energy stockpiling in the asean

Energy Stockpiling in the ASEAN

Energy Studies Institute, 26 November 2013 1


Fernando Oliveira, ESSEC Business School, Singapore

Nahim Bin Zahur, Energy Studies Institute, Singapore

Tilak K. Doshi, KAPSARC, Saudi Arabia

Wu Fulan, Zhejiang University, China

A presentation at

Energy Studies Institute, NUS



Outline

• Introduction and motivation

• Econometric analysis

• Stockpiling Model

• Simulation Results

• Discussion



Introduction

• 1970s oil crises fundamentally altered perceptions of energy

security

• Strategic oil stockpiling among the major strategies available to

oil-importing countries

• IEA countries, China and India have built up large strategic

petroleum reserves (SPRs)

• What about ASEAN countries?



ASEAN’s oil import dependency is growing

-400

-200

0

200

400

600

800

1000

19

86

19

87

19

88

19

89

19

90

19

91

19

92

19

93

19

94

19

95

19

96

19

97

19

98

19

99

20

00

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10 M

MB

of

oil

eq

uiv

alen

t ASEAN Net Petroleum Imports



Oil prices are higher and more volatile

0

20

40

60

80

100

120

US$

pe

r b

arre

l

Real oil prices, 1960 - 2012



Rationale behind strategic oil stockpiling

• Oil market disruptions have a negative impact on economic

growth

– This has been shown to be true even for oil-exporting countries such as

Malaysia (Abeysinghe, 2000).

• Stockpiling can hedge against oil market disruptions by

– Reducing volatility in oil price, lessening impact on GDP

– Providing emergency supplies of oil in case of a supply disruption



Key questions

• Should ASEAN countries set up a regional petroleum stockpile?

• What should be the build-up and drawdown policies for

strategic energy stockpiles?

• How can the stockpile manager balance between mitigating the

impact of oil shocks and preserving oil stocks for future use?



Outline





• Discussion



Econometric Analysis

• We set up econometric models for:

– World oil prices

– World oil consumption

– ASEAN oil imports

– ASEAN GDP

– ASEAN oil production



Econometric Analysis - II

• Primary goal:

• To capture the stochastic nature of the behavior of these

variables

• The econometric models can be used to generate many possible

scenarios and tell us how likely they are



Oil prices The unit root test indicates non-stationarity of oil prices:

shocks to the oil price persist over time

We thus model the oil price pt as a random walk

(1)

2~ (0,14.7 )tu N (2)



Oil prices - II

The large standard deviation (14.7) illustrates that the oil price

is very volatile

(1) and (2) can be used to simulate possible future paths for oil

prices



Oil imports, production and GDP

Oil production, net petroleum imports and GDP in the ASEAN

are co-integrated with one another.

– This means that there is a common non-stationary trend

that explains the behavior of these time series.

World oil production is co-integrated with world GDP, so we

estimate a separate model for these two variables



Scenario generation

We use a scenario tree to simulate evolution of uncertainty

over planning horizon

– E.g. oil price in any node depends on the oil price in the previous node

plus a randomly generated error

– Similar approach for net petroleum imports, GDP and oil production

We construct a “reduced” scenario tree by

– decreasing the branching factor in latter periods

– increasing the time lag between successive nodes in latter periods



Example of tree that grows exponentially

With a planning horizon of 25 years, this tree generates nearly

a trillion scenarios!



Example of “reduced” tree

The same planning horizon can be represented using a smaller

number of scenarios.



Outline





• Discussion



The stockpiling model • We use a dynamic programming approach to model the

stockpiling problem

• Let y(n) = quantity of oil held in the reserve at node n

x(n) = quantity of oil bought for (>0) or sold from (<0) the

reserve at node n

k(n) = the capacity of the oil stockpile at node n

np = the parent node

nc = the child node.



The stockpiling model - II

• Equation (3) describes the build-up of stocks over time

For all nc: ( ) ( ) ( )c p cy n y n x n (3)

• We consider different rules specifying minimum oil reserves

M(n) (≥0) in the final period (i.e. the “leaf” nodes)

For all leaf n: ( ) ( )y n M n (4)



The stockpiling model - III

• The policymaker also selects how much capacity (if at all) to

build at each node

For all nc: (5)

• Oil stock purchases/sales affect the oil price (we assume a

constant elasticity of demand ε (< 0) in the oil market)

For all n:

(6)



The stockpiling model: the cost function

• The cost function in any one time period includes

– Impact of stockpiling on GDP (through oil price changes)

– Net cost of oil stock purchases/sales

– Cost of building stockpile capacity

– Holding costs

– Fill and drawdown costs

For all n: (7)



Optimization

• The full cost function includes the present costs and the

discounted sum of future expected costs, with representing

the discount factor

For all n not a leaf, (8)

( ) ( ) ( ) ( )

c

p p c c

n

n C n n n

• The policymaker’s objective is to minimize the lifetime social

cost of operating a stockpile

0

1

( )min( ),..., ( )N

nx n x n

(9)



Process for selecting optimal strategies

• Simulate uncertain variables forward through time

• Work backwards to decide how much oil to stockpile



Outline





• Discussion



Simulations

• Optimal stockpiling strategies over a 25-year planning horizon

• Initially, we set no constraints on final period stockpile

Parameter Value Unit Description

0.99 - Discount factor

α -0.06 - GDP-oil price elasticity

ε -0.067 - Price elasticity of oil demand

β 5 US$/barrel-

capacity

Cost of building one additional unit

of capacity

h 0.227 US$/barrel Annual holding costs per barrel

u 0.08 US$/barrel Cost of adding one barrel of oil into

the stockpile

d 0.10 US$/barrel Cost of withdrawing one barrel of

oil from the stockpile



Discount factor and Stockpile Size

0

200

400

600

800

1,000

1,200

1,400

0.9 0.95 0.99

Ave

rage

Sto

ckp

ile (

MM

B)

Discount Factor



Benefits of stockpiling are considerable

• Expected benefit over a 25-year planning horizon ranges from

US$ 27-130 billion with discount factor between 0.95 and 0.99

• Greater benefits with longer planning horizon

Table: Total expected cost (in US$ million) of managing the petroleum reserve as a

function of the time horizon and of the discount factor

Discount Factor

0.9 0.95 0.99

Tim

e H

ori

zon

10 years -10 -689 -5,930

15 years -667 -5,339 -29,251

20 years -1,679 -14,490 -68,470

25 years -2,593 -26,396 -131,083



Planning Horizon and Stockpile Size

• The longer the planning horizon, the greater the maximum

stockpile size reached



Why is a minimum target size needed?

• Without a constraint, the stockpile is emptied in the final period

• Yet extending the planning horizon leads to at least some oil

being stockpiled in every period bar the last

• Limited planning horizon leads to sub-optimal behavior

• Hence we need to impose a minimum target stockpile size for

final period: this can serve as a long-run policy target



Impact of different rules for final period stockpile size

• More stringent rules for final period lead to greater stockpile

size in all periods

Figure: Average stockpile size with different decision rules, discount factor = 0.99



Impact of different rules (2)

• Some rules may be too stringent e.g. 120 days rule with 0.95

discount factor

• This raises the question of identifying an adequate rule

Figure: Average stockpile size with different decision rules, discount factor = 0.95



An argument based on inter-generational equity

• The inter-generations rule

The present value of the present generation’s investments in

stocks should at least equal the discounted value of its GDP

benefits

• Intuition: the current generation should not benefit on average

by speculating on oil stocks



Inter-generational equity (2)

• Instead, benefits from oil stockpiling (in the form of avoided

GDP losses and revenue from stock sale) are invested in further

stock purchases, safeguarding all future generations

• Similar to Hartwick’s well-known rule for sustainability in the

context of resource depletion (Hartwick, 1977)



Inter-generations rule for ASEAN countries

• IEA’s suggested 90 days rule makes sense for ASEAN with a

discount factor close to 0.99

Discount Factor Rule Avg. stock

(MMB)

0.9 19 days rule 149

0.95 49 days rule 384

0.99 103 days rule 934



How should the government respond to oil shocks?

Discount factor = 0.99

No rule on leaf

nodes

103-day rule on leaf

nodes

Shock

occurs in:

Year 10 Year 15 Year 10 Year 15

Sto

ck s

ale

in

Year 10 281 228

Year 15 278 652 216 544

Year 20 296 668 237 563

Year 25 406 658 288 472



Responses to oil shocks

• Staggered release of stocks following a shock:

o Because of the possibility of aftershocks, it is in general not

optimal to empty the stockpile immediately

The magnitude of the response is greater if the shock occurs in a

later period (due to greater accumulated stockpile and lower

benefits of keeping the oil in stock)



Sensitivity analysis: GDP-oil price elasticity

The GDP-oil price elasticity measures the extent to which GDP

is reduced by oil price increases



Sensitivity analysis: price elasticity of oil demand

The more elastic the oil market, the greater the stockpile size

0

500

1000

1500

2000

2500

2012 2017 2022 2027 2032 2037

Sto

ckp

ile S

ize

(M

MB

)

Years

Base Case

High price elasticity

Low price elasticity



Outline





• Discussion



Policy Implications

The ASEAN would benefit considerably from building a

regional stockpile.

The expected benefits is about

US$ 27 – 130 billion range

A conservative target for the stockpile towards is about 103

days of net imports.

The optimal strategy in response to an oil price shock involves

an immediate large sale of oil stocks, followed by further releases

in future periods.



Policy Issues: Private Stockpiling

Public stockpiles could displace commercial stockpiling,

reducing effectiveness.

Alternatively, commercial stockpiling could be used to

complement public stocks

Private companies typically reluctant to commit to

minimum stock requirements

– European example: “ticket” system?



Policy Issues: Regional Cooperation

We assume ASEAN countries cooperate in building a

regional stockpile, but this is not a given!

Benefits from stockpiling likely to be much less if each

country acts independently

Some difficulties with cooperation – Unevenly divided benefits

– Free-riding incentives

– Location issues.



TThhaannkk YYoouu

energy stockpiling in the asean

Documents