When the 28th legislature narrowly passed SB21 in 2013, I was working as a policy analyst for the Department of Revenue. I still remember the late night committee hearings and the frantic preparation of fiscal notes.
At the end of the day, the bill was named the “More Alaska Production” act. While the name didn’t stick, it does describe what it was trying to achieve.
In the years since SB21 passed, we have seen an increase in production. So, it is natural for supporters to claim “mission accomplished.” But as I often remind people, correlation doesn’t necessarily imply causation.
As diligent scientists and nonpartisan analysts, we need to test every claim against the data. Let’s explore where the production increased, and see if giving credit to the tax bill makes sense.
Bottom Line Up Front
From what I can tell, we can give some credit to SB21, but most of the credit should go to the operators.
About 10% of the production increases can be attributed to SB21. 50% of the increases appear to be coincidental (production that was already on the way before SB21 passed). And the rest is due to stellar management in the legacy fields.
Is Production Actually Up?
As is often the case, it is good to understand the hypothesis before jumping in to testing it. Here is the production data from AOGCC (I explain why I don’t use DOR data here).
Let’s transform this to annual decline rates to see the point more clearly.
Since production began declining in FY89 until SB21 passed 24 years later, the North Slope averaged a decline rate of a little more than 5% per year. Only once did it post a production increase (when Alpine began producing).
In the 5 years since SB21 passed, we have seen 2 years of increases, 2 years of smaller than normal declines, and 1 greater than normal decline. The average decline over those 5 years is less than 1%
That’s a pretty good indication that something is going on. But it’s not enough to resolve the question. Let’s dig a little deeper.
Hypothesis Testing
Because of the variability in the numbers, we need to make sure these changes are not just a coincidence. It’s possible that the numbers we are seeing are a product of random chance.
For example, maybe production stopped for a month in one year. That would artificially reduce the rate to which we compare the next year. This would end up looking like a smaller decline, but it would be due to completely random factors.
To test for this, I smoothed out the historical production rates prior to SB21 to establish a base production level. The deviations from this level establish what “normal randomness” looks like. Then, I used that randomness to project the 95% confidence interval forward for the years since SB21 took effect.
When I plot the actual production for those years (the purple line at the end), we can see that production rates fall well within the range that could be considered random in 2014 and 2015. In 2016, production is at the top end of the range, but could still be due to randomness. But in 2017 and 2018, production exceeds that range.
Reject the Null Hypothesis
The data show a literal year-over-year increase of 14,436 barrels per day in FY16, and then another increase of 11,535 barrels per day in FY17. When the numbers are official, they will show a decrease of around 7,800 barrels per day in FY18.
But if we account for natural decline, we get a more accurate picture of what is going on. The FY17 increase is clearly a result of something other than randomness. And while the FY18 numbers will show a decline, that production level is still higher than we should expect.
Therefore, we can confirm what I said before. Production did in fact increase after SB21. But, we cannot yet say if it increased because of SB21.
How Much Additional Oil?
As we dig into the data, it’s a good idea to know what we are looking for. To calculate how much oil we are trying to find, I conducted a decline curve analysis on the production data prior to 2014.
Using that output, I added up all the oil that we can estimate would have been produced in 2014-2018 if the trend continued. That number comes out to 857,746,786 barrels of oil. Actual production during that time frame adds up to 953,847,757 barrels of oil.
So, we are looking to explain about 96 million barrels of additional oil. Breaking it down by year, we are trying to explain extra production of these amounts:
FY | Difference from Trend |
2014 | 9,656,440 |
2015 | 3,202,714 |
2016 | 17,472,289 |
2017 | 30,232,677 |
2018 | 35,488,981 |
Known New Developments
There are a few projects that have begun producing oil since SB21 passed. Let’s start by understanding them.
Point Thomson
The Point Thomson Unit began delivering condensate to TAPS in April of 2016. Through June of 2018, the unit has delivered just short of 3 million barrels.
FY | Pt Thomson Oil Production (Barrels of Oil) |
2014 | 0 |
2015 | 0 |
2016 | 76,651 |
2017 | 1,141,495 |
2018 | 1,648,561 |
This production comes from the “Initial Production System” that the owners agreed to install as part of the Point Thomson Settlement Agreement signed in 2012.
Can this oil be attributed to SB21?
No. The commitment to develop this project happened prior to SB21.
Shark’s Tooth
In the South-West corner of the Kuparuk Unit, there is a small project known as Shark’s Tooth. This is a new drill site, often called by its descriptor “DS-2S”. The Petroleum News reported that the prospect was discovered in 2012.
The drill site has 8 producing oil wells, which reached a maximum production rate of 10,021 barrels per day in November of 2016. The AOGCC data shows the drill site producing 7,242 barrels per day so far in 2018. Total production to date adds up to 7 million barrels of oil.
FY | Barrels of Oil Produced from DS-2S |
2014 | 0 |
2015 | 0 |
2016 | 997,217 |
2017 | 3,122,438 |
2018 | 2,763,146 |
Can this oil be attributed to SB21?
Yes. Although this prospect was discovered prior to the bill being passed, it is likely that it got sanctioned as a direct result of SB21.
During the SB21 debates, this project was specifically mentioned as one of the economically marginal projects that could be made viable by passing the bill. It was sanctioned in May of 2013, just a month after the bill passed.
CD-5
ConocoPhillips starting pumping oil from a new drill pad just west of the Colville River in September of 2015 (FY16 began in July 2015). According to AOGCC data, there are currently 14 wells producing oil from the drill pad known as CD-5.
The project is exceeding expectations. While the original estimated peak production was 16,000 barrels per day, the pad produced 36,000 barrels of oil per day so far in 2018. And, the company reports that they intend to expand the operations of this project again next year, adding another 10 well slots.
Total production so far has added up to about 26 million barrels of oil.
FY | Barrels of Oil Produced From CD-5 |
2014 | 0 |
2015 | 0 |
2016 | 5,018,354 |
2017 | 8,930,085 |
2018 | 12,416,376 |
Can this oil be attributed to SB21?
No. According to a webpage presumably sponsored by ConocoPhillips, the project was sanctioned in 2012. Since that occurred prior to SB21 passing, it is difficult to assign credit to the bill.
1H NEWS
ConocoPhillips drilled three new wells on drill pad 1H in North-East West Sak (NEWS) during FY18. A fourth well is expected to be completed within the next few months.
To date, wells 1H-101, 1H-102, and 1H-103 have produced 1,631,132 barrels of oil.
Can this oil be attributed to SB21?
Yes. The company sanctioned the project in 2015. In a press release, it explicitly states that this project was able to move forward because of the tax reform. While that could be PR spin, it is the only evidence we have.
Total From New Developments
Total production from these 4 projects add up to nearly 38 million barrels of oil. Most of which was produced in FY17 and FY18.
Of that production, about 9 million barrels can be credited to SB21.
Retesting the Hypothesis
If we pull the production from these projects out of the total, we can see what happened to the decline curve without them.
As you can see, the production levels since SB21 took effect fall within the range of randomness once we adjust for the new projects.
However, there is a technical challenge we need to address. Because we were using total North Slope production data to establish the baseline, using anything but total production in the analysis is a statistical no-no.
The baseline data includes additions of new fields (like Alpine, North Star, Oooguruk, and Nikaitchuq) at various times over the years. So, the projected range should capture future additions as well. By pulling them out, we would violate statistical protocol.
To do this right, we need to look at unit level data.
Oooguruk
When SB21 passed, the Ooogurk Unit was producing about 7,000 barrels per day from 17 wells. In FY18, the unit produced at a rate of 11,647 barrels per day from 21 wells. I count 9 wells drilled after 2013, and 5 wells shut-in.
I think it is safe to attribute the increased production to successful drilling and fracking activities.
Can this oil be attributed to SB21?
It is unlikely that SB21 tipped the scale from not drilling out the field to continuing the drill plan.
However, Caelus is a small company with limited capital. So, it is possible that the change in tax policy is what gave their investors the confidence to proceed.
Nikaitchuq
Nikaitchuq is a relatively young field. Production peaked in 2016 at 25,000 barrels per day. In 2013, there were 20 oil-producing wells flowing at a rate of just over 10,000 barrels per day.
Production from new wells accounts for about 9 million barrels of additional oil since 2013.
The major increase in production appears to be coming from those other 20 wells. 19 out of the 20 wells that were producing in FY13, showed a production increase in FY14.
This is likely a natural consequence of a principle in fluid dynamics called transient flow. The field was simply too young to start declining when SB21 passed.
Can this oil be attributed to SB21?
Not likely. The production increases at Nikaitchuq are probably a result of physics, rather than economics.
Kuparuk River Oil
If the Kuparuk field would have continued its normal decline from 2014-2018, it should have fallen within the red and yellow lines in the graph above. As you can see, it didn’t.
We already know that DS-2S is part of that increase. But that only accounts for around 8,000 barrels per day. There are at least 12,000 more barrels per day in FY18 that are unaccounted for.
By digging into the pad level detail, I pinpointed the additional production as coming mostly from these 10 pads: 1D, 1G, 1Y, 2G, 2M, 2T, 2U, 2Z, 3C, and 3M. There were also 2 pads that declined much more quickly than the trend: 1C and 3S.
Most of the production increases can be attributed to 16 wells from within those pads. It looks like the rest of the barrels are spread out over a large number of wells that are producing better (or more often) than history would suggest.
My guess is that if you could see the work-over plan, you would find these wells on the list at some point in the last few years.
If you’re interested in which wells I pinpointed, they are named: 1E-33, 1E-18, 1A-20, 1D-01, 1D-04A, 1D-08, 1D-12, 1D-36, 1D-37, 1D-41, 1Y-30, 2U-13, 2X-13, 3H-26, 3M-24, and 3Q-22.
Can this oil be attributed to SB21?
Doubtful. In-fill drilling, shut-ins, and work-overs are routine parts of managing an oil field. As oil production declines, you get more “bang for your buck” when you perform these operations.
For example, if you increase a well’s production from 500 to 600 barrels per day, it’s a 20% increase. But if you take a well from 300 to 600 barrels per day, that’s a 100% increase.
As production declines, it get’s easier to move the needle in terms of decline rates. The result is a decreasing decline rate over time when you aggregate production across the field (this is sometimes called hyperbolic decline).
This increase is more likely to be related to management of a mature asset, rather than improved economics from a tax change.
However, taxes do play a part in every investment decision. So it’s not completely impossible that the tax bill played at least some part.
Prudhoe Bay
Posting flat production rates is a huge accomplishment for an oil field, especially for one of the largest fields in the world.
So, how was BP able to stem the decline at Prudhoe Bay? The logical conclusion would be that they increased drilling activity.
But we know that Prudhoe Bay cut three rigs starting in 2016, running only two rigs for the following years. As a result, you can clearly see a reduction in drilling activity.
I count 195 new wells being drilled since 2013. At the same time, I count 224 wells that were producing in 2013 and showed zero production in 2018.
So the increase is not coming from increased drilling activity.
The only way BP could have replaced rate reductions without drilling new wells is through stellar reservoir and facility management.
Squeezing Old Wells
To me, it looks like BP put some effort into squeezing production from old wells during this time. I count 67 wells that were shut-in during FY13, but posted meaningful production in FY18.
Those wells added nearly 22 million barrels of production over the period.
Increased Uptime
Another possibility is “increase operational efficiency” and “good deferral management.” Basically this means that they were able to keep the wells producing for more days of the year.
We can test that.
I pulled up all 1,898 Prudhoe Bay well records from AOGCC. I adjusted the well count by removing any wells that posted zero production days in each year. Then, I multiplied the well count by 365 days to reach the maximum possible production days for the field.
Dividing the posted production days by the maximum possible production days, I get what I’ll call the “uptime.” While the company will give you different numbers, it probably has the same shape.
Here is what it looks like.
I believe this indicates a change in field management by BP. Before about 2008, I think they would shut-in wells when they became marginal. This meant that only wells that were not having production problems would stay online. That pushed the average production days up.
Starting in about 2008 (when oil prices spiked and the State moved to the ACES tax system), they appear to have become more willing to keep marginal wells online. This drove down the average uptime, but increased total production. As a result, the decline rate of the field fell from nearly 10% per year prior to 2009, down to about 4% since.
Missing Maintenance
But, there’s something else that is interesting here. Take a look at 2017 in the chart above. There’s an appreciable increase in uptime that year. And that year corresponds to an increase in total production for the field.
Let’s look closer.
You should notice the dips in the graph above. Every summer, when the weather warms up enough on the North Slope, the facilities undergo annual maintenance. That is what you are seeing here. In June, July, and August, there is a disruption in production while the facilities are worked on.
What I want to draw your attention to is the 2016 summer. There is a noticeable lack of production disruption (the dip isn’t as far down as it should be). For whatever reason, the maintenance cycle that year was not very deep. And that resulted in a higher uptime than the years before and after.
Because the fiscal year starts in July, that summer falls into FY17. This smaller than normal disruption added about 7,000 barrels per day of production to that year’s average.
But notice that things returned to normal after that year. So, it appears that this was more of a one-time occurrence than something we can count on continuing.
Can this oil be attributed to SB21?
I don’t think the change in reservoir management can be attributed to SB21, because it appears the change in policy happened back around 2008 and continued after SB21 passed.
But, when oil prices tanked in 2015, they may have been motivated to boost rate replacement from old assets as their investment capital for new projects dried up.
You can do that by “popping” old wells to get a short-term boost in rate.
Another possibility is that they were motivated to show an increase in production in order to validate the tax reform. If so, then putting some effort into a bigger work-over plan and deferring maintenance would do it (although laying down rigs would be counterproductive).
The problem is that if this was accelerated production and deferred maintenance, rather than true incremental production, we will see a faster decline rate in the future. Time will tell.
Of course, all of this is just speculation. I’m sure the company representatives could give as a hundred other good reasons.
All I know is that holding Prudhoe Bay decline to zero is a huge reason we were able to notice that production increased elsewhere.
Has SB21 Resulted in Increased Production?
Yes, but only about 10% of the increases we see can be attributed to SB21.
The majority of the production increases appear to be coming from good field management at a lower base rate.
The second biggest factor is coincidental timing. CD-5 and Point Thomson were sanctioned before SB21 passed, but came online shortly afterward. And Nikaitchuq hadn’t reach the point of decline yet.
There are about 9 million barrels that I would say are directly attributable to SB21 so far, and several million more that may be loosely tied to the tax reform.
But, since I can find no fields that did worse under SB21 than we would have expected, the data does support the conclusion: SB21 did result in a production increase.
Future Developments
Starting later this year, we will start to see even more new projects enter production. In fact, FY20 should show a very substantial production increase as GMT-1, Mustang, and Moose Pad come online. In the years that follow, we will see Liberty, GMT-2, Willow, and Pikka adding significant rates.
When those barrels begin to flow, it will be tempting to ask this question again. But as we move further away from the changes that were made, it will be harder to know what would have happened in a world without SB21.
What we do know is that exploration activity increased after SB21 passed. And we know that some of that activity resulted in very promising discoveries. But the ability to make a data driven conclusion about the counterfactual world is weak at best.
So, you will have to make up your own mind about whether or not those new fields are dominos falling from SB21’s passage, or if they are things that would have occurred no matter what.
Conclusion
The data supports the idea that SB21 lead to some new production. And given the price path that unfolded, SB21 certainly did not make the State’s fiscal situation worse.
But, SB21 is not responsible for all of the production increases we are seeing. And whether or not it is responsible for the recent discoveries and subsequent excitement is one that cannot be answered objectively.
Does this mean we should take another look at oil taxes? That’s a conversation for another day.