Traditional asset pricing models typically have assumptions such as symmetrical information and no transaction costs, and work on the aggregate level, where, e.g., a Walrasian auctioneer sets demand equal to supply at the equilibrium price. However, these models do not say much about intra-day trading activity and how prices are formed in reality. In financial markets, information is asymmetric and traders execute in the market at different times and speed.
The research are of financial market structure, that came to be called “market microstructure” began in the 1980s and provide tools and theory to estimate price discovery, transaction costs, and market liquidity. Initially, there were two types of theories forming: information-based and inventory-based models. The information based models analyze the information asymmetry between different market participants, showing, e.g., how the market makers set the bid-ask spread (trading fee) to compensate for adverse selection and inventory risks. The inventory-based models, sets the liquidity providers in focus, and focuses more closely on the relationship between the bid-ask spread and liquidity provider risks.
Asymmetric information comes from the fact that both informed and uninformed traders participate in the market. Private information about an assets price can come from insider information, costly research, or analyzing the order flow. Uninformed traders trade because they have liquidity needs, e.g., portfolio management such as hedging.
One of the successes of market microstructure has been to evaluate the quality of different market designs and understand the impact on trading costs and price.
The model developed by Kyle (1985) has become a foundational model in market microstructure and shows that order flow drives prices, buy orders increase the price and sell orders decrease the price. Kyle (1985) models an informed trader that maximize profits by choosing the size of orders.
Glosten and Milgrom (1985) study a dealership market with informed and uninformed traders. The asymmetric information causes adverse selection for the dealers, who protects themselves by the bid-ask spread.
There are alternative explanations for why the bid-ask spread arise. For example, in a market with symmetric information but buy and sell orders arrive randomly, the dealer has to absorb order flow and needs to be compensated for inventory risks.
Roll (1984) develops an estimator of transaction costs, and shows that transaction costs cause negative serial correlation in returns. The models, shows that it is possible to estimate the bid-ask spread with returns alone.
Glosten and Harris (1988) estimate transaction costs using intra-day price and trade data.
Overview:
Theoretical:
Empirical: