Can pairs trading act like arbitrage to enforce the law of one price?

This thesis confirms empirically that a pairs trading strategy can act to enforce a market price ratio between two closely-related securities, thereby behaving like arbitrage enforcing the law of one price. Any empirical study of arbitrage and pairs trading is difficult because security exchanges report trades anonymously. Trades in two securities cannot be identified as originating from the same trader, nor can the entry and exit trades of a strategy be matched. Much of this thesis is spent designing a methodology for detecting pairs trading through statistical techniques and testing its accuracy in situations where pairs trading can be anticipated. The approach works pleasingly despite being imprecise. It can isolate closely-related pairs in pseudo-random sets of similarly-traded pairs and locate fundamentally similar pairs among large sets formed by pairing index constituents. It can even detect pairs involved in merger arbitrage many months ahead of company mergers. The evidence then shows pairs trading in the twin depositary receipts of BHP Billiton (the world's largest mining conglomerate with dual-listed company structure) act to push the prices apart when the difference falls below a recent measure, instead of reinforcing their trend towards equality. The detected pairs trades enforce the price difference: reducing it when it becomes too large and increasing it when too small. Evidence rejects a trading history of restoring price equality even though these security twins seem to offer identical dividends, risks, and entitlements. Pairs trading may be enforcing the market price ratio as a proxy for an incalculable valuation ratio underlying the law of one price.