Misunderestimations – how accurate are production forecasts from industry and analysts?

We examine how accurate industry participants are in forecasting production in advance, and see how these seem to be consistently over optimistic

Oil exploration and development is an enormously difficult process. To successfully find oil or gas, companies have to undertake complex geological groundwork, including acquiring expensive seismic data. Experienced and highly trained staff have to interpret that data, processed by supercomputers, to try to understand what rocks thousands of metres below the surface and laid down tens of millions of years ago, may contain.

Rigs then have to be contracted (some costing hundreds of thousands of dollars a day to hire) that are powerful and accurate enough to end up within metres of the intended target after drilling more than two kilometres of rock.

If found, the oil than has to be developed and produced, which will often take a decade from discovery to first production, and costs billions of dollars and millions of man hours. Under these conditions it is not surprising that difficulties arise and delays result.

Having said that, these are difficulties inherent in the industry and ones that you would imagine would be at some level improvable over time. So why is it that forecasts of future production by many major oil companies are so consistently over optimistic, and why do the market believe the forecasts this year?

As can be seen below, the forecasts of production (grey lines) for many of the majors have overplayed the actual production over time (dark green), and for many this has been consistent. Perhaps more puzzling is why the analysts covering the companies are comfortable forecasting that production rises so strongly (in line with company forecasts), given the historical suggestion that this seems unlikely.

Below is the production over time split for major oil companies (BG (now subsumed by Shell), BP, Chevron, ENI, Exxon, OMV, Repsol, Shell, Statoil, Total) compared to the forecasts given by these companies over time. It is clear that over the time period examined, companies did over estimate their production markedly.

Factors that affect production that can’t be predicted/divulged to the market include planned acquisitions/divestments (particularly important for BP’s production through the period following the GoM incident), oil/gas prices affecting PSC terms, OPEC requirements and entitlement volumes and these will have played a part in the overestimation. Of these divestments and acquisition are the easiest in theory to normalise for, though in reality production for assets in any year after the acquisition is harder, while for large companies, not every sale is fully documented for investors. The dotted green line represents current analyst forecasts of production (see section later).

We note that the companies’ websites restrict the historical presentations we have been able to access. ENI and Total go back the furthest while Statoil only goes back to 2012. This may make the predictive accuracy of the companies with the most lines look the worst, when this may not be the case. We would like to go back further to test how the accuracy was from 2000 or earlier, but that is not practical at this time.

We would note that the analyses below is almost certainly true for most industries and is a feature of human nature and the reality of business pressures. We are therefore not explicitly criticising the companies directly, but rather trying to point out for investors a common theme that they may seek to take account of, particularly as strategy season continues.

Interestingly, virtually all companies continued to forecast growth through multiple periods even when oil production fell by 1.5% per year between 2006-2014. The trend seems now to be resolutely up, with a combined growth rate of 2.7% out to 2020, but these figures seem to indicate that investors may be wise to re-visit these broad assertions. There is no reason to disbelieve that many companies will grow at the rates that they forecast, but as a whole, we would be surprised (given the historical trend only) if this actually holds for the group.

Accuracy of market analysts
Companies are the first source of data for analysts on production targets and growth assumptions – without information provided by the companies, analysts would have less visibility on which projects will come online when. It is therefore right that analysts rely heavily on these estimates. It is also fair, given what we have seen, that analysts should be cautious of using these numbers as a baseline, given that companies (in the period 2010 onwards anyway) routinely missed their growth targets and instead shrank production by 1.5%. Missed production targets mean less cashflow which means worse financial positions and reduced ability to fund debt, capex, dividends and buybacks.

Below are the analyst revisions for production estimates for the majors from 2010 onwards. This shows material downgrades across the board. As can be seen in the charts above, the consensus going forward is generally in line with company forecasts, and this meant that analysts were much too optimistic in forecasts two years forward (by an average of 8% 2011-2016) but the estimates were unsurprisingly more reliable on the 1st January of the year in question (at 2.3% too optimistic).