I say it’s an unfortunate story for a couple of reasons.
First, it uses some pretty sensational language. The headline for the original article is, “In North Dakota’s Bakken oil boom, there will be blood,” a reference to the Daniel Day-Lewis movie (which, in turn, was based on the novel Oil! by Upton Sinclair). The article also refers to the Bakken oil boom as “a serial killer.”
Oil drilling is a very dangerous occupation – it ranks just behind roofers and just ahead of agriculture on the list of America’s most dangerous jobs – and worker safety is not an issue to be taken lightly. But the use of this sort of language isn’t exactly conducive to a level-headed review of the situation.
Second, the use of fraught language (and what it might tell us about the biases of the article’s authors) aside, there is a fundamental math problem at the center of the article’s argument which undermines its entire premise.
Cutting through the anecdotes and other reporting, this line from the article is the supporting argument for the whole oil-boom-as-serial-killer narrative: “On average, someone dies about every six weeks from an accident in the Bakken – at least 74 since 2006, according to an analysis by Reveal, the first comprehensive accounting of such deaths using data obtained from Canadian and U.S. regulators.”
The authors also provided us with a year-by-year graph of these fatalities (see the graph on the right which I took from the article), but they’re not giving us any context. They’re simply showing us the raw headcount of fatalities in the Bakken region (not just North Dakota but also including data from Saskatchewan, Manitoba, and Montana).
The context here is pretty important, because it paints a picture that is the exact opposite of what the article argues. Yes, worker deaths have increased over those years, but so has oil activity. That context matters. For instance when population increases in a given area so does the number of crimes, which is why we use crime rates (the number of crimes committed in the context of population increases) to describe trends and inform public policy.
To put it another way, a city or county is not necessarily more dangerous simply because there are more crimes committed there. That’s because, depending on population changes, the crime rate may actually be unchanged or even down.a
The same is true with safety in oil industry jobs. If oil activity increases, meaning more workers on more rigs, you’re going to get more fatalities. Nobody is happy about it, but it is what it is.
The CIR/Politico story gives us raw fatality numbers, but not the context in the increase in the number of oil workers. That is a fatal flaw to their story.
Let’s do the math on Bakken oil patch fatalities and let’s use the CIR/Politico story numbers and pretend, for the sake of simplification, as though they all happened in North Dakota (even though some happened in Montana and Canada).
In 2006 there were 6 fatalities in the Bakken oil fields and 351 spudded wells in North Dakota per the Department of Mineral Resources for a rate of 0.0171 fatalities per well.
In 2014 there was a twice the number of fatalities as in 2006, but almost 7.5 times more wells (2,624 again per DMR) for a 0.0045 fatalities per well rate.
That’s a 73 percent decline in the fatalities-per-oil-well rate. And the fatality rate probably declined even further because, as I already explained, we are using the CIR/Politico Bakken fatality numbers from Montana, North Dakota, and Canada but only the the well count from North Dakota.
On the left you can see this data represented in graphical form, with fatality data from the CIR/Politico article and well data from the DMR. The top graph shows us the correlation between oil activity and fatalities. More wells spudded has meant more deaths. But the bottom graph shows us why the context matters. The rate of oil worker deaths per well is way, way down from 2006 and in recent years seems to be on a continued improvement trend.
Again, worker safety is nothing to trivialize. This is an important topic to discuss. But it should be discussed honestly. That this context and data was missing from the sensational CIR/Politico story is telling.