What makes a price index trustworthy.

An index is only useful if people can rely on it.
What earns that trust: governance, data, method and accountability.

Carry Research5 min read

A price index turns many prices into one number. Inflation is measured with one. Oil cargoes, home loans and pension funds are priced against them. When a number is used to settle contracts worth billions, whether it can be trusted is a practical question.

What a price index does

Any market with many sellers has many prices. Grades differ, delivery terms differ, and prices change through the day. An index collects those prices, applies a set of rules and publishes a single value at regular intervals.

The rules matter as much as the data. They decide which prices count, how each one is weighted, how unusual prices are treated and what happens when data is missing. Two indices built from the same prices with different rules can tell different stories.

Once an index is widely used, it becomes a benchmark: a reference price that contracts, budgets and financial products are written against. At that point many people depend on a number that one organization produces.

The data behind an index

Indices draw on different kinds of data, and each brings its own strengths and risks.

  • Transactions. Prices at which trades actually took place. They are the hardest to dispute, but some markets have too few of them.
  • Firm quotes. Prices at which someone is committed to trade. They are close to the market, but may not show where business is actually done.
  • Published prices. Rates that sellers post for anyone to buy at. They are public and easy to check, but some deals are done at a discount to them.
  • Assessments and submissions. Judgments by specialists, or estimates from market participants, used where trading is thin. They fill gaps, but depend heavily on the controls around them.

Whatever the inputs, a benchmark's rules have to say how they are chosen and what happens when the best data is scarce.

A lesson from interest rates

For decades, LIBOR was one of the most important numbers in finance. It set the interest on trillions of dollars of loans and derivatives, and it was based on banks' own estimates of what they would pay to borrow. Investigations after the financial crisis found that some of those estimates had been manipulated.

The response shaped how benchmarks are run today. In July 2013 the International Organization of Securities Commissions (IOSCO), the global body of securities regulators, published its Principles for Financial Benchmarks. Benchmark rules introduced since then in the European Union and the United Kingdom draw on the same ideas.

Commodity benchmarks went through a similar review. Many physical commodities, from crude oil to fuel, are priced against assessments published by price reporting agencies. In October 2012, IOSCO published principles for oil price reporting agencies, setting out how those assessments should be made and governed.

Four foundations of trust

The IOSCO principles, 19 in all, fall into four groups. Together they describe what a trustworthy benchmark looks like.

  • Governance. Someone is clearly responsible for the benchmark, with oversight that is independent of the people who produce it, and with conflicts of interest identified and managed.
  • Quality of the benchmark. The index measures what it says it measures, using enough data from the market it describes, and gives priority to prices from real transactions where they exist.
  • Quality of the methodology. The rules are written down, applied consistently and changed only through a defined process, with users told in advance. There is a plan for what happens if the benchmark has to change or stop.
  • Accountability. There is a way to raise complaints, records that trace every published value back to its inputs, and audits that check the rules were followed.

None of these makes a number right by itself. Together they make it reliable: users can understand what the index represents, trust that it is produced the same way every day, and know that changes will not arrive by surprise.

When a benchmark has to change or end

No benchmark lasts forever unchanged. Markets evolve, data sources disappear and better methods emerge. A trustworthy benchmark plans for this.

Changes to the rules are consulted on or announced in advance, so that users can prepare. Contracts that reference a benchmark include fallbacks: what happens if it is not published on a given day, or ever again. The end of LIBOR showed why this matters. Moving trillions of dollars of contracts to new reference rates took years of work by regulators, banks and borrowers, and contracts with clear fallback terms moved far more smoothly than those without.

Questions to ask about any index

For anyone deciding whether to rely on an index, a few questions go a long way.

  • Who runs it. A named administrator, with oversight independent of the people who produce it.
  • What it measures. A clear definition of the thing being priced, and evidence that the index tracks it.
  • How it handles gaps. Written rules for what happens when data is missing or a market is disrupted.
  • How it changes. A written process for changing the rules, with notice to users.
  • What is kept on record. Each version of the rules and each day's values, so that past results can be traced.
A benchmark earns trust the way a ruler does: by measuring the same way every time.

Why it matters for new markets

Mature markets built their benchmarks over decades. New markets have to build them quickly, and the early choices last. A benchmark designed without clear governance is hard to fix once contracts depend on it.

AI inference is a new market of exactly this kind. Prices change often, the products improve constantly, and companies now spend large sums on it. A reference price for inference is useful only if buyers, sellers and the institutions around them can rely on it. Why AI inference needs a benchmark sets out the case.

How Carry approaches it

At Carry, governance came first. An index governing board provides independent oversight of the methodology, its changes and its use as a benchmark. The methodology, and any change to how the indices are built, is approved before it takes effect, and each methodology version is frozen and kept on record. Each day's values are then calculated by Carry's proprietary algorithm and published automatically by that approved process.

The full rulebook is shared with members of the governing board under a confidentiality agreement. Carry Indices describes the governance in more detail.

This article is a general introduction to price indices and benchmark governance. It is not legal advice, and it does not describe how any particular index is calculated.

Newsletter

New research, in your inbox.