What Capital Discipline?

US rig count is actually higher than expected

In the years following the financial crisis, the diffusion of innovations in horizontal drilling and hydraulic fracturing of shale formations unlocked billions of barrels of previously inaccessible oil deposits to commercial exploitation. What followed was a classic investment cycle: over-investment and over-expansion of capacity quickly led to a price collapse, followed by a prolonged bust.

Proven US oil reserves grew from around 20.7bn barrels (bbl) in 2009 to over 36bn bbl by 2014 and 44bn bbl by 2019. During the go-go boom years of the early 2010s, the number of rigs expanded at an unprecedented rate. US rig count grew between fivefold and sixfold in just five years: from 276 in 2009 to 1,527 in 2014.

The price impact was rapid. WTI collapsed from over $100 per bbl to around $50 in 2015-2016. This was well below the break-even price for most US producers. No wonder then that rig count immediately collapsed as well. However, by this point, the price of crude was no longer being driven by the shale boom. Instead, it was being manipulated by the Saudis, who were waging a price war to drive high-cost American producers out of the market.

In the event, the price of crude remained subdued for years. It was not until the end of 2021, with inflation gathering force across the world economy, that WTI crossed the $80 mark — an important threshold, given that it is the breakeven price for most producers in the world market.

As inflation began gathering pace and oil prices continued to rise, an idea began to circulate that something called “capital discipline” had made US oil production less elastic and less responsive to the oil price than before. The salience of the charge grew dramatically after the Russian attack on Ukraine and the Western sanctions that it triggered. The White House began to charge that US oil companies were not doing their patriotic duty by opening the hose. If you read the financial press, there was no hiding from it.

Like its cognates, capital strike and bond market vigilantes, the notion of capital discipline was woolly-headed and not based on any evidence beyond the claims of market commentators, sell-side analysts and the like. The prima facie case for it was tenuous: even if some investor cabal agreed to not buy instruments issued by shale companies, how could they stop other investors from doing the same? What penalty could they impose on the many thousands of sources of funds that can be tapped by oil & gas companies?

At any rate, the notion that investors had abandoned oil & gas was absurd. Investors have been, rather unsurprisingly, bidding up the shares of oil & gas companies. Indeed, while the SPDR has returned 0.6% since the beginning of 2021, energy stocks have returned 102.0%. So much for capital discipline.

Investor discipline is hard to impose without a coercive central authority. The closest thing that comes to mind is the “self-sanctioning” phenomenon we have seen after the imposition of Western sanctions. But that’s largely firms being unwilling to tolerate the risk of getting caught in the crosshairs of the US economic warfare state. No plausible mechanism has been suggested for how a large number of independent investors could possibly coordinate to impose discipline on a sector.

Yes, financial conditions have tightened as the Fed has hiked. For instance, high-yield spreads have widened from roughly 3% to 5% since Jan 2021. But that is true for all sectors. I have yet to see a single instance of a failed capital raising effort by an oil & gas company.

What about the elasticity of US rig count to the price of crude? If there is anything to capital discipline at all, then the elasticity of US oil production to oil price shocks should have declined. After all, the claim seems to be that they’re being really slow in ramping up production because of “capital discipline.”

We test this hypothesis in the most straightforward way possible. We obtain data on US rig count from the IEA and the spot price of West Texas Intermediate from FRED. We resample to quarterly frequency since higher frequency fluctuations in the oil price are unlikely to affect decisions on capacity expansion and retirement. We first-difference our response, rig count, and use quarterly percent change to detrend our feature, WTI. We lag our feature by a quarter to allow information to flow from the oil market to the oil producers. Then we ask: has the response function of the oil producers changed? More precisely, has the elasticity of rig count to fluctuations in WTI changed after the 2015 crash?

In order to test this hypothesis, we include a post-2015 dummy, as well as an interaction term (to allow for the possibility that the post-2015 slope is different). The data is quarterly over 1988Q1-2022Q3. We also include dummies for three quarters (2015Q1, 2015Q2 and 2020Q2) where the observations are clear outliers. But note that including these outliers does not materially change our results.

Neither the dummy for the post-crash period, nor the interaction term is significant. Indeed, the interaction term bears the “wrong sign” if the capital discipline hypothesis is right. If anything, oil production has been more responsive to changes in the price of crude since the oil price crash of 2015.

We obtain similar results if we estimate separate regressions for different periods (pre-boom, boom-bust, and post-crash). The next figure displays the scatter plots. We can see that US rig count has been, if anything, more responsive to the price of crude since 2016. We do find, however, that the intercept was higher during the investment cycle — that’s the diffusion of fracking; an exogenous shock, independent of the relationship we’re interrogating. Let’s dig a little bit deeper.

Instead of a post-crash dummy, we introduce a dummy for the investment cycle period (2009-2015 inclusive), together with an interaction term. This tests the idea that the boom-bust cycle was exceptional due to the traditional Anglo-Saxon mania surrounding the rollout of the innovations in fracking. Unlike the idea of capital discipline, we find considerable support for this idea. Specifically, we find that there were 41 more rigs than could be expected irrespective of the price of crude, and that rig count was considerably more responsive during the investment cycle, 2009-2015, than before or after. Specifically, we find that the elasticity of rig count to fluctuations in WTI was 280 during the boom-bust cycle, compared to 150 before or after.

Irrespective of the specification of regression model we use, we find that a doubling of the price of crude predicts an increase of 150 rigs in the next quarter. Has recent rig count met this expectation? Given the fluctuations in the price of crude, our model predicts 217 new rigs over the past two years. The actual number was 419. Ie, there are more than 200 rigs too many!

The evidence is dispositive. Notwithstanding the commentary and what insiders may tell each other, there’s no capital discipline. US rig count after the oil price crash has been just as responsive to the price of crude as it has been since the 1980s.