IBP1986_18

EVALUATION OF QUANTITATIVE
INDICATORS FOR PORTFOLIO
SELECTION IN THE OIL INDUSTRY

Valente, Pedro1, Simplício, Jalimar 2, Seize, Patricia3

Copyright 2018, Brazilian Institute of Oil, Gas and Biofuels - IBP

This Technical Work was prepared for presentation at the Rio Oil & Gas Expo and Conference 2018, held from
September 24th to 27th, 2018, in Rio de Janeiro city. This Technical Work was selected for presentation by the event's
Technical Committee, based on the complete paper submitted by the author(s). The organizers will not translate or
correct the received texts. The material as presented does not necessarily reflect the opinions of the Brazilian Institute
of Oil, Gas, and Biofuels, Partners, and Representatives. It is known and approved by the author(s) that this Technical
Work be published in the Proceedings of the Rio Oil & Gas Expo and Conference 2018.

Abstract
Oil companies are constantly re-evaluating their portfolio in order to obtain a
portfolio of projects that bring the maximum return to their shareholders with the lowest
risk. In order to analyze the projects that make up this portfolio, return and risk metrics
should be used that: (i) allow comparability between projects of different maturities and
different businesses, such as the comparison of exploration and production projects and
(ii) are robust and (iii) are understood by managers.
In this sense, the purpose of this article is to evaluate how the choice of different
risk and return indicators for project analysis can affect the selection of a company's
portfolio of projects in this industry.
For this, a typical portfolio of an oil company was estimated, with production
development and oil refining projects, which were evaluated through the return metrics: NPV
and NPV/UI (Net Present Value over Updated Investment) and metrics of risk: standard
deviation, VaR (Value at Risk) and RAROC (Risk Adjusted Return on Capital).
The portfolio resulting from the application of each indicator was evaluated using
the Markowitz approach, which resulted in the efficient border adherence portfolio
being the NPV x VaR.

Keywords: Exploration. Production. Refining. Portfolio. Risk. Return. Markowitz.

1
PhD, Economist - PETROBRAS
2
Master's, Equipment Engineer – PETROBRAS
3
Master's, Economist – PETROBRAS

Confidential C
 Rio Oil & Gas Expo and Conference 2018

1. Introduction
The oil industry has historically shown cycles that can occur due to various external
factors, both from the supply side and the demand side of oil. Supply variations occur,
among other factors, due to geopolitical conditions or new discoveries of exploration
areas. Demand variations can occur due to changes in the economic growth rate in
countries that have an energy matrix dependent on oil and its derivatives. Both generate
fluctuations in oil prices that directly affect the sector's results.
In addition to external factors, each company in the sector has a current internal
position considering its existing assets, ongoing projects, markets it has access to, and its
operational and financial situation. In this scenario, the companies that are part of this
industry are constantly evaluating their strategies and actions to define the direction to
achieve their future objectives.
Once external scenarios are defined, and having evaluated their strengths and
weaknesses resulting from internal factors, the company will seek to identify the strategic
directions it should pursue to move from its current condition to the desired position. In
general terms, this path is built by new investment projects and process improvements.
Thus, the set of projects of a company is a fundamental component for achieving its
strategic goals, and the criteria for selecting the projects that will be part of this portfolio
is the subject of this work. As a selection criterion related to the return of the projects, the
Net Present Value (NPV) is the indicator most commonly used by companies, however,
other indicators can also be used such as NPV/UI (Updated Investment) and internal rate
of return (IRR).
In addition to the economic return indicators, the projects are exposed to different
levels of risk in achieving their objectives. For the purposes of this work, risk is defined
as the uncertainty underlying the forecasts of the main variables responsible for the
project's return. In this case, different indicators can also be used such as standard
deviation, variance, VaR (Value at Risk), RAROC (Risk Adjusted Return on Capital),
among others.
With this, in this work, we will apply methodologies evaluated in different studies
on this topic using simulation data to represent projects that typically make up a portfolio
in an oil company and check how the choice of these risk indicators can affect the final
composition of a portfolio in this industry.
This study seeks to analyze the impact of the choice of different risk and return
indicators on the composition of the project portfolio of an oil company composed of the
production and refining segments. For this purpose, the portfolios resulting from some
combinations of indicators are compared with the result that uses the Markowitz theory.

2. Indicators
For the purposes of selecting the portfolio of assets and projects of the company,
the metrics adopted, both for return and for risk, must cater to the various types of projects
of an organization, whether they are new projects, mature projects, or from different
businesses, allowing for the comparison of production development projects with refining
projects, for example.

2.1. Measurement of Project Portfolio Return

For measuring the portfolio's return, the following indicators were evaluated, with
their respective advantages and limitations:

Confidential C
 Rio Oil & Gas Expo and Conference 2018

2.1.1. Net Present Value (NPV):

It is the algebraic sum of the net costs and benefits of the project over its economic
life, brought to the date of updating the cash flow using the appropriate discount rate: the
weighted average cost of capital (WACC) of the business segment.
NPV is a metric that allows the evaluation of economic viability and also ranks
projects. A positive NPV means that the updating of benefits exceeds the estimated costs,
and its amount represents the value that the company will be adding to its equity by
undertaking the project, compared to not undertaking it, that is, it is the additional
expected gain to the remuneration obtained by applying resources at the company's
weighted average cost of capital.
The main advantages of NPV according to Vitolo 2015 are:
- Simplicity of calculation and understanding by managers;
- Considers the value of money over time;
- Roughly considers the risk associated with the return on invested capital through an
adjustment in the internal rate of return.
NPV also has some limitations, as per Hawawini and Viallet (2007, p.228):
- It does not consider the difference in the investment amount between different projects,
i.e., it does not consider the relationship between the expected value to generate (benefit)
and the initial investment (cost);
- Capital limitations within the horizon of the analyzed projects are not considered, i.e.,
selecting projects purely by the NPV criterion may generate sub-optimal portfolios or
those that exceed the company's budget;
- Opportunities for decision-making that maximize the value of the project in the future
are not considered, i.e., NPV does not measure opportunities for managerial flexibility
associated with the process;
- Significant influence of the cost of the invested capital on the NPV. This discount rate
can be calculated by different methodologies, including or not including different
components of risk and uncertainty.

2.1.2. Net Present Value over Updated Investment (NPV/UI)

This metric is a ratio of NPV to the investment updated by the WACC, providing
a view of investment efficiency.
The main advantages of NPV/UI as attributed by Simplicio (2001) are:
- Simplicity of calculation and understanding by managers;
- Broad technical knowledge;
- Robustness;
- Economic relevance;
- Non-dimensionality of the indicator;
However, NPV/UI also has some limitations:
- The metric can show distortions for assets with low and high cost.

2.2. Measurement of Risk in Project Portfolios

The uncertainties related to the expected return have been the focus of recent
studies that evaluate criteria for selectivity and ordering of investment projects aiming
at the formation of optimized portfolios in the risk-return space.

2.2.1. Variance
In the horizon of modern finance theory, Markowitz (1952), in his seminal article
“Portfolio Selection”, proposes a methodology for forming efficient portfolios in the
3

Confidential C
 Rio Oil & Gas Expo and Conference 2018

risk-return space, associating them respectively with the statistical metrics that define the
mean and variance. The implicit assumption in his study considers the normality of the
random returns of financial assets in the capital market. Variance (V) is a statistical
measure defined by:

Asset Variance: 𝑉 = ∑𝑛𝑖=1 𝑝𝑖 ∗ (𝑦𝑖 − 𝐸)2

The expected return and the risk of the portfolio, in turn, are defined by:
𝑛
𝐸(𝑟𝑝 ) = ∑ 𝑤𝑖 ∗ 𝐸(𝑟𝑖 )
𝑖=1

Portfolio Variance: 𝑉𝑝 = ∑𝑛𝑖=1 𝛼𝑖2 ∗ 𝑉(𝑋𝑖 ) + 2 ∑𝑛𝑖=1 ∑𝑛𝑗>1 𝛼𝑖 𝛼𝑗 𝜎𝑖𝑗

Where,
E(𝑟𝑖) – expected return of asset i;
E(𝑟𝑝) – expected return of the portfolio;
𝑉𝑖 – variance of the returns of asset i;
𝑉𝑝 – variance of the returns of asset i
E – Expected Value; 𝑦𝑖– values observed in a sample of size n
p – probability of V = 𝑦𝑖;
n – nth random return;
𝜎𝑖j - covariance of the returns of asset i relative to the returns of asset j;
𝛼𝑖 – relative proportion of capital allocated to asset i;
𝛼j – relative proportion of capital allocated to asset j

2.2.2. VaR (Value at Risk)

Concurrent with case studies that apply modern portfolio theory to investment
decisions in real assets, still in the 1990s, the financial market began to discuss and
conduct evaluations where market risk is quantified using approaches associating it
with a predetermined percentile of the return distribution. Known as “Value at Risk”,
VaR is synthesized by Jorion, P (1997) as the greatest (or worst) loss expected by an
investor within a given time horizon and confidence interval. Usually, VaR has been
associated with a probability ranging between the 1% and 10% quantiles. The following
figure illustrates the concept of VaR for a Gaussian (normal) distribution:

1 [−
1
(𝑥−𝜇)2 ]
𝑓(𝑥) = ∅(𝑥) = 𝑒 2𝜎2
√2𝜋𝜎 2
Where:
µ – mean; σ² – Variance; X – random variable.
4

Confidential C
 Rio Oil & Gas Expo and Conference 2018

This risk metric, and others of the same nature such as CVaR, has been used in
recent studies applied to real assets.

2.2.3. RAROC (Risk Adjusted Return on Capital)

More recently, the financial credit system has been using RAROC (Risk Adjusted
Return on Capital) and its derivatives to measure the risk of losses in relation to return.
When applied to financial assets, RAROC is defined by the following relation:

Revenue − expense − expected loss + capital revenue

RAROC =
capital
Applied to real assets, the RAROC equation:

𝑁𝑃𝑉
𝑅𝐴𝑅𝑂𝐶 =
𝑈𝐼 + (𝑁𝑃𝑉 − 5𝑡ℎ
𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑖𝑙𝑒 𝑜𝑓 𝑁𝑃𝑉)

Where:
NPV – Net Present Value; UI – Updated Investment to Present Value

2.2.4. Comparative Studies of Previous Risk Metrics

The study by McVean (2000) is motivated to evaluate the impact of using different
risk metrics on the composition of efficient portfolios in the risk-return space, with the
author considering variance, semi variance, VaR (@10%), and the probability of a
predetermined loss as risk measures. The study's results indicate that the efficient
portfolios formed are not neutral to the choice of risk metric. Galeno (2009) combines the
use of portfolio theory (minimization of return for a given return) and preference theory
(certain equivalent) to propose the selection of an oil production project portfolio, with
risk represented by variance and semi variance and two risk tolerance values.

Confidential C
 Rio Oil & Gas Expo and Conference 2018

3. Types of Projects

3.1. Oil Production Development Projects

Oil Production Development (PD) projects, characterized by the goal of
commercially exploiting hydrocarbon accumulations in blocks where oil companies hold
production rights, face uncertainties related to volumetric and reservoir properties, oil and
gas prices, and expenditures on wells and production facilities, which reflect on their
economic indicators.
The sample database for the PD projects used in this study consists of 9 PD projects
in Brazilian sedimentary basins, with 6 of them offshore (sea) and 3 onshore (land). In
conceptual terms, the engineering projects for exploitation of PD projects on land and at
sea are similar. In both, there is a demand for the execution of subsurface activities, which
consist of drilling wells, and installing surface facilities to connect the wells to the
primary hydrocarbon processing plants. In offshore PD projects, the production facilities
include equipment and subsea pipelines. In onshore PD projects, the land production
facilities connect the onshore wells to collector stations and triphasic oil processing
plants.
In the database used, offshore PD projects have significantly longer implementation
times, higher levels of expenditure, and larger recoverable hydrocarbon volumes
compared to onshore projects. The implementation time for offshore projects varied from
2 to 9 years depending on the recoverable hydrocarbon volume and the availability of
production infrastructure. As for onshore PD projects, the implementation time varied
from 6 months to 6 years.

3.2. Oil Refining Projects

In the oil industry, a large part of companies operates in an integrated manner
between their oil production and refining capacity. The degree of integration can vary and
depends on participation in the fuel markets as well as the strategies of each company.
In this work, to compose a typical portfolio of projects for an oil industry, we will
simulate a portfolio of refining projects. A refinery is a quite complex asset, and its
production can vary significantly depending on the complexity and capacity of its units,
as well as the types of crude oils used as input. Moreover, the outcome of this refinery
and also of the projects to be implemented depends on the efficiency of its operation and
other costs that may affect its results.
The production of derivatives from petroleum basically involves three main
processes as described on the Petrobras website:
- Distillation – This is the process of separating derivatives: the oil is heated to high
temperatures until it evaporates. This vapor then returns to liquid state as it cools at
different levels within the distillation tower. At each level, there is a container that collects
a specific petroleum byproduct.
- Conversion – This is the process that transforms the heavier and less valuable
parts of the petroleum into smaller molecules, resulting in more valuable derivatives. This
increases the utilization of the petroleum.
- Treatments – These are the processes aimed at adjusting the derivatives to the
quality required by the market. In one of these processes, for example, sulfur is removed.
In this way, within what we consider as refining projects, we must assess the
separation of investment in these 3 basic units. For the construction of projects for these
units, even in a very generic manner, we need to consider some investment and cost
metrics for each of them. Similarly, to consider the gains, the margin between the initial
products of a distillation unit and the products that can be generated later with conversion 6

Confidential C
 Rio Oil & Gas Expo and Conference 2018

and treatment processes should be considered.

The information for the estimation of these projects is not easily obtained as it is
not normally disclosed by the companies implementing them. Investment and cost
metrics based on Vaillancourt (2014) were used, as per the table below:

Table 1: Parameters Used for the Refining Projects

Project Capacity (bbl/d) Investment (US$/bbl/d) Operation Costs (US$/bbl/d) Depreciation

New Refinery 200,000 Between 25 and 37 Between 3 and 7 10 years
Expansion Project 50,000 Between 17 and 26 Between 3 and 6 10 years
Treatment Project 47,000 Between 7 and 8 1 10 years

(*) The average operating costs include fuel costs and exclude costs of depreciation, amortization, and energy.

For the Refining portfolio of this study, 9 projects were considered: (i) a new
refinery, (ii) an expansion project, and (iii) a treatment project. Each of these projects was
considered in markets with low growth (2% per annum), medium growth (4% per
annum), and high growth (6% per annum), resulting in a portfolio with 9 projects.

4. Analysis of Project Portfolios through Risk and Return Indicators

For the definition of the portfolio by the risk and return method, the 18 projects
were plotted on a graph, which was divided into quadrants. In this division, the projects
were classified as high/low risk and high/low return, according to the following criteria:
- For the return, the P75 of the portfolio was calculated. Projects that showed a
return above this value were classified as high return projects, and those below as low
return.
- For risk, the P25 of the portfolio was calculated. Projects that showed a risk above
the P25 were classified as high risk, and those below as low risk.
Analyses were made of NPV x Standard Deviation (Figure 1), NPV x RAROC
(Figure 2), NPV x VaR (Figure 3), NPV/UI x Standard Deviation (Figure 4), and NPV/UI
x VaR (Figure 5). On all graphs, the Y-axis represents the considered return indicator,
and the X-axis, the risk indicator.
The results can be observed in figures 1 to 5 below:

Figure 1 – NPV x SD Indicator

Confidential C
 Rio Oil & Gas Expo and Conference 2018

Figure 2 – NPV Indicator vs. RAROC

Figure 3 – NPV Indicator vs. VaR

Figure 4 – NPV/UI Indicator vs. SD

Confidential C
 Rio Oil & Gas Expo and Conference 2018

Figure 5 – NPV/UI Indicator vs. VaR

For the analysis of the project portfolio to be compared with the Markowitz
methodology, projects with high risk and low return were excluded from each of these
graphs.

7. Results and Conclusions

When we compare the project, portfolio selected through the application of
Markowitz with the project portfolio selected through NPV/VaR (Figure 6), we see that
the choice considering the risk and return indicators proved to be a method that aids in
selecting the best portfolio for the company.

Figure 6 – Markowitz vs. Analyzed Indicators

For its use, it is necessary for the company to know its assets and projects as well
as its limits for risk appetite and criteria to define the boundaries between high/low return
and high/low risk.
In this article, we use the percentiles of the variables' information from the projects
as criteria for choosing the limits of return and risk. To evaluate the outcome of these
choices, the risk and return information of the portfolios selected through
With our sample, we can observe in the graph that the result of the NPV x VaR
indicator was the closest to Markowitz's efficient frontier. Thus, we consider that this
method, being easier to implement, can be used by companies for selecting their project
portfolio.

Confidential C
 Rio Oil & Gas Expo and Conference 2018

References:

HAWAWINI, G; VIALLET, C. Finance for executives: managing for value creation. 3rd
ed.

Mason, USA: Thomson South-Western, 2007.

GALENO, R.C. Impacto de diferentes métricas de risco na seleção de portfólios de

projetos de produção de petróleo (Impact of Different Risk Metrics on the Selection of
Oil Production Project Portfolios), REM: R. Esc. Minas, Ouro Petro, 62 (3): 305-313, jul.
set., 2009.

JORION, PHILIPPE. Value at Risk: A Nova Fonte de Referência para o Controle de

Risco de Mercado (The New Reference Source for Market Risk Control). © Bolsa de
Mercadorias & Futuros mediante contrato com The MacGraw-Hill Companies, 1997.

MARKOWITZ, HARRY. Portfolio Selection, The Journal of Finance,Vol.7, Nº. 1,77-91,

1952; MCVEAN, J. R. The Significance of Risk Definition on Portfolio Selection, 2000.

SIMPLÍCIO, JALIMAR. Carteira de ativos reais: avaliação da eficiência econômica do

processo de seleção dos projetos de investimento de uma empresa de petróleo (Real Asset
Portfolio: Evaluation of the Economic Efficiency of the Investment Project Selection
Process of an Oil Company). Master's Thesis – Federal University of Rio de Janeiro,
2001.

VAILLANCOURT, KATHLEEN. Oil Refineries, Energy Technology Systems

Analysus Programme, 2014.

VITOLO, GUILHERME. Avaliação de Indicadores para seleção de portfolios de projetos

(Evaluation of Indicators for Project Portfolio Selection). Master's Thesis – Polytechnic
School of the University of São Paulo, 2015.

Available at: http://www.petrobras.com.br/pt/nossas-atividades/areas-de-atuacao/refino/

10

Confidential C