Citation

Material Information

Title:
Socioeconomic Dimensions of Oil & Gas Development in Colorado
Creator:
Hegg, Alex
Physical Description:
Presentation Video

Notes

Abstract:
Video presentation of Data 2 Policy project submission for Fall 2020 symposium
Acquisition:
Collected for Auraria Institutional Repository by the Self-Submittal tool. Submitted by Alex Hegg.
Publication Status:
Unpublished

Record Information

Source Institution:
Auraria Institutional Repository
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.

Auraria Membership

Aggregations:
Auraria Library
Added automatically

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

In a previous D2P submission, we examined whether a spatial relationship exists between the placement of oil and gas well drilling sites and asthma emergency room visits in Colorado. As an extension of this research, we now wish to investigate whether oil and gas wells are more likely to be placed in and around various communities with certain economic and demographic characteristics. We will b e utilizing data available from the Colorado Department of Public Health and Environment and the Colorado Oil and Gas Conservation Commission websites as well as from the American Community Survey for the years 2013 to 2017. We will be analyzing this data in a Bayesian G eneralized Linear Model framework.



PAGE 1

Socioeconomic Dimensions of Oil & Gas Development in ColoradoA Bayesian AnalysisBy: Alex Heg g7393 – Bayesian Statistics Dr. Joshua French

PAGE 2

Introduction and Backgroun dPrevious Project: Analysis of Respiratory Illness and Oil & Gas Drilling Sites•Positive relationship between asthma ER visits and number oil and gas drilling sites in a census tract•If oil and gas drilling may be contributing to respiratory illness, which Colorado communities may be affected?•We consider relationships between the rate of oil & gas wells drilled during 20132017 and two census tract characteristics: educational attainment and median household income

PAGE 3

Model SpecificationBayesian Normal Error Regression using a Markov Chain Monte Carlo (MCMC) method•Bayesian: Specify Probability Distributions for both Data and Parameters•Prior Distributions chosen to be non-informative•MCMC Method:•Samples from Target Distribution (our normal error model)•Samples create a Markov Chain•Multiple chains are used in analysis Two census tract explanatory variables:•Median household i ncome•Percentage of population age 25+ with HS Diploma/Equivalent

PAGE 4

Results cont. •Beta1 is almost entirely above zero, but the effect is very small (centered around $0.09)•Beta2 also mostly positive, centered around 1.75%•Chains overlap almost entirely, suggesting convergence (good).

PAGE 5

Results cont.•Interpretation:•Though a very small effect, census tracts with higher median household income are associated with slightly more well drill sites per •Conditioned upon the data, an increase in high school diploma attainment rate by around 1.75% is associated with an increase of one oil well per in a census tract.•These results seem unexpected. This is most likely due to poor model specification (see below)•Policy Suggestions:•Though we cannot conclude on causality, these relationships may inform possible policy decisions:•We may wish to consider the public health risks associated with drilling as noted in the previous project with respect to higher income and/or educational attainment and consider either:•Providing mitigation, if we wish to reduce the public health risk, through increased public health spending•Providing economic incentives since drilling may be having a positive impact on the economies of the tracts in which it occurs.

PAGE 6

Final Thoughts:•Model Mis -Specified•As noted above, the results seem unexpected. We might expect drilling to occur in lower income areas since it is typically viewed as “undesirable” and, as such, households with higher incomes have more ability to choose to live elsewhere.•We need to consider that you can’t drill successfully anywhere either.•If we examine our data, we see in the below histogram that there are a very large number of tracts with zero wells (1,143 of 1,249 tracts have zero wells):•We should consider using a Negative Binomial model which allows for a higher number of zeros in the data.

PAGE 7

Bilbiography :1.Colorado Asthma Data: Colorado Department of Public Health and Environment: https://data cdphe.opendata.arcgis.com/datasets/a176548521c546f0b9be512197d7d8f4_1/data?geometry= 126.227%2C32.804%2C84.875%2C44.746 2.Colorado Oil & Gas Conservation Commission: Open Data https://data cdphe.opendata.arcgis.com/datasets/a176548521c546f0b9be512197d7d8f4_1/data?geometry= 126.227%2C32.804%2C84.875%2C44.746 3.American Community Survey data: 2013 2017. Obtained via tidycensus package in R. https://cran.r project.org/web/packages/tidycensus/index.html