Labor Market Dynamics and The Unconventional Natural Gas Boom: Evidence from the Marcellus Region Tim Komarek, Old Dominion University Prepared for the USAEE Resource Boom and Labor Dynamics USAEE 2015
Source: Energy Information Administration 6,636 increase Natural Gas Production in United States Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Shale Production in United States Source: Energy Information Administration 9,299 increase Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Local Labor Market Effects -Growing literature examining the jobs and income resulting from the fracking boom -Short-run benefits through the multiplier process -Long-run implications of the “resource curse” -Bidding up factor prices -Crowding-out in the traded sector (Dutch Disease) Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Objectives -Econometrically estimate the impact of resource extraction activity on the labor market -Utilize the timing of fracking activity to trace out the dynamics of the labor market response -Exploit a natural experiment from differing policies related to fracking between NY and PA, OH and WV in the Marcellus region Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Exploit a natural experiment in the Marcellus region -New York ban on unconventional gas drilling -Pennsylvania, Ohio, and West Virginia that permitted fracking -Led to dramatically different natural gas production in each state -Comparing counties in the Marcellus region before and after the fracking boom Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Shale Formations in the Northeastern United States Source: Energy Information Administration Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Total County Level Unconventional Wells Spudded Source: Author’s calculations from Drillinginfo.com Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Timing of the Fracking Boom (50 + wells): PA, OH, WV Source: Author’s calculations from Drillinginfo.com Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Exploit a natural experiment in the Marcellus region -New York ban on unconventional gas drilling -Pennsylvania, Ohio, and West Virginia that permitted fracking -Led to dramatically different natural gas production in each state -Comparing counties in the Marcellus region before and after the fracking boom Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Natural Gas Production For New York and Pennsylvania Source: Author’s calculation from ERS County-level Oil and Gas Production Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
County level data for NY, PA, OH and WV 2001 – ,026 county-years Labor market data from BEA Regional Economic Accounts -Total employment, population, wages and wages per job -Employment and wages by sector -Manufacturing, construction, transportation retail trade, accommodations Horizontal wells from Drillinginfo.com Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Unconventional wells from Drillinginfo.com -# of bore holes drilled per year in unconventional formations -Production data is highly variable from year to year due to measurement and reporting challenges -Wells drilled more closely tied to economic activity related to fracking Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Distribution of Unconventional Natural Gas Wells Source: Author’s calculations from Drillinginfo.com Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results 50
Y cst = Σ i=0 β -i HighFracking cst + α c + W st + tZ c + ε ct Y cst = log of labor market variable HighFracking cst = 1 if wells >50 LowFracking cst = 1 if 0 < wells < 50 Unweighted OLS with standard errors clustered by county Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Economic theory suggests potentially different effects over time Estimates show a positive employment and wage effect that increases over the course of the boom period See wages being bid up in manufacturing, construction and transportation sectors No evidence of crowding-out in the traded sector (manufacturing) Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Resource Boom and Labor Dynamics USAEE 2015
Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
The current shale gas boom differs from the energy boom-bust cycle of the 1970s and 1980s -Based on technological innovations (hydraulic fracturing and horizontal drilling) -Extracts gas from shale formations -EIA estimates the U.S. reserves hold 70 years of natural gas supply Local impacts on communities “boomtowns” where extraction takes place Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results
Resource Boom and Labor Dynamics USAEE 2015 Motivation Identification Strategy Data Empirical Model Results VARIABLES Total Employment Total Employment Earnings per Worker Earnings per Worker High Fracking *** *** ** ** ( )(0.0107)(0.0165)(0.0177) Low Fracking ( )( ) ( ) ( ) Observations1,9113,0161,9113,016 Control GroupNYAll CountiesNYAll Counties County FEYes State-Time FE Yes County*Year Trend Yes