How We Think About Commerce

A cheese maker choosing between cheese and powder, a grain merchant deciding whether to store or ship, a resin buyer timing an order and a freight network pricing a lane look like four different problems. They are one problem: choosing a route through a web of conversions, at prices the market sets. Get the structure right once, and each new problem is a variation.

This page is the map of that structure. Each dot is one idea. Select one to read it, or regroup the map to see the same ideas from another side. Later posts take them one at a time.

Group by

The connected firm

An Erlangen programme for commerce

Klein sorted geometries by what stays the same. We sort commerce by what a conversion keeps: mass, components, value.

Concept: The Erlangen view. Buying, making, selling and hedging look different but follow the same rules. Learn the rules once.
Origin: Our viewBuilds on geometry (Klein, 1872).
Module: The connected firm
Value: Hours
Thinkers: Klein

Figure 1

The idea map

The ideas behind commercial intelligence, arranged around one: every business is a web of conversions. Regroup them by theme, module, novelty or value channel, focus a group, and select an idea to read it.

Read the map by its links

A list of topics is not insight. The insight is in the links between groups that sit far apart on the map, where two teams in one company solve the same problem without knowing it. These are the six links we will test first.

From	To	The link
Proof and method	Markets and price	Purchase price variance is the profit and loss of a position taken by default.
Selling finished goods	Decisions under uncertainty	A fixed price quote is an option, and it can be priced.
The physical world	Arbitrage and replication	An empty truck and liquid whey are the same thing: something a business makes anyway, worth little until someone finds its route.
Decisions under uncertainty	Six Sigma outside the walls	Flexibility is long volatility, and control limits decide how much of it to keep.
The connected firm	Markets and price	One clock for every team lifts profit and cuts risk at the same time.
The ecosystem and its rules	Selling finished goods	Whoever writes the rules of a tender decides much of the price before the bids arrive.

Four claims

Each claim below changes a number on the profit and loss, and each comes with a diagram that shows why, drawn the way mathematicians draw them: things as words, changes as arrows.

1. Price and decision are one number. An input is worth the value of the best thing you can do with it. That is standard in the mathematics of decisions and in option pricing. It is rare in operations, where an input costs what was paid for it.

Figure 2

Two readings of one calculation

One calculation, the Bellman equation, gives two answers. Its value is the price of what you hold. The route that reaches that value is the decision.

2. A hedge is a conversion. It moves price risk to someone who wants it, and leaves the local gap and the cash with you. Trading houses run their books this way. Many manufacturers treat a hedge as a separate financial act, judged by how much of the book it covers.

Figure 3

Who holds which risk

Before the hedge you carry both risks. Selling futures passes the price risk, through the exchange, to the buyer of your futures. After it, you keep only the local gap, called basis.

3. Hedging is quality control. The discipline operations already use for quality, limits, charts and root causes, applies just as well to the prices that cross their walls. The table further down pairs each tool with its twin.

4. The recipe should drive the hedge. When one machine makes two products, they are one exposure, not two. A recipe and a hedge portfolio are the same arithmetic in different words, so operations and the risk desk should read one model.

Figure 4

The recipe sets the hedge

Size the futures with your own yields, and their gain or loss offsets that of operations. Together they fix the margin at today's futures prices. What remains is basis: the gap between local prices and the exchange.

The concepts, and how they connect

The first map lists ideas. This one shows the dozen things every idea is about, and how each leads to the next. Start at the word closest to your job, and follow the arrows.

Figure 5

The concept map

The kinds of thing every idea above is about, and how one leads to another. Blue arrows are our own claims. The dashed pair is the central one.

What we borrow, what is new, where we disagree

Kind	Example	Where it is standard
Borrowed	Forward curves, basis, value at risk	Trading houses and banks
Borrowed	Control charts, root cause analysis	Quality teams in operations
Borrowed	Selling at the split off point	Cost accounting
Our view	Six questions about every resource; price as the value of the best route	Our synthesis
A different view	Judge a hedge by the risk left, not the share covered	Differs from common practice in manufacturing
A different view	Volatility pays the flexible	Differs from the view that price swings only do harm

Definitions

Word	In plain words	What to do
Conversion	Something goes in, something else comes out.	List what goes in and out of every site, truck and warehouse.
Value stream	Every route from a raw input to cash.	Compare the margin of each route.
Offramp	A chance to sell now instead of making more.	At each step, compare selling now with the next step.
Forward curve	Today's price for each future month.	Give every team the same curve, every day.
Basis	Why the same thing costs different amounts in two places.	Track it by location.
Price range	When no one trades your exact product, the honest price is a range.	Quote a floor, a target and a ceiling.
Hedge	Hands the price risk to someone who wants it. You keep the local gap.	Record which risk left and which stayed.
Return on risk	What you expect to earn for each dollar you could lose.	Judge every hedge by it.
Real option	A second recipe, route or supplier you can switch to.	Put a price on every switch.
Arbitrage	Buy low, change it, sell high, after costs.	Rank the gaps after freight, storage and processing.
Earnings at risk	How much profit a bad month could take.	Set a limit and check it daily.
Agreed baseline	The yardstick agreed before work starts.	Count only gains measured against it.

Quality control, outside the walls

Inside operations	In the markets
Specification limits	Risk limits
A defect	A decision made on an old price
Control chart	Daily risk against its limit
Root cause	What moved the profit, and why
Yield	Margin per unit of raw input

Where the ideas come from

Each idea we borrow, and what it changes for an operator.

Things and conversions

Antoine Lavoisier, conservation of mass: nothing is lost, nothing is created. Cream and skim leave the separator together; price them apart and one of the two numbers is wrong.
Felix Klein, the Erlangen programme: you understand a field by what survives its moves. Buying, making, selling and hedging preserve the same things, so they are one subject, not four departments.
Coecke, Fritz and Spekkens, a theory of resources: a price must respect what a thing can become. Milk can become cheese, so milk is never worth less than cheese minus the cost of making it.
Brendan Fong and David Spivak, hypergraph categories: boxes with many inputs and many outputs still connect, and the totals add up. A whole operation, intake to delivery, fits on one page and adds up.

Decisions and prices

Richard Bellman (the Bellman equation) and Warren Powell (sequential decisions): what you hold is worth the best thing you can still do with it, decided step by step as news arrives. Price and decision come out of the same sum.
Black, Scholes and Merton, option pricing: an option is worth what it costs to copy. An operation that can switch, cheese to powder or one lane to another, owns one, and it has a price and a hedge.
Harrison, Kreps and Pliska, the fundamental theorem of asset pricing: when no trade makes money from nothing, prices hang together. In thin markets they hang together only within a range: quote a floor, a target and a ceiling.
E. T. Jaynes, maximum entropy: probability is logic with missing facts, so claim no more than you know. A scenario weight is a stated belief, open to challenge, not a setting buried in a model.
Bouchaud and Potters, heavy tails: real markets jump more often than a bell curve allows. Set limits for the month the textbook calls impossible.

Markets and firms

Craig Pirrong, the economics of commodity trading firms: trading houses earn from moving goods through space, time and form, not from betting on price. They hedge the level and keep the gap, and any operation can do the same.
Ronald Coase and Oliver Williamson, transaction costs: firms own the steps where depending on others costs too much. That is why storage and terminals, the bottlenecks, are where margin collects.
Hurwicz, Maskin and Myerson, mechanism design: design the rules, and the players' own interests deliver the result. Whoever writes a tender sets much of the price before the first bid.
Hal Varian, computer mediated transactions: once every transaction is measured, contracts that were impossible become routine: price formulas, cost pass-through, menus of terms.
Michael I. Jordan: prediction is half of intelligence; the other half is deciding while customers and rivals decide too. A forecast that ignores how they respond is wrong the day you act on it.

Quality

Walter Shewhart and W. Edwards Deming, control charts: separate noise from signal before you act. A daily risk chart with limits does for profit what a control chart does for fill weight.
Six Sigma, from Motorola, DMAIC: define, measure, analyse, improve, control. The loop operations already trust runs just as well on market exposure.

What we do, and what we do not

We help operators decide, and we help them hedge inside their own limits. We do not trade, and we never hold market risk of our own.

The argument behind the map is in Commercial Intelligence: An Erlangen Programme for Commerce.