The Commodity Value Chain

Introduction

We can come up with a framework for thinking about commodity markets. This makes use of the humanistic framework in my other post (Metacognition: Thinking about Thinking). The key idea is to build a mental model of flows across nodes, but with a physical overlay: geography, quantity, etc.

Value Chain

• Humanistic framework, we ask who (nodes), then build a nice network.
• In commodities, we overlay a physical/geographical element over the nodes: what (type), quantity (how much), where (geography).
• Logistics/necessity premium - The spread captured by the trader, reward for providing liquidity to consumers for utility
• Need to visualise it in our head: the nodes, their context, and their physical overlay.
• Build a mental model in your head. React to information, simulate or predict it. Then you act on it to make the trade to capture logistics premium.
• When reading or studying or interacting with commodities, you really are enhancing your mental model and making it richer (e.g more nodes, more context).

Format

• Node Levels
  • Country, state, company, facility
• Node Types
  • Production
  • Processing
  • Storage
  • Distribution
  • Consumption
• Node Context Per Node
  • Location (where)
  • Ownership (who)
  • Quantity
  • Flows (goes where). How it evolves over time (e.g historical state, current state).
• Node Predictors
  • State & future events. Aka predict flows and as a trader capitalize on the logistics premium.

Examples

• Trader of crude/products in ME. You make money by spread.
• Situation - Israel conflict.
• Analysis - If Israel strikes Iranian oil export terminal (e.g Kharg Island), that wipes out their exports. We know crude flows from Iran to China heavily (Orrin). So you say OK, those ‘lost barrels’ will source more from elsewhere.

Oil & Gas

• Node Types
  • Production - Fields (production), refineries, terminals, caverns, pipelines, ports, ships, distribution, petrol stations, airports, facilities
  • Processing - Refineries
  • Storage - Terminals, caverns
  • Distribution - Pipelines, ships, ports
  • Consumption - Airlines, cars, industrial vehicles, burners, power plants, petrochemical plants, etc.
• Node Context
  • Location (where)
  • Ownership (who)
  • Quantity/Operations/State
  • Flows (goes where)

Supply & Demand Quantities/Factors

We can further split each supply or demand quantity into individual quantities in a hierarchical fashion, e.g based on geography (a natural grouping).

A possible hierarchy for quantities could look like this. A quantity would be measured by some variable: say demand/supply/imports/exports/inventory.

Region (Q) $\rightarrow$ country (Q) $\rightarrow$ companies/players (Q) $\rightarrow$ production/consumption node (Q) $\rightarrow$ factors.

Then these quantities have beta to individual factors. These are latent drivers of quantities. For example, seasonality, weather, interest rates, OPEC policy, geopolitical events, etc. The problem is, while betas to these factors are probably estimable via statistical modelling, the factors themselves are suited to discretionary modelling, and by extension so are the quantities. For example, random shocks could occur (e.g political upheaval or natural disaster) that are not forecastable.

I’d like to come up with a nice diagram for this.

Generalizing further, we can view these as supply nodes and demand sinks and a normal state of flows (arb) that bring commodities over. Whenever factors change, they drive changes in supply/demand nodes, and the flows change accordingly.