Forecasting rig demand accurately is something that has eluded the industry for some time.  Traditionally forecasts were based on logic and factors such as:

• Known production plans from which the number of rigs required was calculated for new fields and projects with assumptions used to fill in where information is not available
• Established decline rates were used, with assumptions, to determine demand from fields already in production
• Demand estimates for consumption of oil and/or natural gas from EIA, IEA or other bodies and how much drilling is required to achieve this
• Known lease requirements for exploration and/or production to retain mineral rights
• Budgets and spending plans for oil companies based on surveys or published statements
• Future price data of oil or natural gas
• Stated plans from oil companies or drilling contractors
• Linear regressions and trends
• Moving averages
• Time series analysis

Thus the traditional bottoms-up approach makes a series of assumptions for each field, project and region in the world (wherever it is possible to) and prepares a forecast of demand.  However, when all these numbers are added, the assumptions made simultaneously in each step accumulate.  This makes the errors in each step progressively grow bigger thus depriving the eventual forecast number of reliability. This could cause the model to miss the actual demand, sometimes significantly.

The RigOutlook model does not use any of the above methods in preparing forecasts.  Some of these methods are used by us to only check our forecast, but our modeling methods are very different.

RigOutlook's forecasts are based upon non-linear differential equations which were developed exclusively for each region and rig type depending upon the region's and rig type's unique realized demand and affective forces. Using the real data and math software tools such as MathLab, MathCAD or SAS, these models are estimated and forecasts are developed with those estimates.  The data used are commonly observable industry activities and are built on data from RigLogix for offshore rigs and RigOutlook research with Baker Hughes data for land rigs.

There is a phenomenon known as system dynamics that drives demand for drilling rigs.  An overview is shown below of the process from drilling rig to consumer.

When trying to forecast how many rigs will be employed in future periods, our model implicitly takes into account the whole chain starting from the end consumer who purchases and uses the products shown at the far right above.

Demand at each stage fluctuates due to a variety of reasons that could include seasonal effects like winter heating or economic activity driven by summer driving season. There are many other discrete factors that cause demand to vary through each major step from the final consumer to refineries to pipeline/ tanker companies to E&P companies and finally to drilling contractors.

Sometimes this process is referred to as the bullwhip in the supply chain and at the very end is the manufacturers of drilling rigs and equipment for drilling rigs. The conflicting signals from downstream are part of what causes the cycles in the industry. A typical cycle should have a modest increase in demand, but unfortunately, in most cases, the fluctuating signals cause a larger than needed peak. A simple graph is shown below where the large peak is appropriately labeled overshoot.

The spike in demand during the overshoot period causes an earlier, harder, faster fall during the down market.  This process creates a volatile ride for participants, often because the downturn comes unexpectedly early and many projects were committed to, especially during the overshoot time that is not economically justifiable after demand falls.  Some of these projects are canceled, others are delayed or slowed down and the rest are carried to completion.

The balance carried to completion typically expand capacity beyond what is required and compound rationalization of capacity by making the bottom of the cycle lower, and creating a fatal price war on its way down. For each cycle, rig type and region the magnitude is determined by a variety of factors that can be unique to that sub group or applicable to all rigs worldwide.

The RigOutlook model is designed to capture the inherent characteristics of the system dynamics in determining rig demand.  During development of the model past demand was studied and the model was tailored for each region.  However, during testing of the past cycles a variety of methods were used including mock forecasting based on data for a portion of a cycle as well as modeling with and without external factors (such as oil and natural gas prices or storage levels) to determine if a factor is driving a market.

For the forecasts provided the external factors (oil and/or natural gas prices, storage levels etc) are not necessarily used to determine demand in future time periods. This is done for several reasons:

• Using future data for external points (oil and/or natural gas prices or storage levels) ties the accuracy of the model to another forecast for these data points
• While these factors play a critical role in determining demand, there are factors that can mitigate this that are unique to each cycle. These factors include:
• Long commitments that don't allow for quick reactions through actions such as canceling contracts
• Changing regulations that can cause industry changes independently of the state of a cycle
• Strategic moves being made by companies (i.e. fleet expansion, mergers etc)
• The model accuracy is high without including these factors. This is measured by R2 which is typically between 0.75 and 0.96. Note that 1 is a perfect match and 0 is no match at all and our model results are typically considered very high.

After a forecast is prepared these factors are used to check the output of the math model to ensure consistency and verify that the forecast makes sense. In the drilling industry there is variability in demand for a variety of reasons and because of this three forecasts are provided: Low, Expected and High Demand.

A typical forecast looks like the graph below. The blue line represents the number of rigs actually contracted with known future contracts shown after the date of forecast (November 2006 in this case). The red line is the model results for the past and current cycles. The three scenarios, Low, Expected and High, for demand are shown in Green, Orange and Purple lines respectively. A macro model for all cycles combined may also be shown in some regions with a black line. 

For a more detailed presentation of the model please contact us by calling (281) 345-4040 or email us at tbeebe@rigzone.com.