Hilde C. Bjørnland, Frode Nordvik, and Maximilian Rohrer, "How
Flexible is US Shale Oil? Evidence from North Dakota", Journal of
Applied Econometrics, Vol. 36, No. 3, 2021, pp. 273-292.

The data used in the article were retrieved from the database of the
North Dakota Industrial Commission (NDIC), Oil and Gas Division.
Additional data were retrieved from the Energy Information
Administration (EIA), from the New York Mercantile Exchange (NYMEX)
retrieved through Datastream, and from the Federal Reserve Bank of St.
Louis (FRED). 

The file "BNR_2020.csv" is a panel dataset consisting of well-level monthly
observations. It is zipped in the file bnr-data.zip.

The following variables are included: 

d.oil:               change in oil produced per well
d.oil.l:             one month lagged d.oil
frack:               indicator variable turning 1 if fracking well
wti.d:               change in the West Texas Intermediate crude oil price
fut3.spread.d:       change in the 3-month future spread
msci.wo.d:           change in the MSCI World index
vix.d:               change in the VIX index
fed.fund.fred.d:     change in the fed fund rate
copper.fred.d:	     change in the copper price
tw.usd.major.fred.d: trade weighted USD rate
well.id:	     individual well index
year:                year
age:		     age of the well
date:		     date, year-month

The analysis was conducted in R. Below you can find the code to
replicate the main table given the supplied data.

# package
require(lfe)

# load data
final.data <- read.csv("BNR_2020.csv")

# Main Regression: Table 2
# Note that the age fixed effects lead to very long execution time. 
# The results are almost identical when omitting the age fixed effects
summary(felm( d.oil ~ 
                d.oil.l*frack
              + msci.wo.d + vix.d + fed.fund.fred.d + copper.fred.d
	      + tw.usd.major.fred.d + wti.d*frack + fut3.spread.d*frack
              | well.id + year + age    
              |0 
              | date + well.id, 
              data = final.data))