The Department of Revenue puts out a forecast of state revenues each December, which serves as a companion to the budget. The latest numbers are a few months old now, and an update should be released next month. However, the Spring update usually focuses more on adjusting the budget year than the long-term outlook.
The numbers serve as a guide from which budget decisions can be made. But, like any forecast, they are out of date as soon as something approximated in the forecast can be replaced with an observation.
Forecasting oil prices is an impossible task, so it is hard to blame the economists doing the analysis for the large errors that inevitably occur. The recent move to using futures data rather than subjective assessments is a welcome improvement in transparency, regardless of how little it changes accuracy.
While the document is a staple of the budgeting world, it receives weight beyond what it can carry. Almost any discussion about the state’s financial future is built from these revenue assumptions. As I’ve pointed out before, those assumptions are certain to be wrong and the conclusions drawn from them are formed on a faulty foundation.
Risk-Weighted Forecast
The Department rightfully uses a risk-weighted average when providing numbers from which to build a budget. This “expected value” approach is the best way to capture the fact that the actual numbers may be higher or lower than the forecast amount.
Over several years, this approach of minimizing the average error will result in better long-term planning. But, only if the legislature uses the years of higher than expected revenues to backfill the years of poor outcomes.
That is, a risk-weighted forecast requires decision-makers to stick to the forecast revenues rather than the actual revenues in order for it to work. Unfortunately, that doesn’t tend to happen. And, budgeting to a declining annual cash flow is a sure way to fail in the long term.
Still, this is a useful number and is worth calculating. When I run my own numbers, I get something close, although slightly less pessimistic, to what the Department projects.
By the way, I did not include Permanent Fund earnings in these graphs. I’ll come back to this topic at a later date. The short story is that the State has a huge inconsistency in the way we treat those numbers.
Why are these numbers different than DOR?
These numbers are a little higher based on two key assumptions.
First, I’ve put more faith in the oil industry’s ability to develop our resources. My production forecast assumes that Hilcorp will have success in Prudhoe Bay similar to what they are accomplished in every other asset they have purchased in Alaska.
I’ve also put a higher likelihood on known discoveries coming online in scenarios that the oil price justifies them. I applied explicit delay risks and significant resource uncertainty, but added very few scenarios in which a delineated prospect is abandoned for non-economic reasons.
I’ve also built in a small chance that prospects like Icewine, Yukon Gold, Horseshoe, Cairn, and Peregrine will have exploration success that leads to future development.
Second, I used a Brownian Motion Mean Reversion technique to capture future price movements.
Futures markets tend to anchor to whatever the current prices are. So, long-term futures can be significantly moved by short-term factors (is there any reason to believe that the Coronavirus will still be depressing prices in a year, much less a decade?). That anchoring leads to large and unnecessary movements in long-term forecasts, usually leading to results that swing back at forth as the market moves through commodity cycles.
However, because the BMMR is too much of a “black-box,” I don’t recommend the state use such an approach (something low cost and transparent is better for their purposes). Plus, an approach like this is worthless in a deterministic forecast – like the one the department uses.
I use this approach because it does a much better job of simulating volatile price paths (which are critical in determining a mean value) while minimizing long-term revisions.
Scenario Planning
While using a risk-weighted value is the right tool for budgeting, it is the wrong tool for scenario planning. Take this example:
I am going to flip a coin. You must predict the outcome. Assume a “heads” is a 1 and a “tails” is a 0. Using a risk-weighted approach, you should call 0.5 heads as the predicted outcome. That way you will be wrong by 0.5 either way.
If you predict a 1 or a 0, you will either be exactly right or completely wrong. If you play this game enough, your error will average out to zero using the risk-weighted approach. Conversely, it’s theoretically possible for you to have an error of 100 or -100 after calling head or tails 100 times.
The risk-weighted forecast ensures you are half correct every time. However, this approach is never exactly right. The coin will never fall on 0.5 heads.
That brings us back to our production forecast. Either these new fields are going to come online, or they aren’t. Either the fiscal climate will support development or it won’t. So, the middle case that assigns partial weight to each possible outcome is unlikely to occur. We need to understand the outcomes in those conditional states, and have a plan in place when one or the other reveals itself.
Success Case
In the success case, the long-term inflation-adjusted average oil price represents where prices will remain going forward. There will be peaks and troughs, but they will average out to around $70 in 2020 dollars.
This price supports the development of several projects, including Pikka and Willow. It also supports developments inside the core fields, including increased activity by Hilcorp in Prudhoe Bay and Milne Point, and by ConocoPhillips in Kuparuk and Colville River.
This is what the production profile looks like when you take economic uncertainty off the table.
With that production profile, a $70 mean real oil price, and under the current tax law, the State’s revenue picture looks like this:
As shown, there is an increase in revenues that helps defer our immediate problem. But, that problem resurfaces when those major developments begin to decline.
Failure Case
And, if we land on tails, the situation is a lot worse than we think. Here is how the production picture plays out in a world that averages $40 oil moving forward.
In this low price environment, Prudhoe Bay declines much more quickly. Badami, Endicott, NorthStar, Oooguruk, and Point Thomson are all abandoned within 10 years. The Willow prospect still works out, but only because the development costs can migrate to other fields under the current tax law. The Pikka project does not get developed in this scenario, due to the slow recovery of capital.
Still, production stays relatively flat for the next 10 years, as new developments offset natural declines in the other fields. Then the issue accelerates into terminal decline. By the mid-2040s, the cost of production and transportation no longer justify continued production.
This price environment locks us into the minimum tax. As a result, revenues stay relatively flat at around $1.5 billion. In this world, there are no reinforcements over the hill. The budget problem is painful and perpetual. And the oil stops flowing in just 25 years.
Conclusion
When we arrive in 2030, we will probably be living in a world that resembles one of these two scenarios. It is unclear which right now, but we should have a better vantage point within the next couple of years. So, we need to have contingency plans in place that help us prepare for either eventuality.
As the future unfolds, we are likely to realize that the people in 2020 either overreacted to a temporary downturn in revenues, or that they failed to realize the gravity of the situation. I tend to believe the success case is more likely. And in that scenario, we get one more chance to truly set the state up for long-term success. So long as we don’t blow that money on things we can only temporarily afford, we probably don’t have to take drastic actions today.
Regardless, we need to start laying the groundwork now if we are going to manage how things end up. In either case, we are heading toward an iceberg – we just don’t know how far in front of us it is. If we wait until we can see it, we will lack the ability to steer around it. The only question is how hard we need to turn the rudder today. I advise a series of small turns over the next three years as what’s on the horizon gets clearer.
But, if that failure leg is where our future is heading, we have a bigger problem than we think. In that case, we have already squandered our opportunity to sail into the sunset. Consequently, we must decide between enjoying the time we have left, or preparing a life raft for our children.