Overview
The no-arbitrage principle states that in competitive, well-functioning markets there should be no way to construct a costless (or negative-cost) position that yields a nonnegative payoff in every possible state of the world and a strictly positive payoff in at least one state. Put simply: there should be no “free money” trades. This principle disciplines relative prices across goods, locations, dates, and states of the world. It underpins internal consistency in price lists, the law of one price, and the valuation of financial contracts by replication, and it guides practical corporate decisions on pricing, financing, and procurement.
Origins and Credit
Arbitrage reasoning appears in classical price theory (Cournot, Marshall) and in general equilibrium via Arrow and Debreu’s state prices (1954–59). In modern finance, the principle is formalized by the fundamental theorem of asset pricing (Harrison and Kreps; Harrison and Pliska, late 1970s–early 1980s), which links no-arbitrage to the existence of a linear pricing rule or “risk-neutral” probabilities. It also anchors Ross’s arbitrage pricing theory (1976) and many parity relations used in corporate finance and derivatives.
Core Idea and Mechanics
An arbitrage is a trading strategy with three features: (1) zero or negative initial cost, (2) payoffs that are never negative across all states, and (3) strictly positive payoff in at least one state. If such a strategy exists, competitive traders would scale it up, pushing prices until the opportunity disappears. The no-arbitrage principle thus imposes tight relationships among prices:
- Law of one price: Economically identical payoffs must have the same price. If a payoff X can be replicated by a portfolio Y, then Price(X) = Price(Y).
- Linear pricing: With no-arbitrage, there exists a set of state prices (or a “stochastic discount factor”) such that the price of any payoff equals the sum of its state-contingent payoffs times those state prices. In derivative markets, this is often expressed as risk-neutral valuation: price equals discounted expected payoff under an adjusted probability measure.
- Bounds and inequalities: If payoff A is at least as good as payoff B in every state, then Price(A) ≥ Price(B). Super- and sub-replication provide price bounds when exact replication is not possible.
- Parity relations: Familiar identities—such as put–call parity for options or covered interest parity in FX—are direct consequences of no-arbitrage and replication.
Key Assumptions and Conditions
- Comparable, enforceable contracts: Payoffs are clearly defined; claims are honored (no hidden default risk unless priced).
- Low frictions: Transaction costs, taxes, and trading delays are small relative to price gaps; otherwise, only “within a band” no-arbitrage applies.
- Feasible trading technologies: Ability to go long/short, borrow/lend, and scale trades; position limits or short-sale constraints can block arbitrage.
- Common information timing: Prices refer to the same time and information set; stale quotes can create temporary disparities.
- Limited balance sheet constraints: Arbitrageurs have capital and risk management capacity to execute trades.
Implications
- Internal consistency checks: Price lists across SKUs, bundles, service levels, and financing terms should not permit costless “buy low, sell high” constructions, including cross-channel or cross-border resale.
- Valuation by replication: Forward prices, swaps, and many insurance-like contracts can be priced from more primitive instruments (spot, financing, storage, and carry costs) without invoking investor risk preferences.
- Risk-neutral pricing in finance: Derivative values are present values of expected payoffs under risk-neutral probabilities; this isolates pure pricing relations from risk premia estimation.
- Policy and contract design: Well-designed policies (e.g., deposit insurance premia, emissions markets) avoid built-in arbitrage that would generate windfalls or unintended gaming.
- Operational guardrails: Terms and conditions (warranties, geo-fencing, non-resale clauses) are often needed to prevent gray-market arbitrage when firms pursue price discrimination.
Example in Practice
Trade credit and early-payment discounts. A supplier offers “2/10, net 30”: a 2% discount if the invoice is paid within 10 days; otherwise, the full amount is due in 30 days. For a $100 invoice, paying on day 10 costs $98; delaying to day 30 costs $100. The implied 20-day return from not taking the discount is $2 on $98, roughly 2.04% for 20 days. Annualized (simple) this is about 2.04% × (365/20) ≈ 37%.
If the buyer can borrow at, say, 8% per year, it can borrow $98, pay on day 10, and repay the loan on day 30 for about $98 × (1 + 0.08 × 20/365) ≈ $98.43. This locks in a saving of roughly $1.57 versus paying $100 on day 30—a near-arbitrage after considering minimal credit risk and administrative costs. In a competitive environment, such rich terms will not persist broadly: either buyers take the discount (eliminating the “free” return), suppliers adjust terms (smaller discount or shorter window), or financing frictions (credit limits, risk) prevent full exploitation.
The same logic scales to capital markets. Put–call parity (for a non-dividend-paying stock) requires C − P = S − PV(K). If this fails, a trader can replicate the underpriced side with the overpriced side and a bond to earn a riskless gain until prices realign.
Limitations and Common Misunderstandings
- Frictionless ideal: Transaction costs, taxes, inventory risk, and delays create “no-arbitrage bands.” Small deviations are consistent with no-arbitrage once costs are included.
- Limits to arbitrage: Capital constraints, short-sale bans, model risk, and “noise trader” risk can let apparent mispricings persist, even for extended periods.
- Not about expected profit: True arbitrage is riskless in all states. “Stat arb” or carry trades earn positive expected returns but can lose money; they are not arbitrage in the strict sense.
- Does not fix absolute prices: No-arbitrage pins down relationships among prices. Risk premia and preferences still determine expected returns and levels where replication is not exact.
- Product non-equivalence: Differences in service, warranties, or legal restrictions mean two items are not identical; price gaps may then be justified and non-arbitrageable.
- Credit and counterparty risk: If payoffs are uncertain due to default or settlement risk, apparent arbitrage may vanish once those risks are priced.
Related Concepts
- Law of one price
- Fundamental theorem of asset pricing
- Put–call parity
- Covered interest parity
- State prices and Arrow–Debreu securities
- Risk-neutral valuation
