VoxEU Column Energy

Revisiting the relationship between oil prices and costs in the upstream industry

The recent dramatic decline in the price of oil runs counter to the argument that oil prices should be high because of the high costs. This column presents new evidence on this relationship. Using a representative global dataset, the authors find that upstream costs follow oil prices with a time lag. In particular, a sustained 10% increase in the price of oil leads to an increase in upstream activity of about 4%, and in this way triggers a sustained 3% increase in global upstream costs after a lag of one to two years.

Exploration and production (E&P) costs in the oil and gas industry increased by some 100% between 2000 and 2012 (IHS 2014). The higher cost of hydrocarbons production has previously been put forward in much of the economic literature as one of the primary reasons for structurally higher oil prices. However, the recent dramatic decline in the price of oil (the price of Dated Brent fell by over 50% between mid-July 2014 and early 2015) proved that the popular argument that oil prices could not fall because of high costs is incorrect.

Two questions about current oil price environment: New evidence

These developments raise two important questions relevant to the current oil price environment:

  • What is the true relationship between the price of oil and the cost of production per barrel?
  • And, what are the implications for the oil industry of the current price and cost environment?

On the supply side, when prices fall it is reasonable to expect that either more expensive production would be reduced to ensure profitability at new prices, or cost would fall, or some combination of the two. Indeed, cost inflation in the oil industry has stalled in the past few years and has started falling more recently as the industry adjusts to the new, lower oil price environment (see Figure 1). This holds implications for the industry as cost deflation could materially improve the profitability of projects. The prospects for a continuation of this trend are of particular relevance in an industry where project scales are typically large and long-term, and require considerable upfront capital investment.

Figure 1. Growth in three-year moving average of the real oil price and drilling costs (measured by the growth in the upstream capital cost index which is provided by IHS)

Furthermore, in response to the fall in prices all major independent oil and gas companies have announced spending cuts on the order of 20-25% for 2015 as opposed to 2014; and according to Wood MacKenzie (2015), they have delayed final investment decisions on a number of large projects. Lower investment today should mean lower output and higher prices tomorrow, keeping other factors constant. How much will costs fall? What will happen to activity in the oil and gas industry? To answer these questions, in a recent study (Toews and Naumov 2015) we explore the relationship between costs in the oil and gas industry and the price of its most important commodity – the price of oil. This column summarises our main findings.

Classical economic theory would assert that in a competitive industry the cost of producing one extra unit of output (marginal cost) will determine its price. In other words, if it takes $100 to explore for, produce, and deliver an additional barrel of oil globally, the global price of oil should be $100/bbl. However, in the global oil and gas industry (taken as a whole), a number of central preconditions for this mechanism are absent; perhaps the most important among them being perfect competition. Due to access restrictions, the additional barrel of oil remains untouched in places such as the Saudi desert. OPEC, a producer cartel, manages production through adjusting its spare capacities, which can affect prices and so complicate their relationship with marginal cost. This therefore calls for a more robust empirical approach to determine how costs and prices interact in this industry, which forms the basis for further insights and allows us to make predictions.

We construct a simple empirical framework and use a previously unexplored global data set to estimate the statistical relationship between costs and prices in the upstream sector of the oil and gas industry. Our three main findings can be summed up as:

  • Following a 10% increase (decrease) in the oil price, drilling activity increases (decreases) by 4% and the cost of drilling increases (decreases) by 3% with a lag of 4 and 6 quarters respectively (note that this lag can already be observed in Figure 1).
  • Conversely, exogenous shocks to costs have only a small and statistically insignificant impact on the oil price in the short to medium term.
  • Finally, exploration activity shocks negatively affect the oil price.

Previous research

Our analysis is most closely related in methodology to Kilian (2009) who uses a similar econometric framework to understand the different nature of shocks driving the oil price, distinguishing between aggregate demand shocks, oil supply shocks, and so called precautionary demand shocks. His main conclusion is that after accounting for aggregate demand, it is variation in precautionary demand for oil rather than supply shocks that has been historically responsible for the majority of variation in oil prices. We extend his work, treating the variation in the oil price as given, and we use it to study the observable variation in exploration activity and costs. Our work is also related to that of Anderson et al. (2014). Using data from Texas on drilling activity and rig rents, they present evidence that drilling activity and drilling costs significantly respond to changes in the oil price. On the other hand, they are not able to find any significant relationship between oil price changes and the contemporaneous extraction of oil. They use these results to build a theoretical model in which drilling activity is at the centre of an optimal oil extraction path. Our work differs from their contribution in two main aspects. First, their main contribution is theoretical whereas our focus is on the identification of causality and estimation of dynamic responses to shocks. Second, we use global data on drilling and cost of drilling, whereas their data is constrained to Texas. Finally, our research is also related to a more applied literature which focuses on the identification and the estimation of the price elasticity of drilling in the oil and gas sector; see Dahl and Duggan (1998) for a comprehensive review of the literature.

Investigating the data on costs and exploration

Reliable long-term data on exploration and production activity and cost developments are hard to come by. A common source of such cost data is IHS Energy’s Capital Costs Analysis Forum. It has the advantage of breaking costs down by region as well as by components on a quarterly basis. However, this dataset records only cost and not investment activity in the oil and gas sector.

Another source of data is a time series of upstream expenditure from Barclay’s E&P Spending Survey, covering the period 1978- present. The drawback here is that one has to use a proxy for exploration and production activity (such as the rig count, adjusted for changes in rig quality) to calculate unit costs from these spending data. The US provides the longest run of oil price and cost data, with an index of oil and gas investment costs going back over 100 years. This dataset is ideal for our purposes; however, it does not cover the ‘global’ picture.

To overcome these data limitations, we assembled time series data on drilling activity from Wood Mackenzie (Woodmac). This dataset makes it possible to record activity as well as cost in oil and gas exploration and appraisal for the period 1995 through 2014, approximated by the number of wells drilled and the real costs of drilling a well. Drilling of exploration wells has accounted for 40-50% of capital expenditure in recent years; it is therefore a good indicator of overall cost inflation.

The data distinguish individual wells by company, country, and onshore/offshore status. Most importantly, this enables us to separate onshore from offshore drilling activity and costs of drilling (henceforth referred to as the ‘location’ of drilling). We construct two quarterly time series capturing (i) the total number of exploration and appraisal wells drilled by upstream exploration companies, and (ii) the average cost of drilling these wells. In combination with the oil price, we therefore have three time series consisting of 75 time periods. In Figure 2 and Figure 3 the total number of wells drilled and the average cost of drilling is presented by location.

Figure 2. Total number (logged) of drillings (exploration and appraisal wells) across locations

Figure 3. Average cost (logged in thousand USD, real) of drilling a well across locations

To our knowledge, this is the most detailed global data source on drilling activities and costs in the oil and gas industry. Nevertheless, some limitations within these data should be noted. In particular, and due to the large volume of information available, Woodmac’s dataset does not record information on US onshore drilling activity. However, in order to conduct a robust analysis we used the same estimation procedure with the US-specific data described above, and as expected the results were largely consistent with our reasoning.

To investigate the relationship between oil prices and upstream cost, we use a standard method in time series econometrics (structural Vector Auto Regression) to decompose the variation in the data into three different shocks. In particular, we identify the following shocks.

  • Oil price shocks are changes in the price of oil caused by oil market dynamics.

Examples are numerous and include most oil price changes (for instance the surge in oil price between 2003 and 2008) related to an increased global demand.

  • Activity shocks are unanticipated changes in exploration activity (measured by the number of wells drilled) which are not driven by a change in the oil price.

Such activity may increase if countries suddenly open up their upstream sector for private investment as is happening currently in Mexico. Advances in technology, such as ultra-deep water drilling, could also allow for exploration in parts of the world that were not accessible previously. Similarly, nationalisation of an oil industry and tougher regulations may decrease activity.

  • Cost shocks are defined as variations in the cost of drilling after accounting for changes in drilling activity and the oil price.

Exogenous shocks to drilling costs may occur due to technological advances which make drilling more efficient and therefore cheaper. Alternatively, increasing project complexity could drive the costs up. These shocks can also originate in the supply chain all the way down to the manufacturing of rigs.

Once the shocks have been identified we can estimate what effect (if any) they have on our variables of interest – oil price, cost of drilling, and exploration activity. Finally, we note that our main results are generally consistent across all of the datasets described in this section and for a range of alternative estimation techniques.

Oil prices drive costs

Our results show that a one-off 10% increase (decrease) in the price of oil (oil price shock) increases (decreases) global exploration activity by 4%, which in turn leads to an increase (decrease) in cost of 3%, but with a time lag of 1-2 years. The responses of the variables in the system to an oil price shock are presented in the first row of Figure 4. The logic behind this result is relatively straightforward. An increase in oil prices will boost cash flows and raise the expected rate of return on marginal projects for the industry. Both channels will prompt exploration and production investment to rise. An increase in investment in more costly projects clearly affects the composition of exploration wells drilled by adding more costly wells and thus pushing up the average cost of wells drilled. Additionally, due to the fact that supply chain capacity is limited in the short to medium term, many input prices and therefore total upstream costs will rise, but with a lag reflecting, for example, the nature of existing contractual arrangements.

An increase in exploration activity (activity shock) does not seem to have a statistically significant effect on cost. On the other hand, as new oil supplies become available, the oil price is negatively affected, reaching a new lower level within 1 year. The responses of the variables in the system to an activity shock are presented in the second row of Figure 4.

A change in drilling cost (cost shock) due to rising project complexity might reduce profit margin and available cash, which in turn may translate into a reduction in capital spending. Consequently, cutbacks in the capital expenditure of oil companies would lift oil prices through lower future supply. This intuition sounds very plausible, but the data tell us otherwise. Our estimation results are surprisingly clear that in the time period studied (1995-2014) exogenous shocks to industry cost (proxied by the cost of drilling after controlling for the variation in oil price and drilling activity) did not have a statistically significant and lasting effect on either exploration activity or the price of oil.

  • In other words, as far as the data and method are concerned, oil prices drive costs, whereas we are not able to find compelling evidence that the reverse is true.

The responses of the variables in the system to a cost shock are presented in the third row of Figure 4.

The notion that cost does not affect price may sound strange at first, but in fact, the working assumption in the oil and gas industry in the 1990s and 2000s was that industry can ‘eat’ cost inflation by being more efficient and constantly innovating. In our view, this is reflected in the data.

Figure 4. Responses of the variables in the system to one-standard-deviation structural shock

Policy implications

Upstream costs depend on many factors. One major driver, which is often overlooked, is the price of oil. This column introduces the results of new research on this relationship; using a representative global dataset and appropriate econometric techniques, we find that upstream costs follow oil prices with a time lag.

  • In particular, a sustained 10% increase in the price of oil leads to an increase in upstream activity of about 4%, and in this way triggers a sustained 3% increase in global upstream costs after a lag of 1-2 years.

These results have several important implications.

They could improve the planning and forecasting of long-term project economics in the oil and gas industry. Linking cost estimates to oil prices will produce more realistic estimates of project economics, particularly for projects where capital spending is distributed relatively evenly across long project lifetimes, that is, where costs are not entirely locked in at the beginning of a project’s life cycle.

Another implication concerns the timing and planning of long-term supply contracts, especially for those projects with large upfront spending; if past movements in the price of oil help to anticipate the future direction of exploration and production costs, then (for example) costs could be negotiated down where they have not yet adjusted to a decline in oil prices, to avoid locking in costs associated with high prices – and vice versa. In fact, in April 2015 a US oil services company signed a five-year rig contract with a subsidiary of a major IOC where the day rate for the rig was linked to the price of oil.1 Our results suggest that we are likely to see more follow suit.

Last but not least, the results reported here contribute to the literature on the determinants of the price of oil. One of the more persistent hypotheses often heard in times of rising prices is that oil prices cannot fall because costs are too high. Our results clearly demonstrate that such statements are a fallacy. Not only have we showed that costs follow oil price, we also cannot present any evidence for the reverse. Meanwhile, oil prices tend to be affected by shocks in global exploration activity.


Anderson, S T, R Kellogg,  and S W Salant (2014), “Hotelling Under Pressure," Working Paper.

Dahl, C,  and T E Duggan (1998), “Survey of price elasticities from economic exploration models of US oil and gas supply," Journal of Energy Finance & Development, 3(2), 129{169.

FT (2015): “Oil groups have shelved $200bn in new projects as low prices bite”. http://www.ft.com/cms/s/0/d6877d5e-31ee-11e5-91ac-a5e17d9b4cff.html#axzz3j6W5OhtH

IHS (2014), “IHS Upstream Spend Report," Report.

Kilian, L (2009), “Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market," The American Economic Review, pp. 1053-1069.

Toews G and A Naumov (2015), “The Relationship Between Oil Price and Costs in the Oil and Gas Sector”, Working Paper.

Wood Mackenzie (2015), “Upstream Data Tool," http://www.woodmac.com/, Accessed: 2015-05-03.


12,598 Reads