Transcript Slide 1

Measuring The Economic Value
of Shale Energy Development
Presented for the
BU/SRSA Shale Workshop
Mark Partridge & Amanda Weinstein
Presented at Bucknell University
July 31, 2012
Swank Chair in Rural-Urban Policy
([email protected])
Department of Agricultural, Environmental & Development Economics
Ohio State University Extension
Shale: Economic Development
Game Changer?
 I will follow Weinstein and Partridge
(2011) and Farren et al. (2012).
 I will focus on economic development.
Of course, shale development also has
major implications on world and US
energy markets—see map.
 For those interested in local/regional
growth, long-term economic outcomes
receive more weight than temporary
booms revolving around construction.
2
Shale Energy is found all over US and the world.
Is Shale a Game Changer?
 Commenting on Ohio’s shale energy
development: “This will be the biggest thing
in the state of Ohio since the plow…This is
truly extraordinary.” Aubrey McClendon
CEO of Chesapeake Energy of Oklahoma.
 Quoted in the Columbus Dispatch “Realism on Renewable Energy.”
September 22, 2011, Pp. B1-B2.
 Economists have 150 years of evidence on natural
resource booms and the evidence is often negative
(e.g., Papyrakis, E. and R. Gerlagh, 2007; Kilkenny and Partridge, 2009;
James and Aadland, 2011).
 E.g., Natural Resources Curse & Dutch Disease

More cases like LA, WV, Venezuela, Nigeria vs Norway
4
Is shale a game changer?
 In the latest year, PA has gained about
6,000 mining jobs (minus coal mining)
and about 40,000 total jobs.
Since 2006, PA has gained about 18,500
mining jobs (minus coal mining).
 PA’s
total employment is over 5.7million
 Ohio has gained about 500 mining jobs
and 100,000 total jobs.
Source U.S. Bureau of Labor Statistics
CES measured over June 2011-June 2012.
5
Figure 7: Total Employment and Previous Oil
Booms in the U.S.: 1969=100
6
Shale: Game Changer?
1. Economists point out that ‘projects’ and
policies should be judged on their net
benefits and costs, and NOT net job creation.
 E.g., CO2 content of coal vs natural gas.
 E.g., lower energy costs (but energy security is
not a large issue since NG replaces US coal).
2. The best source of an industry’s actual
economic impact is NOT the industry itself,
studies paid for by the industry, or
sympathetic politicians and newspapers.
 This is not a surprise .
 In serious research, we use peer review to weed
out poor studies. We create counterfactuals.
7
Shale: Game Changer?
A counterfactual is what would have happened if
there was no shale industry. The difference
between the number of jobs that happened and
the counterfactual is the actual jobs created.
3. So-called ‘impact studies’ that estimate direct
and indirect effects are over-estimates of new
job creation and serious regional economists
have not viewed them as anywhere near best
practice for decades. NOT COUNTERFACTUALS!
At best, an impact study should tell how many
jobs are ‘supported’ by an industry, not how
many jobs it ‘created’ and explain the difference.
Energy is a capital intensive industry—fewer jobs.
8
Shale: Game Changer?--#3 cont.
 Even in good impact studies, the
“employment” effects are not continuous
but in a piecemeal fashion. Construction,
then drilling, then pipelines, and so on.
 They are usually based on slightly dated
national input-output estimates. Heavily
weigh the Oil-Patch supply chain
response, not actual PA/OH response.
9
New drilling activity and its capital intensive nature in PA.
Taken from: http://www.donnan.com/Marcellus-Gas_Hickory.htm
10
Shale: Game Changer?--#3 cont.
“Penn State (Considine) Impact Studies” funded by
the shale industry is an example. It predicts
111,000 jobs in 2011 and 212,000 in 2020 using
the IMPLAN software. {see Kelsey et al. (2011) for a
different point of view}
 Kleinhenz
& Associates (2011) funded by the industry
predicted over 200,000 jobs in Ohio by 2015.
 Ohio Shale Coalition (2012) predicted 66,000 by
2014.
11
#3 Continued
 Impact studies typically ignore displacement
effects and do not compare development
impact to the counterfactual.
Example of a coal counterfactual is Black et al.
2005 in Economic Journal. Multiplier of 1.25.
 PA and OH studies estimate 95% of shale
industry purchases are in PA and 90% in Ohio.
Examples of other problems:
No Price Effects or crowding out.
Entrepreneurs do other things.
Nationally, more natural gas means less
coal needed for electricity and fewer coal
jobs. [oil is new jobs]
12
Example of displacement or labor shortages
elsewhere in the economy in North Dakota
Bakken region.
http://graphics8.nytimes.com/images/misc/pixel.gif
What we do in our comparison
Ohio to Pennsylvania?
 (1) Assessment of impact analysis
 (2) Statistical regressions on the entire state




of Pennsylvania
(3) Compare PA to North Dakota’s Bakkan
shale region which has had a similar
employment change in ‘mining.’
(4) Examine the employment life cycle
effects of natural gas and coal per kwh.
(5) Compare drilling counties with similar
non-drilling counties in PA.
(6) Show the industry is too small to
materially affect Ohio/PA employment.
14
Findings for OH based on PA
 We conclude Ohio’s expected
employment effects are near 20,000
workers (not counting displacement).
 There are relatively “large” income
effects in affected counties.
 We do a difference in difference
assessment of those with heavy mining
vs similar counties w/o mining to get a
handle on the actual income and job
creation.
15
Estimates of the number of jobs required to produce a kWh by energy source
Source: Weinstein et al. (2010) chart using data from Kammen et al. (2004)
Table 3: Effects of Displacing Coal with Natural Gas
Total kWh from Coal
2009
Change in
Jobs
Change in Energy
Costs (millions)
Change in Emissions
(lbs)
Ohio
113,711,997,000
-195
-$491,804
-23,822,663,372
Pennsylvania
105,474,534,000
-181
-$456,177
-22,096,914,873
Source: EIA and Weinstein et al. (2010)
16
OH and PA Natural Gas Related Employment
2001=100
170
160
150
140
130
120
110
100
90
80
70
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
OH
PENN
Source: U.S. Dept. of Labor QCEW http://data.bls.gov/pdq/querytool.jsp?survey=en
Note: 21111-Oil and gas extraction 213111 - Drilling Oil and Gas Wells 213112 - Support Activities for Oil and Gas Operations 541360 - Geophysical Surveying and Mapping Services 238912 Nonresidential Site Preparation Contractors 333132 - Oil and Gas Field Machinery and Equipment Manufacturing
486210 - Pipeline Transportation of Natural Gas 237120 - Oil and Gas Pipeline Construction
333911 - Pump and Pumping Equipment Manufacturing for natural gas wells
17
OH and PA Natural Gas Related Employment
30000
28000
26000
24000
22000
20000
18000
16000
14000
12000
10000
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
OH
PENN
Source: U.S. Dept. of Labor QCEW http://data.bls.gov/pdq/querytool.jsp?survey=en
Note: 21111-Oil and gas extraction 213111 - Drilling Oil and Gas Wells 213112 - Support Activities for Oil and Gas Operations 541360 - Geophysical Surveying and Mapping
Services 238912 - Nonresidential Site Preparation Contractors 333132 - Oil and Gas Field Machinery and Equipment Manufacturing 333911 - Pump and Pumping Equipment
Manufacturing for natural gas wells
486210 - Pipeline Transportation of Natural Gas 237120 - Oil and Gas Pipeline Construction
18
Percent Natural Gas Non-Farm Employment Share: OH and PA
0.6
0.5
0.4
0.3
0.2
0.1
0.0
2001
2002
2003
2004
2005
2006
OH
2007
2008
2009
2010
PA
Source: U.S. Dept. of Labor QCEW http://data.bls.gov/pdq/querytool.jsp?survey=en and U.S. Bureau of Labor Statistics CES, Total Nonfarm Employment by state, www.bls.gov.
Note: 21111-Oil and gas extraction 213111 - Drilling Oil and Gas Wells 213112 - Support Activities for Oil and Gas Operations 541360 - Geophysical Surveying and Mapping Services 238912 Nonresidential Site Preparation Contractors 333132 - Oil and Gas Field Machinery and Equipment Manufacturing
486210 - Pipeline Transportation of Natural Gas 237120 - Oil and Gas Pipeline Construction
333911 - Pump and Pumping Equipment Manufacturing for natural gas wells
19
Table 1: Pennsylvania County Descriptive Statistics
Population Per Capita Employment Employment
Income
Income
2005
Income Growth Rate Growth Rate Growth Rate Growth Rate
2005
2001-2005
2005-2009
2001-2005
2005-2009
NonDrilling
Counties
Drilling
Counties
255,508 $32,187
5.3%
-0.4%
12.6%
13.6%
124,928 $27,450
1.4%
-0.6%
12.8%
18.2%
Source: BEA
PA Counties considered in our simple
difference in difference counterfactual
Southern PA Matched Employment Pairs
110
105
100
95
90
85
80
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Three Mining Counties
Three Non-Mining Counties
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
22
Northeastern PA Matched Employment
Pairs
110
105
100
95
90
Three Mining Counties
Three Non-Mining Counties
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.go
23
Southern PA Matched Per Capita Income Pairs
130
125
120
115
110
105
100
95
90
85
80
75
70
65
Three Mining Counties
Three Non-Mining Counties
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
24
Northeastern PA Matched Per Capita
120
115
110
105
100
95
90
85
80
75
70
65
Three Mining Counties
Three Non-Mining Counties
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
25
Regression Approaches—the
current gold standard
 The need to establish a counterfactual in
peer-reviewed research.
 Many approaches. The three I consider are
(1) matching, (2) 2SLS, and (3) difference in
difference.
 Matching uses trend and level attributes to
identify otherwise equal locations when the
treatment began—identify counter factual.
Regress the sample of matched and non-
matched counties using an indicator for
treatment (drilling). Complex methods can
use trends as well (GHM, JPE, 2010).
26
Regression Approaches—the
current gold standard
 Drilling may be endogenous because
places that accept drilling may be
different than other places.
 2SLS—find an instrument that is only
related to the outcome through
indirectly affecting the treatment.
E.g., geology only affects economic
activity through indirectly affecting shale
energy employment. Instrument is then
used to predict where drilling takes place.
My results are OLS.
27
Regression Approaches—the
current gold standard
 Difference in Difference
Take the change in the economic outcome
and regress it on the change in treatment.
E.g., difference in the percent change in
employment growth regressed on the
change in the number of wells in periods t,
and 0. [period t is say 4 years after shale
drilling and period 0 is say 4 years prior
to drilling.
 Need
to pick initial periods carefully due to
anticipation effects.
28
Regression Approaches—the
current gold standard
 You can include trend effects.
 D in D’s key advantage is it
differences out county fixed effects
that influence long-term growth—
e.g., culture, or demographics.
 One can still condition on other
variables that may affect near term
economic growth. For instance,
initial income or population.
29
The Model
C is county fixed effect
Period 0:
Yi0= β0 + β1(Number of Wells)i0 + Ci + εi0
Period 1:
Yi1= β0 + β1(Number of Wells)i1 + Ci + εi
Difference the 2 equations and C falls out:
Yi1- Yi0= β0 + β1(Δ Number of Wells) + εi
Add in other X variables to condition on.
β1 measures the positive indirect and induced
effects net of any displacement
30
Short-term Employment and Income Effects of Drilling
Change in Percent Employment Growth 2005-2009
Compared to 2001-2005
Parameter Estimate
t-value
Total Wells 05-09 min. 01-05 Wells
2001 Log Population
2001 Log Per Capita Income
N
R2
Adjusted-R2
1.769E-05
1.14
0.023
-0.096
67
0.118
0.076
2.64
-1.55
Change in Percent income Growth
Parameter Estimate
t-value
Total Wells 05-09 min 01-05 Wells
2001 Log Population
2001 Log Employment
N
R2
Adjusted-R2
2.515E-05
2.11
0.084
-0.086
67
0.205
0.167
Source: BEA and PA DEP Date
2.53
-2.76
31
Employment Growth Using
National Data.
1. Yi0= β0 + β1(Oil and Gas Employment Growth)i0 + Ci + εi0
2. Yi1= β0 + β1(Oil and Gas Employment Growth)i1 + Ci + εi1
3. Yi1- Yi0= β0 + β1(Δ Oil and Gas Employment Growth) + εi
 Equation 3 estimates the impact of the difference
in oil and gas employment growth (from the 20012005 time period to the 2005-2011 time period)
on the difference in employment growth between
period 1 and period 0 (2001-2005).
 The county fixed effect is differenced out and thus
there should not be omitted variable bias.
 β1 measures the positive indirect and induced
effects net of any displacement.
32
Change in Employment Growth Rate:
2005-2011 minus 2005-2001
Dependent Variable
Employment Growth
Variable
Rate
t-statistic
Oil and Gas Employment Growth
0.00011
2.10
2000 Log Population
-0.01000
-3.91
2001 Log Average Wage
0.13218
7.43
2000 Percent College
-0.17863
-3.45
2000 Unemployment Rate
-0.00034
-0.31
Industry Controls
Yes
State Fixed Effects
Yes
R2
0.2292
Adj-R2
0.2143
N
3065
Earnings Growth Rate:
2005-2011 minus 2005-2001
Variable
Oil and Gas Employment Growth
2000 Log Population
2001 Log Average Wage
2000 Percent College
2000 Unemployment Rate
Industry Controls
State Fixed Effects
R2
Adj-R2
N
Dependent Variable
Earnings Growth Rate t-statistic
0.00027
-0.02100
0.28632
-0.36531
0.00156
Yes
Yes
0.2494
0.2350
3065
2.90
-4.66
9.13
-4.00
0.79
Growth Rate of Establishments :
2005-2011 minus 2005-2001
Variable
Oil and Gas Employment Growth
2000 Log Population
2001 Log Average Wage
2000 Percent College
2000 Unemployment Rate
Industry Controls
State Fixed Effects
R2
Adj-R2
N
Dependent Variable
Establishments
Growth Rate
t-statistic
0.89
0.00006
-15.33
-0.04986
5.13
0.11602
0.42
0.02788
4.91
0.00699
Yes
Yes
0.4480
0.4373
3065
Tentative Estimated Impact on the
Pennsylvania Marcellus Region
Average county employment in the Pennsylvania Marcellus
Shale region that increased the number of wells drilled (34
counties) was 56,885 in 2005 (total employment in the region
was 1.9 million).
Direct oil and gas employment [NAICS 2111 and 2131] in the
Marcellus drilling region of PA went from 3,911 in 2001 to 4,922
in 2005 to 15,335 in 2011 (an increase in the percent growth of
186%)
Expected percent growth in total employment is 0.038%
(1.86*0.00011) which amounts to 22 (0.00038*56885) workers per
county (735 total workers for the region)
Expected percent growth in total earnings 0.049% or $985,821
($33.5 million for the region)
Expected percent growth in total establishments 0.011% or 0.35
establishments per county (12 for the region)
Top Counties by Percent Oil and Gas
Employment Growth
Rank
County
State
2000
Population
Shale
Oil and Gas
Growth
Oil and Gas
Growth
1
White
AR
67,165
63%
166,413%
1,664
2
Bradford
PA
62,761
100%
105,321%
1,053
3
Lycoming
PA
120,044
93%
75,981%
760
4
Faulkner
AR
86,014
59%
36,452%
1,059
5
Wyoming
PA
28,080
100%
27,209%
272
6
Cleburne
AR
24,046
80%
23,308%
233
7
Barrow
GA
46,144
0%
12,069%
121
8
Conway
AR
20,336
83%
11,827%
118
9
Tioga
PA
41,373
100%
11,598%
116
10
Robertson
TX
16,000
1%
9,142%
102
Top Counties by Oil and Gas
Employment Growth
Rank
County
State
2000
Population
Shale
Oil and Gas
Growth
Oil and Gas
Growth
1
Harris
TX
3,400,578
0%
30%
20,054
2
Williams
ND
19,761
100%
404%
5,663
3
Midland
TX
116,009
0%
56%
5,183
4
Oklahoma
OK
660,448
0%
63%
5,089
5
Ector
TX
121,123
0%
109%
4,206
6
Tarrant
TX
1,446,219
100%
105%
4,145
7
Denver
CO
554,636
100%
72%
2,880
8
Kern
CA
661,645
20%
28%
2,343
9
Dallas
TX
2,218,899
32%
34%
2,188
10
Lafayette
LA
89,974
0%
17%
2,174
Conclusions
 PA Shale natural gas is associated with
significant income effects but modest
employment effects.
 The real question of shale investment is not
job creation, but net benefits vs costs
including pollution costs.
In this question, natural gas should be
compared to coal, the true alternative.
Shale natural gas is lower cost, less carbon,
and like coal has local pollution impacts.
 States should consider higher
severance tax for long-term needs.
Schools, infrastructure, environment.
39
Mark Partridge
Swank Chair in Rural-Urban Policy
Dept. Agricultural, Environmental &
Development Economics
The Ohio State University
Google “Partridge Swank” and you will get my website
(614) 688-4907
([email protected])
40
Individual County Graphs Follow
41
Washington-Cumberland Matched Employment
Pair
120
115
110
105
100
95
90
85
80
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Washington(Mining)
Cumberland(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov.
Greene-Perry Matched Employment Pair
120
115
110
105
100
95
90
85
80
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Greene(Mining)
Perry(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov.
Fayette-Franklin Matched Employment Pair
120
115
110
105
100
95
90
85
80
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Fayette(Mining)
Franklin(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov.
Susquehanna-Carbon Matched
Employment Pair
110
105
100
95
90
85
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Susquehanna(Mining)
Carbon(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov.
Tioga-Union Matched Employment Pair
110
105
100
95
90
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Tioga(Mining)
Union(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
Bradford-Columbia Matched Employment
Pair
110
105
100
95
90
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Bradford(Mining)
Columbia(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
Washington-Cumberland Matched Per Capita Income
Pair
125
120
115
110
105
100
95
90
85
80
75
70
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Washington(Mining)
Cumberland(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
48
Greene-Perry Matched Per Capita Income
Pair
135
130
125
120
115
110
105
100
95
90
85
80
75
70
65
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Greene(Mining)
Perry(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
49
Fayette-Franklin Matched Per Capita Income Pair
125
120
115
110
105
100
95
90
85
80
75
70
65
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Fayette(Mining)
Franklin(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
50
Susquehanna-Carbon Matched Per Capita
Income
Pair
125
120
115
110
105
100
95
90
85
80
75
70
65
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Susquehanna(Mining)
Carbon(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
51
Tioga-Union Matched Per Capita Income
Pair
120
115
110
105
100
95
90
85
80
75
70
65
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Tioga(Mining)
Union(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
52
Bradford-Columbia Matched Per Capita Income
Pair
120
115
110
105
100
95
90
85
80
75
70
65
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Bradford(Mining)
Columbia(Non-Mining)
Source: U.S. Bureau of Economic Analysis, REIS Data, Downloaded Oct. 7, 2011. www.bea.gov
53