This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Slide 1
By Amanda Weinstein [email protected] Nov. 26, 2012 Local
Labor Market Restructuring in Shale Booms
Slide 2
Outline Introduction to Shale Regional Shocks and Natural
Resource Booms Methodology Results Conclusions
Slide 3
Motivation Commenting on shale energy development, Aubrey
McClendon CEO of Chesapeake Energy of Oklahoma was quoted in the
Columbus Dispatch saying, This will be the biggest thing in the
state of Ohio since the plow. Various impact studies have estimated
large employment effects for Ohio, Pennsylvania, and other areas.
We are concerned that job numbers may be overinflated by the
industry (or any industry) Policy makers often use these job
numbers to justify supporting the industry through tax breaks and
other measures We need to create a counterfactual to estimate what
would have happened if there was no shale development. The
difference between what did happen and the counterfactual is the
shale development effect
Slide 4
Shale Booms Innovations in oil and gas extraction along with
rising oil and gas prices have led to shale development across the
U.S. Hydraulic fracturing and micro-seismic technology Impact on
local employment and earnings The nature of local adjustments to
economic shocks Natural resource curse Restructuring in the local
labor market due to displacement effects (including Dutch disease)
and other spillovers
Slide 5
Hydraulic Fracturing
Slide 6
Drilling Tower and Capped Well Marcellus Shale horizontal
drilling tower in Lycoming County, PA.
Slide 7
U.S. Shale Plays
Slide 8
Shale Gas Production
Slide 9
Tight Oil production
Slide 10
Actual and Projected Production (EIA)
Slide 11
Actual and Projected Production
Slide 12
The Employment Boom The boom in employment generally preceded
the boom in production as many areas have a significant
construction period before drilling began
Slide 13
North Dakota North Dakota oil and gas employment has shot up
from holding steady at about 1,800 in 2004 to11,700 in 2011.
Slide 14
Regional Shocks The shale boom may be viewed more as a
transitory shock whether it is or not Wages and prices adjust more
in booms than busts and are more flexible in a transitory shock
than a permanent one (Blanchard and Katz, 1992; Topel, 1986) After
a shock, states return to the same growth rate on a different
growth path (Blanchard and Katz,1992) Long run impacts are often
negligible Military base closings (Dardia et al., 1996; Hooker and
Knetter, 1999; Popper and Herzog, 2003) Large plant openings
(Greenstone and Moretti, 2004; Edmiston, 2004)
Slide 15
Previous Natural Resource Shocks
Slide 16
Natural Resource Shocks Evidence of a resource curse has been
found across nearly all levels of geography (Papyrakis and Gerlagh,
2007; Kilkenny and Partridge, 2009; James and Aadland, 2011)
Reasons for the poor performance have generally been focused on
their institutions (Mehlum et al., 2006; Rodriguez and Sachs, 1999)
U.S. counties may be affected through mechanisms other than local
institutions Specialization in natural resource extraction may lead
to a less diverse and more volatile economy Natural resources have
been found to affect agglomeration at the state level (but not
lower levels of geography), natural resources used for energy are
found to have no significant effect (Rosenthal and Strange,
2001)
Slide 17
Natural Resource Shocks Black et al. (2005) analyze the impact
of the coal boom in the 1970s and the subsequent coal bust in the
1980s. Little evidence of crowding out Less than 2 jobs created for
every 10 coal jobs created during the boom but 3.5 jobs lost for
every 10 coal jobs lost during the bust. Marchand (2012) found that
for every 10 energy extraction jobs created in Western Canada,
there were 3 construction jobs, 2 retail jobs, and 4.5 service jobs
created Recent analysis of the impact of mountaintop mining find
that it may reduce poverty rates in the short term but not long
term (Partridge et al., forthcoming; Deaton and Niman, 2012) Weber
(2012) finds that $1 million in shale gas production results in
2.35 jobs within counties in Texas, Colorado, and Wyoming
Slide 18
Data: Setting up the Counterfactual Economic Modeling
Specialists Intl. (EMSI) data from 2001-2011 covers the general
boom period from about 2005 onward and the years leading up to the
shale boom EMSI data provides detailed employment and earnings data
at the county level Counties in the lower 48 states (3,060
counties) Controls population and education (U.S. Census Bureau),
industry composition, county fixed effects, economic trends Define
the boom counties and boom period
Slide 19
Data: Setting up the Counterfactual Measuring the boom Both
Black et al. (2005) and Marchand (2012) measure the boom using
proportion of earnings derived from natural resource extraction
which misses development in new counties such as in Pennsylvania
and many other areas Weber (2012) uses data on gas production and
earnings from production which may miss the benefits of initial
construction as well as the tapering off when the drilling period
ends EMSI data allows us to measure the boom using employment
Measuring the boom period General boom period for U.S approx 2005 -
2011 Define boom period by state
Slide 20
Boom Periods by State
Slide 21
Change in Oil and Gas Employment Direct oil and gas employment
is measured as the sum of industry codes 2111 (oil and gas
extraction) and 2131 (support activities for mining).
Slide 22
Shale Boom Counties In a shale booming state (defined by oil
and gas production and employment) At least 10% increase in oil and
gas employment growth and at least 20 additional oil and gas
workers during the boom period.
Slide 23
Descriptive Statistics
Slide 24
Boom vs. Non-boom Counties: Employment
Slide 25
Boom vs. Non-boom Counties: Earnings Boom counties seem to be
benefitting in terms of employment and earnings though pre-boom
trends varied between the two
Slide 26
Methodology At best, a well done impact study should tell you
how many jobs are supported by an industry, not how many jobs it
created. Not a counterfactual The goal of the
difference-in-difference methodology is to set up this
counterfactual The difference-in-difference approach Y it is
ln(employment or earnings) Parameter of interestCounty Fixed
Effect
Slide 27
Difference-in-Difference Results Boom counties were associated
with an annual increase in employment of 1.59% and Increase in
earnings of 3.08%
Slide 28
Difference-in Difference with Trends From Greenstone et al.
2010 (large plant openings) When 2 = 3 = 5 = 7 = 0, reduces to
equation 1 As shown in the previous employment and earnings growth
graphs the economic trends leading up to the boom period are
different for boom counties and non-boom counties
Slide 29
Results with Trends The effect on earnings is again nearly
double that of the effect on employment The impact of shale
development decreases over time
Slide 30
Endogeneity The OLS methodology assumes that a change in oil
and gas production (and employment) is driven by an exogenous shock
Concerns that shale development may occur in pro-business counties
biasing our results Shale development may be occurring in
struggling communities trying to attract economic development of
any kind Or in communities that have done well in the past because
of their pro- business policies. Instrument for boom counties using
the percent of the county covering shale resources which we would
expect to be endogenous Although there may be endogeneity in the
location choice of drilling firms, we expect that county fixed
effects (and various other controls are sufficient)
Slide 31
Instrument: Percent Shale
Slide 32
Instrumental Variables Results Significant first stage results
using percent shale Instrumental variables coefficient estimates
similar in sign to previous estimates though not significance
Hausman tests suggest that our identification of shale boom
counties is not endogenous
Slide 33
The Size of the Boom The binary variable used for boom counties
may miss some of the variability in the size of the boom and the
impact of shale development We would also like to estimate the
employment multiplier associated with shale employment growth
Equation 3 below incorporates the size of the boom measured by
ln(oil and gas employment) Direct oil and gas employment is
measured as the sum of industry codes 2111 (oil and gas extraction)
and 2131 (support activities for mining).
Slide 34
Finding the Employment Multiplier There is on average 1 oil and
gas worker for every 87 non-oil and gas workers in shale boom
counties in 2005 1 additional oil and gas worker is associated with
0.46 additional jobs (or a multiplier of 1.46) Calculations similar
to Moretti (2010)
Slide 35
Impact on the Traded and Nontraded Sectors The impact on
tradable sectors other than oil and gas is 1.08 The employment
multiplier for nontradable sectors is 1.42
Slide 36
Conclusions Labor Market Restructuring Although we find little
evidence of crowding out, the multiplier effect is only significant
for the nontradable goods sector With an employment multiplier less
than 2, the local labor market is restructuring by shifting the
share of employment toward oil and gas extraction jobs The average
percent of mining employment in boom counties increased from
approximately 4% before the boom period to 6.8% in 2011. The
average percent of mining in non-boom counties remained steady
before and during the boom at about 0.89% of total employment.
Slide 37
Conclusions The Economic Impact Boom counties experienced an
increase in both employment growth and earnings growth As Blanchard
and Katz (1992) found, growth rates seem to be returning to their
original levels after the initial increase in growth The Employment
Multiplier At 1.46, the employment multiplier is lower than
previously expected and than impact studies suggest Policy
implications Importance of realistic expectations Future Research:
multipliers by region or state, border counties and spatial
spillovers 37
Slide 38
Amanda Weinstein Graduate Research Assistant The Swank Program
in Rural-Urban Policy Dept. Agricultural, Environmental &
Development Economics The Ohio State University
([email protected]) Thank You 38
Slide 39
Extra Slides
Slide 40
Difference-in-Difference Methodology [E(Y b0 )-E(Y n0 )]-[E(Y
b0 )-E(Y n0 )] The difference-in-difference approach Y it is
ln(employment or earnings) Through asymptotics it can be shown that
the probability limit of b3 is
Slide 41
U.S. Shale Plays
Slide 42
Prices - Booms and Busts
Slide 43
Oil Prices
Slide 44
Major Holders of Utica Shale Right in Ohio (April 2012)