report, Washington,
DC. http://www.saveusenergyjobs.com/wp- content /uploads/2010/07/
'The_Economic_Cost_of_a_Moratorium_on_Offshort%20_0il_and_Gas_
|Exploration_to_the_Gulf_Region. pdf,
National Commission on the BP Deepwater Horizon Oil Spill and Offshore
Drilling (2011). Deep water: The gulf oil disaster and the future of offshore
drilling, report to the president. Technical report, Washington, DC.
29
National Oceanic and Atmospheric Administration (n.d). List of coastal counties
for the bureau of the census statistical abstract series. Technical report.
http: //www.census.gov/geo/landview/lv6help/coastal_cty. pdf,
National Pollution Funds Center (2013). Fpn n10036 mississippi canyon block
252 daily financial summary. Technical report. Provided by Jonathan Abram-
son, Regional Manager, U.S. Coast Guard National Pollution Funds Center,
via email January 9, 2013.
Oxford Economics (2010). Potential impact of the gulf oil spill on
tourism: Report prepared for the u.s. travel association. Technical re-
port. (http://www.ustravel.org/sites/default/files/page/2009/11/,
|Gulf_0i1_Spill_Analysis_Oxford_Economics_710. pdf]
Power, S. and L. Eaton (2010). U.s. saw drill ban killing many jobs. The Wall
Street Journal. [Online; posted 20-August-2010] http: //online.wsj.com/)
larticle/SB10001424052748704488404575441 760384563880 . html,
Richardson, J. (2010). Economic impact of offshore moratorium on
louisiana. Technical report, Baton Rouge, LA. http://www. |
jrallyforeconomicsurvival .com/PDF/Richardson . pdf!
Topel, R. H. (1986). Local labor markets. The Journal of Political Economy,
$111-S143.
U.S. Bureau of Transportation Statistics (n.d.). Air carrier sum-
mary: T3: U.s. air carrier airport activity statistics. http:
//www.transtats.bts.gov/DL_SelectFields.asp?Table_ID=273&DB_
Short_Name=Air/20Carrier/20Summary
U.S. Department of Commerce (2010). Estimating the economic effects of the
deepwater drilling moratorium on the gulf coast economy. Technical report,
Washington, DC. Interagency Economic Report.
U.S. Internal Revenue Service (n.d.). Soi tax stats — migration data.
\//waw.irs.gov/uac/SOI-Tax-Stats-Migration-Data,
Vigdor, J. (2008). The economic aftermath of hurricane katrina. The Journal
of Economic Perspectives 22(4), 135-154.
30
Figures and Tables
Figure 1: 2010 Employment for Oil Parishes, Non-Oil Louisiana Coastal
Parishes, Florida Coastal Counties, Control (Inland) Counties
Source: Constructed by author using QCEW employment data for total covered
employment across all industries.
31
Figure 2: Gulf Coast Region, Coastal Counties and Parishes, and the Louisiana Oil Parishes
32
Figure 3: Estimated Employment Impacts by Coastal Region
33
Notes: The employment estimates are based on the 2010 panel model presented in Table 5 column 1 (left) and the 2008-2010
panel model presented in Table 5 column 2 (right) and average 2009 employment. The 95% condence intervals are presented
as error bars.
Figure 4: Estimated Louisiana Employment Impacts in Statistical Models and Ex Ante Predicted Employment Impacts from
2010
34
Notes: The Model `10 and Model `08-`10 estimates are based on the estimated combined employment impact for non-oil
Louisiana coastal parishes and Louisiana oil parishes for the 2010 panel and 2008-2010 models presented in Table 5, columns
1 and 2, respectively, and average 2009 employment. The 95% condence intervals are presented as error bars for these two
statistical model estimates. Condence intervals or other representations of the distribution around the central estimates are
not available from the sources published in 2010.
Table 1: Ex Ante Estimates of the Employment Impacts of the Offshore Drilling Moratorium
Estimated Job Loss Region Source
8,169 Gulf States [Mason |(2010
9,462 Louisiana
17,464 Louisiana ichardson | (2010
23,247 Not Specified Department of the Interior (DOT; |Power and Eaton (2010 )
30,000 Louisiana Louisiana Mid-Continent Oil and Gas Association GA)
50,000 Not Specified Quoted by John Hofmeister, former CEO, Shell (Desel| 2010 )
200,000-1,000,000 Not Specified [Kotok|(2010}
cE
Table 2: Treatment by 2010 Spill-Related Exogenous Shocks
Region Oil Spill Spill Response Oshore Drilling Moratorium
Alabama Coastal Counties X X
Florida Panhandle Coastal Counties X X
Florida Non-Panhandle Coastal Counties X
Louisiana Oil Parishes X X X
Louisiana Non-Oil Coastal Parishes X X
Mississippi Coastal Counties X X
Texas Coastal Counties X X
Gulf Coast Inland Counties
36
Table 3: Number of Counties and Average County Employment, Wages, and Establishments in 2009, Gulf States
Number of Counties Employment (1000s) Weekly Wage (2010$) Establishments (1000s)
Region Coastal Inland Coastal Inland Coastal Inland Coastal Inland
Gulf Coast States 139 395 57 36 693 648 3.7 2.4
(182) (121) (139) (138) (9.6) (7.9)
Alabama 8 59 34 26 681 655 2.4 1.6
(53) (51) (134) (121) (3.2) (2.8)
Florida 40 27* 66 163 645 720 5.3 13.7
(111) (243) (87) (145) (8.1) (20.4)
Panhandle 18 29 630 2.1
(42) (72) (2.7)
Non-Panhandle 22 95 657 7.9
37
(139) (96) (9.9)
Louisiana 38 26 39 14 772 634 2.6 1.0
(57) (25) (153) (146) (3.6) (1.6)
Oil Parish 5 56 850 3.6
(39) (36) (2.8)
Non-Oil Parish 33 36 760 2.4
(59) (161) (3.7)
Mississippi 12 70 18 12 618 585 1.1 0.8
(24) (18) (127) (92) (1.2) (1.2)
Texas 41 213 81 32 692 658 4.2 1.8
(308) (128) (138) (145) (15.2) (6.4)
Source: BLS (nd).
Notes: Standard deviations presented in parentheses. * Some inland Florida counties refer to those on the Atlantic coast.
8E
Table 4: Estimated Employment and Wage Impacts, All Industries, Gulf Coast States, 2010
(1)
(2)
(3)
(4)
(5)
(6)
Region In(employment) In(employment) In(employment) | In(wage) In(wage) In(wage)
Gulf Coast -0.0032 -0.0016
(0.0020) (0.0051)
Gulf Coast ex Oil Parishes -0.0034 -0.0024
(0.0021) (0.0052)
Oil Parishes 0.012*** 0.012*** 0.021*** 0.020***
(0.0040) (0.0040) (0.0044) — (0.0043)
LA Coast ex Oil Parishes 0.0026 -0.025**
(0.0025) (0.012)
AL Coast 0.013** 0.038**
(0.0061) (0.016)
FL Panhandle Coast -0.0075 -0.0013
(0.0048) (0.0011)
FL Non-Panhandle Coast -0.027*** -0.0079
(0.0040) (0.012)
MS Coast -0.0016 -0.015
(0.0054) (0.016)
TX Coast 0.0026 -0.0022
(0.0029) (0.0066)
Sample Period 2010 2010 2010 2010 2010 2010
N (counties) 534 534 534 534 534 534
N 5,340 5,340 5,340 1,602 1,602 1,602
Notes: Robust standard errors clustered by county. All models include fixed effects by county and month (or quarter).
** * denote statistical significance at the 1, 5, and 10% levels.
2K K
’
6¢
Table 5: Estimated Employment Impacts, All Industries, Gulf Coast States, Seasonality Robustness Checks
(1) @) 3) 4) ©) ©
Region In(employment) In(employment) In(employment) In(employment) In(employment) In(employment)
Oil Parishes 0.012*** -0.0021 0.022*** 0.0030 0.010*** -0.0050
(0.0040) (0.0074) (0.0048) (0.0076) (0.0038) (0.0074)
LA Coast ex Oil Parishes 0.0026 0.012** 0.0064** 0.016*** 0.0045* 0.011**
(0.0025) (0.0051) (0.0030) (0.0055) (0.0023) (0.0050)
AL Coast 0.013** 0.0022 0.017*** 0.0061 0.020*** 0.0026
(0.0061) (0.021) (0.0063) (0.021) (0.0060) (0.020)
FL Panhandle Coast -0.0075 -0.0056 -0.0036 -0.0011 -0.0032 -0.0052
(0.0048) (0.0065) (0.0051) (0.0066) (0.0047) (0.0065)
FL Non-Panhandle Coast -0.027*** -0.017* -0.023*** -0.013 -0.023*** -0.017**
(0.0040) (0.0086) (0.0043) (0.0090) (0.0039) (0.0085)
MS Coast -0.0016 -0.0053 0.0022 -0.0014 -0.0020 -0.0064
(0.0054) (0.013) (0.0056) (0.0013) (0.0053) (0.0013)
TX Coast 0.0026 0.016*** 0.0064* 0.020*** -0.00012 0.015***
(0.0029) (0.0053) (0.0035) (0.0057) (0.0030) (0.0053)
Sample Period 2010 2008-2010 2010 2008-2010 2008-2010 2008-2010
Seasonal Effects Month Month-by-Coastal Month-by-State
N 5,340 18,156 5,340 18,156 5,340 18,156
Notes: Robust standard errors clustered by county. All models include fixed effects by county and month (2010 panel) or county, month,
and year (2008-2010 panel). Each sample includes 534 counties/parishes. ***, **, * denote statistical significance at the 1, 5, and 10%
levels.
OF
Table 6: Estimated Wage Impacts, All Industries, Gulf Coast States, Seasonality Robustness Checks
(1) (2) (3) (4) (5) (6)
Region In(wage) In(wage) In(wage) In(wage) In(wage) — In(wage)
Oil Parishes 0.020*** 0.020** 0.0083 0.025*** 0.030*** 0.020**
(0.0043) (0.0087) (0.0095) (0.0088) (0.0040) —(0.0087)
LA Coast ex Oil Parishes -0.025** -0.0032 -0.038*** -0.0020 -0.0097 -0.0036
(0.012) (0.0083) (0.013) (0.0081) ~—- (0.012) ~~—- (0.0083)
AL Coast 0.038** 0.058*** 0.026 0.063*** 0.060*** 0.054***
(0.016) (0.012) —- (0.018) ~—(0.012) (0.015) (0.012)
FL Panhandle Coast -0.0013 0.0031 -0.013 0.0080 0.015 0.0010
(0.0011) (0.0083) (0.014) ~— (0.0081) ~—s- (0.011) ~~—(0.0083)
FL Non-Panhandle Coast —_-0.0079 -0.0014 -0.020 0.0034 0.0083 -0.0035
(0.012) (0.0073) (0.014) ~—- (0.0068) ~—s- (0.011) ~—_—(0.0073)
MS Coast -0.015 -0.0051 -0.027 -0.00026 -0.011 -0.00029
(0.016) (0.014) —- (0.018) _~—-(0.014) (0.016) (0.014)
TX Coast -0.0022 0.014*** -0.014 0.019*** -0.0040 0.018***
(0.0066) (0.0051) (0.010) (0.0051) ~—- (0.0070) ~~—-(0.0059)
Sample Period 2010 2008-2010 2010 2008-2010 2008-2010 2008-2010
Seasonal Effects Quarter Quarter-by-Coastal Quarter-by-State
N 1,602 5,874 1,602 5,874 1,602 5,874
Notes: Robust standard errors clustered by county. All models include fixed effects by county and month (2010 panel) or county,
month, and year (2008-2010 panel). Each sample includes 534 counties/parishes. ***, **, * denote statistical significance at
the 1, 5, and 10% levels.
IV
Table 7: Estimated Employment Impacts, All Industries, Gulf Coast States, Geographic Robustness Checks
(1) (2) (3) (4) 5) 6)
Region In(employment) In(employment) In(employment) In(employment) In(employment) = In(employment)
Oil Parishes -0.0048 -0.010 0.022** 0.010 0.0097* -0.0032
(0.0039) (0.0072) (0.0044) (0.0076) (0.0040) (0.0075)
LA Coast ex Oil Parishes -0.0090** 0.0043 0.0072** 0.022** 0.014** 0.029
(0.0024) (0.0049) (0.028) (0.050) (0.0050) (0.019)
AL Coast 0.0016 -0.0060 0.018** 0.013 0.027* 0.019**
(0.0061) (0.021) (0.0062) (0.020) (0.013) (0.0065)
FL Panhandle Coast -0.019** -0.013* -0.0029 0.0055 0.00090 0.00039
(0.0048) (0.0064) (0.0050) (0.0065) (0.0058) (0.010)
FL Non-Panhandle Coast -0.039** -0.024** -0.023** -0.0061 -0.031** -0.032**
(0.0040) (0.0086) (0.0042) (0.0085) (0.0053) (0.0072)
MS Coast -0.013* -0.013 0.0029 0.0052 0.0084 0.017
(0.0053) (0.013) (0.055) (0.013) (0.082) (0.015)
TX Coast -0.0091** 0.0085 - - -0.00036 0.019
(0.0029) (0.0052) (0.) (0.) (0.0044) (0.013)
Sample Period 2010 2008-2010 2010 2008-2010 2008-2010 2008-2010
Sample Region National Gulf States excluding TX Gulf States excluding buffer counties
N(counties) 2,865 2,865 280 280 444 444
N 28,650 97,410 2,800 9,520 4,440 15,096
Notes: Robust standard errors clustered by county. All models include fixed effects by county and month (2010 panel) or county, month,
and year (2008-2010 panel). ***, **, * denote statistical significance at the 1, 5, and 10% levels.
GV
Table 8: Estimated Wage Impacts, All Industries, Gulf Coast States, Geographic Robustness Checks
(1) (2) (3) (4) (5) (6)
Region In(wage) In(wage) — In(wage) In(wage) In(wage) In(wage)
Oil Parishes 0.032*** 0.025*** —0.022*** 0.022** 0.020*** 0.019***
(0.0035) (0.0085) (0.0049) (0.0088) (0.0044) (0.0088)
LA Coast ex Oil Parishes -0.022* -0.0048 — -0.036*** -0.0066 -0.034 -0.037
(0.012) (0.082) ~—- (0.012) (0.0082) (0.029) (0.032)
AL Coast 0.050*** 0.063*** 0.040** 0.061*** 0.020* 0.019**
(0.015) (0.012) ~—- (0.016) (0.012) (0.011) (0.0087)
FL Panhandle Coast 0.010 0.0079 0.00044 0.0053 -0.0053 -0.0011
(0.011) (0.0082) ~— (0.012) (0.0082) (0.011) (0.011)
FL Non-Panhandle Coast 0.0036 0.0033 -0.0061 0.00071 -0.023*** -0.014***
(0.011) (0.0071) ~—- (0.012) (0.0073) (0.0060) (0.0052)
MS Coast -0.0033 -0.00039 -0.013 -0.0030 -0.015*** -0.0096
(0.016) (0.014) ~—- (0.016) (0.014) (0.0052) (0.0072)
TX Coast 0.0092 0.0419*** - - -0.012 0.012
(0.061) (0.0050) (0.) (0.) (0.013) (0.010)
Sample Period 2010 2008-2010 2010 2008-2010 2008-2010 2008-2010
Sample Region National Gulf States excluding TX Gulf States excluding buffer counties
N(counties) 2,867 2,867 280 280 444 444
N 8,601 31,537 840 3,080 1,332 4,884
Notes: Robust standard errors clustered by county. All models include fixed effects by county and quarter (2010 panel) or
county, quarter, and year (2008-2010 panel). ***, **, * denote statistical significance at the 1, 5, and 10% levels.
eV
Table 9: Estimated Industry-Specific Employment and Wage Impacts, Gulf Coast States
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Region Mining Support 213 Accommodation 721 Mining Support 213. Accommodation 721
In(employment) In(wage)
Oil Parishes -0.063*** 0.0014 0.048** 0.017 -0.047* -0.0030 0.015 0.042*
(0.0017) (0.039) (0.024) ~—- (0.037) _-| (0.027) (0.020) ~—- (0.035) ~— (0.024)
LA Coast ex Oil Parishes 0.013 0.14** 0.014 0.0078 -0.048 -0.0023 -0.020 0.022
(0.014) (0.0062) (0.017) ~—«- (0.038) | (0.033) (0.034) ~—- (0.023) _~— (0.019)
AL Coast -0.11*** -0.11*** 0.026 0.052 -0.18*** -0.0022 0.12 0.12
(0.0060) (0.018) (0.040) ~—- (0.043) _-| (0.015) (0.014) (0.10) (0.073)
FL Panhandle Coast - - 0.0089 0.034 - - 0.036 0.030
(0.022) (0.037) (0.040) (0.027)
FL Non-Panhandle Coast - - -0.047*** — -0.037** - - -0.062*** — -0.036***
(0.012) (0.017) (0.021) (0.011)
MS Coast -0.072*** 0.0025 -0.048*** 0.0012 0.033** 0.016*** -0.0081 0.011
(0.0050) (0.018) (0.016) ~—- (0.042) _-| (0.015) (0.012) ~—-(0.077) ~—« (0.030)
TX Coast 0.014* -0.032 -0.027** -0.052 -0.017 0.082*** -0.0034 0.024
(0.0083) (0.028) (0.013) ~—- (0.034) _-| (0.027) (0.027) ~—- (0.042) ~—S (0.041)
Sample Period 2010 2008-2010 2010 2008-2010 2010 2008-2010 2010 2008-2010
N(counties) 118 118 247 247 118 118 223 223
N 1,180 4,012 2,470 8,398 354 1,298 669 2,453
Notes: Robust standard errors clustered by county. All models include fixed effects by county and month (quarter) (2010
panel) or county, month (quarter), and year (2008-2010 panel). ***, **, * denote statistical significance at the 1, 5, and 10%
levels.
VV
Table 10: Estimated Establishment Impacts, All Industries, Gulf Coast States, Seasonality Robustness Checks
(1) (2) 3) @) () 6
Region In(establishments) In(establishments) In(establishments) In(establishments) In(establishments) In(establishments)
Oil Parishes 0.0080*** 0.039*** 0.0020 0.038*** 0.0087*** 0.040***
(0.0018) (0.0036) (0.0025) (0.0036) (0.0017) (0.0036)
LA Coast ex Oil Parishes 0.0097*** 0.040*** 0.0054*** 0.041*** 0.011*** 0.040***
(0.0022) (0.0037) (0.0020) (0.0037) (0.0021) (0.0037)
AL Coast -0.0017 -0.030*** -0.0076*** -0.031*** 0.00073 -0.029***
(0.0022) (0.0050) (0.0029) (0.0050) (0.0020) (0.0050)
FL Panhandle Coast -0.0099*** -0.021*** -0.016*** -0.021*** -0.0083*** -0.020***
(0.0032) (0.0048) (0.0036) (0.0051) (0.0031) (0.0048)
FL Non-Panhandle Coast -0.019*** -0.037*** -0.025*** -0.037*** -0.018*** -0.036***
(0.0045) (0.0070) (0.0047) (0.0070) (0.0044) (0.0069)
MS Coast -0.00014 -0.011** -0.0061* -0.012** -0.00027 -0.011**
(0.0027) (0.0051) (0.0032) (0.0052) (0.0028) (0.0051)
TX Coast 0.00064 0.0090* -0.0053* 0.0083 -0.00032 0.0083
(0.0026) (0.0052) (0.0030) (0.0053) (0.0027) (0.0052)
Sample Period 2010 2008-2010 2010 2008-2010 2008-2010 2008-2010
Seasonal Effects Quarter Quarter-by-Coastal Quarter-by-State
N 1,602 5,874 1,602 5,874 1,602 5,874
Notes: Robust standard errors clustered by county. All models include fixed effects by county and quarter (2010 panel) or county, quarter,
and year (2008-2010 panel). Each sample includes 534 counties/parishes. ***, **, * denote statistical significance at the 1, 5, and 10%
levels.
Table A.1: Counties by Treatment Region
Region Counties/Parishes by FIPS Code
Alabama Coast 1003*, 1025, 1039, 1053, 1061, 1097*, 1099, 1129
12005*, 12013, 12033*, 12037*, 12039, 12045*, 12059, 12063, 12065*, 12073, 12077, 12079, 12091*, 12113*, 12123*, 12129*,
Florida Panhandle Coast 12131*, 12133
Florida Non-Panhandle Coast 12015*, 12017*, 12021*, 12027, 12029*, 12041, 12043, 12049, 12053*, 12057, 12067, 12071*, 12075*, 12081*, 12083, 12087*,
12101*, 12103*, 12105, 12115*, 12119, 12121
45
Louisiana Oil Parishes 22045*, 22055*, 22057*, 22101*, 22109*
22001, 22005, 22007, 22009, 22011, 22019, 22023*, 22033, 22037, 22039, 22047, 22051*, 22053, 22063, 22071, 22075*, 22077,
Louisiana Non-Oil Coast 22079, 22085,22089, 22091, 22093, 22095, 22097, 22099, 22103, 22105, 22113*, 22115, 22117, 22121, 22125
Mississippi Coast 28005, 28039, 28045, 28047*, 28059*, 28073, 28091, 28109, 28113, 28131, 28147, 28157
48007*, 48015, 48025, 48039*, 48047, 48057*, 48061*, 48071*, 48089, 48123, 48131, 48149, 48157, 48167*, 48175, 48201,
Texas Coast 48215, 48239, 48241, 48245*, 48247, 48249, 48261*, 48273*, 48285, 48291, 48297, 48321*, 48351, 48355*, 48361, 48391,
48409*, 48427, 48457, 48469, 48473, 48477, 48479, 48481, 48489*
Notes: * refer to counties classied as coastal under the more narrow denition of coastal in the robustness checks.
File and source
- File
- the-labor-market-impacts-of-the-2010-deepwater-horizon-oil-spill-and-off-shore-oil-drilling-moratorium.pdf
- Size
- 674,771 bytes
- SHA-256
- e16d9e410c9e0a2bf6467d83c7772fc37add35f271b56da520f795f7e3539464
- Original
- www.nber.org