When is a trend not a trend? Decline in the Legacy Production of the Permian

A longer version of my question….

The EIA kindly publishes detailed production data from the US. The monthly data is generally considered the most reliable. Here I am just looking at the Permian subset.

It is presented as total production – with the increment month-on-month being the net of a (big) decline in all previous production (“Legacy”) and a usually bigger increase in new production (although this recent example is of a small net decrease)

Rinse and repeat, month after month and the overall production increases.

However, there seem to be two unknowns:

  1. The actual decline of the Legacy and,
  2. The real “new” production

As far as I can see, the link between the two is the rig-count, and the rather odd metric of “Production per Rig”. As we all know, drilling rigs don’t produce. However this looks like a placeholder which represents how much production is brought on, on average, by each rig… so you would expect it to be the “new production” divided by the number of active rigs… and it sort of is. This won’t be exact as the rig count is changing daily, but generally it has been close.

For example: August 2019, the incremental production from July was 4,310,660 bopd rising to 4,420,811 bopd for a net increase of 110,151 bopd. The stated Legacy Decline was -247,022 bopd, which implies a total addition in the month of August of 357,172 bopd.

The product of Rigs (438) and Production per Rig (830) is 363,450 bopd, which is pretty close to the calculated month-on-month addition. So far so good.

The Legacy Production

Is the decline in the legacy production a calculation based on reversing the above (i.e. you start with known month-on-month incremental production and you estimate the total “new” production from the rig count)? Or is it somehow the reverse where the Production per Rig is an output? Or something else I have missed? Somewhere here there seem to be too many unknowns.

Why am I asking this?

Well the last time I looked at this data there was a logical progression of the decline in legacy production gently increasing as the total production accelerated. This makes sense: you add significant amounts of steeply declining production each month and the average decline increases slightly.

However, since February-March 2019, something odd is happening. The amount of decline in the huge production wedge is apparently decreasing (ignore the big “V” in Covid months, but look at the reversed trend from 1Q19 to 1Q20).

I have no idea how to interpret this. It is conceivable that the legacy production is actually declining as per the previous trend, but the calculation method is missing something. I immediately thought of DUCs, which could contribute additional production but not show up in the (Number of Rigs * Production per Rig). If these were contributing production volumes it could explain an underestimation of the legacy decline. There is an element of this with some months showing a net reduction in DUCs during this period, but it is not clear-cut by any means.

Any ideas?