Commodity Trading Value Chain

Craig Pirrong

Bauer College of Business

University of Houston

Commodity Transformations

• All commodities undergo transformations through the value chain

• Transformation in space (transportation)

• Transformation in time (storage)

• Transformation in form (processing)

Some Examples

• Power plants transform fuel into power

• Pipelines transform gas in one location to gas in another

• Storage terminals convert oil today to oil tomorrow

• Refining oil into products

• Blending grain, oil or metals concentrates

Complexity

• Most commodities go through numerous transformations of all 3 types

• Think of the process of transforming oil at the wellhead to gasoline at the pump

• Multiple spatial transformations (VLCC, pipeline, truck)

• Multiple physical transformations (at refinery)• Storage at “break points”

Transformations are Costly

• Transformations are costly• Space: transportation costs• Time: Storage costs• Form: refining/processing costs

Constraints and Bottlenecks

• Marginal transformation costs tend to be increasing, and sometimes non-linear

• Constraints (“bottlenecks”) in the transformation process can create these non-linearities

• As constraints become binding, transformation costs tend to rise steeply in the quantity of transformation undertaken

Some Examples: Spatial Transformations

• Pipeline capacity

• Transmission capacity (e.g., thermal, voltage limits)

• Rail-loading capacity

• Port capacity

Some Examples: Temporal Transformations

• Grain or oil storage capacity limits

• Limits on the rate of inflow to or outflow from natural gas storage facilities

• The impossibility of time travel: can take a commodity from the present to the future via storage, but can’t take the commodity from the future to the present

Some Examples: Transformations in Form

• Generating capacity constraints in electricity

• Refining capacity constraints

• Blending

Regulatory Bottlenecks

• Regulatory factors are an increasingly important source of bottlenecks.

• Gasoline formula regulations that vary by geographic region (e.g., Midwest)

• Renewable Fuel Mandates

• Export bans

• Rail safety regulations

Pricing

• Understanding commodity pricing requires an understanding of the transformation process and the role of bottlenecks

• It also requires an understanding of the role of the price system

The Role of the Price System

• A competitive price system aggregates the information held by millions of economic actors

• Competitive prices adjust to direct resources to their highest value uses

• In particular, they adjust to reflect relative scarcity and the importance of constraints/bottlenecks

Spreads Price Bottlenecks

• Transmission/congestion charges price transmission bottlenecks (example: PJM)

• Price of NG transportation and storage prices pipeline and storage bottlenecks

• Crack spread

• Spark spread

• Basis

Spreads Provide Signals on Resource Allocation

• Basis prices quality/locational value differences

• Locational basis will adjust to reflect changes in spatial supply and demand patterns and transportation constraints

• Example: CL basis. Basis relations in WTI (and between WTI and other crudes) have changed dramatically in recent years

Constraints and Spreads

• Since constraints/bottlenecks affect the marginal cost of transformation, they will affect spreads

• Spreads widen as constraints become more binding

• Spreads are more volatile when constraints bind tightly than when they do not

Binding Constraints and Spreads: Transformations in Space

• Natural gas basis widens and becomes more volatile when pipeline constraints bind

• Locational spreads in power markets are wide and volatile when transmission constraints bind

• Grain basis wide and volatile when rail system is congested

Binding Constraints and Spreads: Transformations in Time

• Calendar spreads widen and become volatile when storage space is almost completely utilized (e.g., oil, wheat)

• Backwardations become extreme when inventories drawn down to very low levels (and hence the non-negativity constraint on inventory starts to bind)

Binding Constraints and Spreads: Transformations in Form

• Refining margins (e.g., crack or crush spreads) wide and variable when refining capacity utilization rates are high

• Spark spreads (the margin between the value of power and the cost of fuel to generate it) wide and variable when generation capacity utilization is high

COMMODITY TRADING

Light-Heavy Differential Example

• At the height of the oil price spike in summer, 2008, light-heavy price differentials were very wide and inventories of heavy crude were accumulating (e.g., Iran storing heavy crude in VLCCs)

• Combination of regulation-induced demand (low sulfur diesel), restrictions on supply of light sweet crude due to Nigerian disruptions, and limitations on capacity to process heavier crudes to satisfy demand for low sulfur diesel caused the differential to blow out

Trading

• Spreads and pricing relationships are the essence of much commodity trading

• Trading and managing the risk of such price exposures requires an understanding of the value chain

• There is a big potential payoff to understanding the intricacies of the value chain

US Oil Markets: An Extended Example

• Major changes in North American oil markets in the past decades

• Major supply and demand shocks have affected pricing relationships

• Marginal barrel determines price: where the marginal barrel comes from depends on shifting supply and demand conditions

• Seasonal and secular shifts

An Overview• 1990s-early 2000s: US Midcontinent became a deficit

supply region: marginal barrel was from imports. Oil flow mainly south-to-north. Midwest supplied from Canada

• 2008: Financial crisis led to a substantial decline in demand

• Post-2008: huge increases in output in Midcon, S. TX and W. TX.

• Complete shift in pricing relationships due to bottlenecks

Midcon Infrastructure

Midcon Infrastructure II

Midcon Flows

Cash Basis Relationships

• Midcon prices above prices at GOM by cost of transportation prior to shale boom

• Shale boom created huge bottlenecks: excess supply in Midcon and parts of TX, no way to get it to Gulf, and Midcon refineries operating at full capacity

• Pricing relationships flipped: Midcon at a huge discount to GOM until bottleneck alleviated by reversal of pipelines, addition of rail and barge capacity

• Now the bottlenecks is legal: GOM-Midcon=cost of transportation, but GOM at a discount to foreign crude due to export ban

US Oil Output

0

50000

100000

150000

200000

250000

300000

350000

Jan-

2005

May

-200

5

Sep-

2005

Jan-

2006

May

-200

6

Sep-

2006

Jan-

2007

May

-200

7

Sep-

2007

Jan-

2008

May

-200

8

Sep-

2008

Jan-

2009

May

-200

9

Sep-

2009

Jan-

2010

May

-201

0

Sep-

2010

Jan-

2011

May

-201

1

Sep-

2011

Jan-

2012

May

-201

2

Sep-

2012

Jan-

2013

May

-201

3

Sep-

2013

Jan-

2014

May

-201

4

Sep-

2014

Jan-

2015

May

-201

5

Sep-

2015

PADD 2 Output

0

10000

20000

30000

40000

50000

60000

70000

Jan-

2005

May

-200

5

Sep-

2005

Jan-

2006

May

-200

6

Sep-

2006

Jan-

2007

May

-200

7

Sep-

2007

Jan-

2008

May

-200

8

Sep-

2008

Jan-

2009

May

-200

9

Sep-

2009

Jan-

2010

May

-201

0

Sep-

2010

Jan-

2011

May

-201

1

Sep-

2011

Jan-

2012

May

-201

2

Sep-

2012

Jan-

2013

May

-201

3

Sep-

2013

Jan-

2014

May

-201

4

Sep-

2014

Jan-

2015

May

-201

5

Sep-

2015

PADD 2 Crude Production

North Dakota Output

0

5000

10000

15000

20000

25000

30000

35000

40000

Jan-

2005

May

-200

5

Sep-

2005

Jan-

2006

May

-200

6

Sep-

2006

Jan-

2007

May

-200

7

Sep-

2007

Jan-

2008

May

-200

8

Sep-

2008

Jan-

2009

May

-200

9

Sep-

2009

Jan-

2010

May

-201

0

Sep-

2010

Jan-

2011

May

-201

1

Sep-

2011

Jan-

2012

May

-201

2

Sep-

2012

Jan-

2013

May

-201

3

Sep-

2013

Jan-

2014

May

-201

4

Sep-

2014

Jan-

2015

May

-201

5

Sep-

2015

North Dakota Crude Production

PADD 3 Output

0

20000

40000

60000

80000

100000

120000

140000

160000

180000

200000

Jan-

2005

May

-200

5

Sep-

2005

Jan-

2006

May

-200

6

Sep-

2006

Jan-

2007

May

-200

7

Sep-

2007

Jan-

2008

May

-200

8

Sep-

2008

Jan-

2009

May

-200

9

Sep-

2009

Jan-

2010

May

-201

0

Sep-

2010

Jan-

2011

May

-201

1

Sep-

2011

Jan-

2012

May

-201

2

Sep-

2012

Jan-

2013

May

-201

3

Sep-

2013

Jan-

2014

May

-201

4

Sep-

2014

Jan-

2015

May

-201

5

Sep-

2015

PADD 3 Crude Output

Crude Price Differential

-10

-5

0

5

10

15

20

25

Jan-

2004

Apr

-200

4

Jul-2

004

Oct

-200

4

Jan-

2005

Apr

-200

5

Jul-2

005

Oct

-200

5

Jan-

2006

Apr

-200

6

Jul-2

006

Oct

-200

6

Jan-

2007

Apr

-200

7

Jul-2

007

Oct

-200

7

Jan-

2008

Apr

-200

8

Jul-2

008

Oct

-200

8

Jan-

2009

Apr

-200

9

Jul-2

009

Oct

-200

9

Jan-

2010

Apr

-201

0

Jul-2

010

Oct

-201

0

Jan-

2011

Apr

-201

1

Jul-2

011

Oct

-201

1

Jan-

2012

Apr

-201

2

Jul-2

012

Oct

-201

2

Jan-

2013

Apr

-201

3

Jul-2

013

Oct

-201

3

$/bb

l

PADD 3-PADD2 RAC Differential

US Prices vs. Brent

-35

-30

-25

-20

-15

-10

-5

0

5

10

1/1/

2004

5/1/

2004

9/1/

2004

1/1/

2005

5/1/

2005

9/1/

2005

1/1/

2006

5/1/

2006

9/1/

2006

1/1/

2007

5/1/

2007

9/1/

2007

1/1/

2008

5/1/

2008

9/1/

2008

1/1/

2009

5/1/

2009

9/1/

2009

1/1/

2010

5/1/

2010

9/1/

2010

1/1/

2011

5/1/

2011

9/1/

2011

1/1/

2012

5/1/

2012

9/1/

2012

1/1/

2013

5/1/

2013

9/1/

2013

Axi

s Ti

tle

PADD II and PADD III RAC Minus Brent

RAC2-Brent

RAC3-Brent

WTI-Brent Futures Spreads

-25

-20

-15

-10

-5

0

5

10

1-Ju

l-02

1-D

ec-0

2

1-M

ay-0

3

1-O

ct-0

3

1-M

ar-0

4

1-A

ug-0

4

1-Ja

n-05

1-Ju

n-05

1-N

ov-0

5

1-A

pr-0

6

1-Se

p-06

1-Fe

b-07

1-Ju

l-07

1-D

ec-0

7

1-M

ay-0

8

1-O

ct-0

8

1-M

ar-0

9

1-A

ug-0

9

1-Ja

n-10

1-Ju

n-10

1-N

ov-1

0

1-A

pr-1

1

1-Se

p-11

1-Fe

b-12

1-Ju

l-12

1-D

ec-1

2

1-M

ay-1

3

1-O

ct-1

3

1-M

ar-1

4

1-A

ug-1

4

1-Ja

n-15

1-Ju

n-15

1-N

ov-1

5

1-A

pr-1

6

CL-CB Spread (Nearby & 12 Month)

Nearby

12 Month

WTI-Brent Spot Spread

-30

-25

-20

-15

-10

-5

0

5

10

Jan

02, 2

009

Apr

02,

200

9

Jul 0

2, 2

009

Oct

02,

200

9

Jan

02, 2

010

Apr

02,

201

0

Jul 0

2, 2

010

Oct

02,

201

0

Jan

02, 2

011

Apr

02,

201

1

Jul 0

2, 2

011

Oct

02,

201

1

Jan

02, 2

012

Apr

02,

201

2

Jul 0

2, 2

012

Oct

02,

201

2

Jan

02, 2

013

Apr

02,

201

3

Jul 0

2, 2

013

Oct

02,

201

3

Jan

02, 2

014

Apr

02,

201

4

Jul 0

2, 2

014

Oct

02,

201

4

Jan

02, 2

015

Apr

02,

201

5

Jul 0

2, 2

015

Oct

02,

201

5

Jan

02, 2

016

Apr

02,

201

6

Jul 0

2, 2

016

Oct

02,

201

6

Jan

02, 2

017

Apr

02,

201

7

Jul 0

2, 2

017

Oct

02,

201

7

Jan

02, 2

018

Apr

02,

201

8

WTI-Brent

Bottlenecks and Spread Volatility

-4

-2

0

2

4

6

8

WTI-Midland Spread Change

Commodity Flows & New Infrastructure

• Dramatic reversal in direction of oil flows• Utilization of rail, barge and even truck to circumvent

bottlenecks• Displacement of imports, especially from Nigeria (shale

oil light and sweet)• Reversal of pipelines, construction of new pipelines• Increased product exports to circumvent ban on crude

exports, increased investment in refining capacity• Building of “splitters” (“mini-refineries”) to circumvent

ban on crude exports

Crude Shipments

0

10000

20000

30000

40000

50000

60000

70000

80000

90000

100000

Jan-

2007

Apr

-200

7

Jul-2

007

Oct

-200

7

Jan-

2008

Apr

-200

8

Jul-2

008

Oct

-200

8

Jan-

2009

Apr

-200

9

Jul-2

009

Oct

-200

9

Jan-

2010

Apr

-201

0

Jul-2

010

Oct

-201

0

Jan-

2011

Apr

-201

1

Jul-2

011

Oct

-201

1

Jan-

2012

Apr

-201

2

Jul-2

012

Oct

-201

2

Jan-

2013

Apr

-201

3

Jul-2

013

Oct

-201

3

Jan-

2014

Apr

-201

4

Jul-2

014

Oct

-201

4

Jan-

2015

Apr

-201

5

Jul-2

015

Oct

-201

5

PADD 3 to PADD 2 Crude Shipments

Crude Shipments

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

Jan-

2008

Apr

-200

8

Jul-2

008

Oct

-200

8

Jan-

2009

Apr

-200

9

Jul-2

009

Oct

-200

9

Jan-

2010

Apr

-201

0

Jul-2

010

Oct

-201

0

Jan-

2011

Apr

-201

1

Jul-2

011

Oct

-201

1

Jan-

2012

Apr

-201

2

Jul-2

012

Oct

-201

2

Jan-

2013

Apr

-201

3

Jul-2

013

Oct

-201

3

Jan-

2014

Apr

-201

4

Jul-2

014

Oct

-201

4

Jan-

2015

Apr

-201

5

Jul-2

015

Oct

-201

5

PADD 2 to PADD 3 Crude Shipments

Rail Shipments

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

May

-201

0

Jul-2

010

Sep-

2010

Nov

-201

0

Jan-

2011

Mar

-201

1

May

-201

1

Jul-2

011

Sep-

2011

Nov

-201

1

Jan-

2012

Mar

-201

2

May

-201

2

Jul-2

012

Sep-

2012

Nov

-201

2

Jan-

2013

Mar

-201

3

May

-201

3

Jul-2

013

Sep-

2013

Nov

-201

3

Jan-

2014

Mar

-201

4

May

-201

4

Jul-2

014

Sep-

2014

Nov

-201

4

Jan-

2015

Mar

-201

5

May

-201

5

Jul-2

015

Sep-

2015

Nov

-201

5

PADD 2 to PADD 3 Crude Oil by Rail

The Bottlenecks are Back

-35

-30

-25

-20

-15

-10

-5

0

5

10

15

WTI Cushing v. WTI Midland, LLS & Canadian Select

Midland-WTI LLS-WTI Canada-WTI

Contango Example

• Demand collapse in aftermath of financial crisis and inflexibility of supply response in the short run caused huge crude inventory builds, including in US Midcontinent, especially Cushing

• Storage space effectively constrained• Contango (the implicit price of storage) on WTI

blew out• Also blew out on WTS—so it was a storage capacity

issue, not a WTI/futures issue

Supercontango

CL Supply of Storage

Market Responses

• Substantial increase in storage capacity at Cushing

• Using VLCCs to store oil

Grains in the US

• In 2006-2008, spreads between the grain futures prices (and the prices of shipping certificates) and cash grain prices in the country (e.g., elevator bids) reached historically high levels

• This differential should reflect the cost of transforming stored grain to grain on a barge

• What is the bottleneck?

• Huge inventory build (driven in part by impending renewable fuel mandates)

• Anything else?

RINs

• Renewable ID Numbers (“RINs”) provide an example of how a regulatory bottleneck can affect pricing

• Congress mandated increasing use of biofuels (e.g., ethanol) but decline in gasoline usage and technical limitations on the amount of ethanol standard engines can use (“the blend wall”) caused dramatic increase in the demand for unused certificates issued in prior years

• Huge price spike

Hitting the (Blend-) Wall

0

0.5

1

1.5

2

2.5

RINS Prices

D6-1

D6-2

D4

LNG: EVOLUTION OF A NEW TRADED MARKET

LNG

• First cargoes of LNG shipped in 1964

• LNG liquefaction characterized by expensive investment, large scale

• In early stages of development, one plant’s capacity sufficient to satisfy uncontracted demand

• Lumpy additions to capacity

Transactions Costs

• This situation creates two sided small numbers bargaining situations

• Buyers can hold up sellers and vice versa

• Long term contracts a way of mitigating holdup problems

• Use oil price linkage to reduce pricing disputes

Contractual Specificity

• Feedback cycle

• Long term contracts tie up buyers and sellers, so when a new plant comes on-line, new plant and buyers are in a small numbers bargaining situation

• “Contractual specificity”

• So new plant enters into long term contracts, and the process is self-perpetuating

Coordination Game

• There are enough buyers and sellers to break the small numbers bargaining situation

• But it is necessary to coordinate a movement from the long term contract equilibrium to the short term contracting equilibrium

• Coordination always hard to achieve, and existing contracts impede that coordination

Breaking the Cycle

• Need to have multiple plants competing for multiple buyers

• Market developments have facilitated this

• Slower than anticipated demand growth

• Large amounts of capacity coming online (e.g., Australia, US)

• Excess supply: “free cargoes”

Breaking the Cycle II

• Shocks created short term needs for LNG (Egypt gas decline, Amazonian drought)

• Japanese overbought and desire to escape contractual obligations: government is obliging by trying to eliminate “destination clauses”

• This has created additional short-term trading

A New Cycle

• The liquidity cycle

• Liquidity builds liquidity

• In liquid markets buyers (sellers) can rely on markets for security of supply (demand)

• This breaks the small numbers bargaining problems

• If there is a liquid spot market, even big, long-lived assets don’t need long term contracts (e.g., refineries)

An Experimental Space

• Blockchain has burst upon the scene, and interest has spread rapidly from the cryptocurrency space to virtually every industry

• Many elaborate claims about how blockchain will revolutionize everything from A to Z

• FOMO

• Now businesses are in the process of experimenting to develop practical applications

Blockchain & Supply Chains

Many Attempts, Many Failures

• GitHub (a software collaboration platform) data shows the extent of experimentation

• Since 2009, 86,034 BC projects, 9,375 by companies & research institutions

• 28,885 in 2016 alone—pace obviously accelerating

• Only 8 percent of projects started are currently active

• Average longevity 1.22 years

Signs of a Burgeoning Industry

• These patterns (many attempts, many failures) are characteristic of a newly developing, rapidly evolving industry

• Indicative of a highly promising technology that is not fully understood, and perhaps over-hyped

• Much trial-and-error• Evolutionary explosion• Likely to be followed by a narrowing and deepening

evolutionary process

Alternative Visions: Disintermediation

• Blockchain will disintermediate—that is it will render existing intermediaries like banks obsolete and superfluous: peer-to-peer supplants reliance on trusted third parties– Will probably depend on function, e.g., payments vs.

supplying credit

• This is a more radical vision, and there is room for considerable skepticism (more on this later)

Alternative Visions: Improving the Efficiency of Intermediation

• Using blockchain to change the way existing firms/intermediaries perform their legacy functions

• Performing the same basic functions, but doing so more efficiently

• The most mature blockchain initiatives are of this type

A Conceptual Framework

• The conceptual underpinnings of blockchain can help one understand its most immediately promising applications

• A secure, immutable ledger (database) that multiple parties can view and add to—a way of sharing information – Open/unpermissioned

– Closed/permissioned

Supply Chain Within a Single Entity

• Tracking a product or part that moves through many processes/locations within a single firm

• Employees update the blockchain as product/part moves through the firm– Cargill (turkeys)

– Maersk (shipping containers)

• Better inventory and quality control

• Source of data to identify process bottlenecks

Commodity Trade Involving Multiple Parties

• Commodity trades are complex, and involve many parties– Buyer, seller (and a given lot can change hands multiple

times), banks (usually two or more), transporters (often several, e.g., truck, rail, ship), inspectors, warehouses

• Information currently diffused, often on paper documents that can be lost or altered, and which are cumbersome to transfer

• Participants can document their actions by updating blockchain

• Experiments in oil, cotton, and grain trading

Recording and Transferring Asset Ownership

• Record asset ownership on a blockchain

• Update ownership record when asset is sold

• Digitize assets

• Subdivide assets into fractional shares

• Digital gold (CME & Royal Mint)

Smart Contracts

• Many transactions involve contingencies: if X happens, do Y– Simple: make a fixed payment on the 15th of every

month

– More complex: size of payment depends on a market price (a swap contract, variation margin)

• Smart contracts implemented on blockchain can automate this process

Devilish Details

• Participation

• Permissioning: who can do what?

• Organization, ownership, control, and governance

• Scaling and network effects

• The ongoing experimentation is in large part focused on attempting to master these details in particular applications

Trust and Disintermediation

• One of the hyped benefits of some BC applications (e.g., Bitcoin) is the elimination of need for trusted intermediaries

• No free lunch• These applications involve relatively simple forms of

opportunistic behavior (e.g., double spend)• Many commercial applications (e.g., commodities

trades) involve more varied and complex misbehaviors• This is why I believe most successful implementations of

BC will involve trusted intermediaries to operate more efficiently, rather than disintermediation

Functional Analysis

• When evaluating BC applications, it is essential to ask:– What economic function is being performed?

– How does BC improve the efficiency of performing this function?

– Does BC permit a change in who performs the function?