Fundamental Analysis of Petroleum Industry by mte65681


More Info
									New Developments in the Use of Valuation to Support Real Asset Decisions
with a CO2 Management Opportunity as an Example

             David Laughton, DLC Ltd. and the University of Alberta, +1-780-454-8846,

In 1997, I was asked to edit an issue of The Energy Journal (Laughton 1998) on how new approaches to asset valuation could be
used to improve decisions about real asset design, selection and management in the upstream petroleum industry. In this paper, I
report on some of the developments since then on this topic, including some organisational insights, and discuss the potential for
future progress.

The organisational insights are of two types. The first has helped to clarify the appropriate role for asset valuation in the overall
decision-making process along side physical and financial simulation and fundamental economic analysis. The second has to do
with the implications of organisational structure and of issues of change management for the appropriate design of new valuation

The technical developments and potential for future research are best classified by the aspect of the valuation process they affect:
the five modelling tasks involved in estimating an asset value (the decision model, the cash-flow model, the asset-level uncertainty
model, the economy-wide uncertainty model, and the valuation model), the computation of the value estimate using these models,
and the presentation of results.

On the modelling side we have much better phenomenological models of the uncertainty in competitive market commodity prices,
particularly those with public futures markets. There are also much more data to support parameterisation of these models. This
will continue to be an active field of research. This work gives us insight into the output price side of industry and some of the
input prices. However, we are only beginning to understand how to model general input price uncertainties in a useful way.

In the absence of sequential decision-making, asset-level uncertainty can be modelled just as it is for the physical simulation of an
asset. This was known in 1997. We were then and are still at the early stages of modelling asset-level uncertainties, and their
resolution, in a way that can be used for estimating asset values when sequential decision-making is important. For example, some
initial exploratory work is being done on how to package geostatistical information for these purposes.

There have been very few developments since 1997 in cash-flow modelling or in the range of approaches by which uncertainty and
time are valued.

Because of developments in computational technique, the class of decision models that can be addressed has expanded greatly. In
1997, we could analyse simple discrete sequential decisions where the best choices were contingent on one or (with some difficulty)
two dimensions of continuous time series uncertainty and a few discrete possible asset states. We can now look at decision models
with discrete controls that depends on several continuous dimensions of time series uncertainty and a large number of discrete asset

This was made possible by the optimisation methods developed by Longstaff and Schwartz (2001). The key is that optimisation is
done within a randomly selected sample of the population of possible futures instead of the full tree of possible states. We have yet
to determine the limits of this method as we apply it to situations where the underlying probabilistic processes are controlled (as can
be the case for continuously distributed asset-level variables), and to situations with more and more continuous dimensions of

This approach to optimisation also makes the specification of control very flexible. For example, simulation can be over more
variables than are used to determine optimal control. We discuss reasons why this may be useful.

The voluminous amounts of data that can be obtained from complex high-dimensional optimisations means that an area of active
research is how to package that data into information that is as useful as possible for groups of decision-makers. One approach that
has been found by some to be helpful is based on 2-D scatter plots of the optimal controls based on the two most important sources
of uncertainty. These are sequenced though sections of the other sources of uncertainty, continuous or discrete, and through time.
Some people have also found useful the information in conditional and unconditional plots of the probability of different actions
through time.

These developments are illustrated by an expanded version (Laughton et al. 2008) of the CO2 management example used in a
session I ran at the IAEE Aberdeen and Vancouver meetings in 2002 (Laughton et al. 2002) to illustrate the use in asset valuation
of models jointly for the prices of CO2 emissions rights and energy. In this example, CO2 extracted from an offshore natural gas
field is vented to the atmosphere or, if the facilities are built, can be stored in a nearby reservoir. The energy required to operate
this facility would be natural gas. The amount required is currently uncertain. We look at two situations where the price of
emission permits is uncertain and determined by a cap and trade system, or is known, arising from a “carbon tax”.

There is an initial decision to make the platform storage-ready or not. There is a subsequent decision each year while CO2 is being
produced to build the facilities, to do a last bit of appraisal of the storage reservoir, or to wait. If the appraisal is done, the amount
of energy required to inject the CO2 into the reservoir is reduced and determined. There are two possible designs for the facilities,
one of which costs more to build but is more energy efficient than the other. If the appraisal is done, there is a remaining annual
decision to build the facility or wait. If the facilities are built, there is an annual decision to use them or vent the CO2 and pay for
the permits to do so.

Approaches to valuation currently in common use cannot deal explicitly with this type of sequential decision problem. We also
show in the analysis of this example that the types of biases in DCF estimates discussed in Laughton (1998) and Laughton et al.
(2002) are also present here.

The approach to estimating asset values described here is ready for use now in commercial organisations to help with specific
decisions in specific situations. It is not yet ready for general use. However, that time is approaching. Given the lead times for a
large organisation to learn and understand a new form of analysis, any organisation that wants to be ready to use this type of
analysis when the methods are ready for use may want to begin learning about them now.

Given that, the organisational issues of change management are important. So the paper concludes by:

1)        summarising the benefits from the proposed change for a large commercial organisation that uses typical DCF valuations
as part of its analytical process;

2)     showing how this approach has been designed to conform as much as possible to the current work-flow in a typical large
commercial organisation, so that the barriers to, costs of, and disruptions from its implementation are as small as possible; and

3)        describing the different routes by which change may be managed in organisations in different situations and with different
cultures, strengths and weaknesses.

The full paper will be posted by 04 Oct 2010 at

If you decide to download it, please send me an email message at

Laughton, D. (1998), (ed.) The Energy Journal, vol.19-1

D Laughton, R Hyndman, A Weaver, N Gillett, M Webster, M Allen, and J Koehler (2002), “A special session on GHG price
uncertainty and energy project evaluation”, International Association for Energy Economics (IAEE), Aberdeen/Vancouver

D Laughton, R. Guerrero and D. Lessard (2008), “Real Asset Valuation: A Back-to-basics Approach”, Journal of Applied
Corporate Finance, vol. 20-2, 46-65

FA Longstaff and ES Schwartz (2001), "Valuing American Options by Simulation: A Simple Least-Squares Approach", Review of
Financial Studies, vol. 14, 113-147
3   Intelligent Well Technology: Status and Opportunities for Developing Marginal Reserves   SPE

To top