The Map of the Field: Biais, Glosten and Spatt's Microstructure Survey

Some papers make a discovery. Others do something rarer and, in their way, just as valuable: they take a sprawling, argumentative, half-understood field and impose order on it.

That is what Biais, Glosten and Spatt did in 2005. Three people who had personally built large parts of market microstructure sat down and wrote the map. If you want to understand how the whole field fits together rather than one corner of it, this is the document.

The problem: a field that had grown faster than its own understanding

By 2005, market microstructure was thirty-odd years old and had accumulated a great deal of apparatus.

There was the inventory tradition, from Demsetz through Stoll and Ho, saying spreads exist because dealers get stuck with unwanted risk. There was the adverse selection tradition, from Copeland and Galai through Glosten and Milgrom and Kyle, saying spreads exist because dealers fear the informed. There was a huge empirical literature measuring spreads, decomposing them, tracking price discovery. There was a growing body of work on limit order books, and on how the design of a market shapes the outcomes in it.

And these pieces did not obviously talk to each other. A student could easily learn Kyle's model and Ho and Stoll's model and not really understand which one to reach for, or what happens when both forces operate at once, or what any of it implied for whether an exchange should run a call auction or a continuous book.

Regulators had the same problem, more urgently. Real decisions were being made about market structure, about decimalization, about consolidation versus competition among venues, about transparency rules. What did the theory actually say?

The key idea via analogy: the anatomy lesson

The survey's organizing insight is that the price you actually transact at is not a single thing. It is the fundamental value of the asset, wrapped in several layers of friction, and each layer has its own theory.

You can think of it like peeling an onion inward from the price you pay:

The outer layer is the order-processing and inventory layer. Somebody must be persuaded to take the other side right now and hold the position. They must be paid for their costs and for bearing risk they did not want. That is Demsetz, Stoll, Ho. It is a real resource cost, and competition can shrink it but not eliminate it.

The next layer is adverse selection. Whoever quoted that price is exposed to the possibility that you know more than they do. That is Copeland and Galai, Glosten and Milgrom, Kyle. It is not a cost in the same sense at all: it is a transfer from the uninformed to the informed, mediated by the market maker who prices it in. This distinction matters intensely for policy, and the survey hammers it. Competition cannot compete away adverse selection, because it is not a margin, it is a loss that must be funded.

The next layer is strategic behaviour. Traders are not passive. Informed traders hide, splitting orders and timing them. Liquidity traders choose when and where to trade to avoid the informed. Market makers infer, and the informed anticipate the inference. The equilibrium is a game, not a mechanism.

The outermost frame is market design. All of the above happens inside a set of rules that somebody chose: continuous trading or periodic auctions, a dealer market or a limit order book, transparent or opaque, one venue or many, what the tick size is, who gets priority. The rules are not neutral. They determine which frictions bite and who bears them.

That last point is the survey's real thesis, and it is why the paper is titled the way it is. The three sections, microfoundations, empirical results, and policy implications, are not three separate topics. They are one argument: theory tells you what forces are at work, evidence tells you how big they are, and together they tell you how to design a market.

Why it mattered

• It is the standard entry point to the field. For twenty years, "read Biais, Glosten and Spatt" has been the answer to "how do I get up to speed on microstructure?" Alongside O'Hara's textbook, it is how a generation of researchers and practitioners learned the landscape.
• It made the policy connection explicit. Microstructure is unusual among finance subfields in that its findings directly determine regulation. Tick sizes, transparency rules, consolidation, best execution, the treatment of dark venues: all of these are microstructure questions with real money on them. This survey is one of the clearest statements of what the theory does and does not license regulators to do.
• It clarified the central dichotomy. The survey is very good at insisting that an inventory cost and an adverse selection cost are different animals, even though both show up as a wider spread. One is a payment for a service, the other is a redistribution. Confusing them leads to bad policy, and the survey is the standard citation for keeping them apart.
• It legitimized the limit order book as a theoretical object. The synthesis of the work on order-driven markets, much of it by the authors themselves and their collaborators, helped consolidate the modern view that the limit order book is not a plumbing detail but the central institution of price formation.

The honest limitations

• It is a map, not a discovery. Nothing in it is new. Its value is entirely in the organization and synthesis, which is a real contribution, but if you are looking for a result, look elsewhere.
• It predates the machines. This is the big one. Published in 2005, the survey was written just before the high-frequency era truly landed. It does not contain the arms race, latency arbitrage, colocation, maker-taker fee structures, the explosion of venue fragmentation, the flash crash, or the modern debate about speed bumps and batch auctions. O'Hara's 2015 survey exists precisely because so much changed so fast after this one.
• It reflects its authors' priors. It is a survey by three theorists who built the theory, and it is, unsurprisingly, most confident and most detailed on the parts of the field they built. The econophysics tradition, with its empirical laws of impact and order flow, gets short shrift.
• The empirical section has aged. Estimates of spread components, of price impact, and of adverse selection from the 1990s describe a market that no longer exists. The mechanisms endure. The numbers do not.

The one-line takeaway

Biais, Glosten and Spatt drew the field's map: the price you pay is the true value plus a payment for someone's inventory risk, plus a charge for the chance you are better informed than them, all playing out inside a set of trading rules that somebody chose, and the point of microstructure is to understand all three layers well enough to choose those rules wisely.