UC Berkeley

Many articles examine the quality of financial markets, and propose how to improve it. Most often, the goal is to improve the efficiency of the candidate market, specifically operational efficiency, i.e. the goal is to reduce the transaction costs. Although operational efficiency is one of the most important factors that contribute to the quality of financial markets (or at least the perception of it), several other factors play a role. Fairness, informational efficiency, and stability of financial markets are ultra important. My research exhibits the need to set a standard for quality assessment of financial markets. This purpose demands the illustration of the multidimensional nature of the quality of financial markets and the inadequacy of commonly used measures of market quality to capture it.

When exploring the quality of financial markets, contrary to common practice, I do not limit my attention to the sell-side. I have looked at both the sell-side and the buy-side, the impact of market design changes and new policies on their behavior, and the quality implications of it. However, since there is a lot to say on this enormous subject, I have chosen a principle of selection. I have selected three scenarios in which market quality is adversely affected by endogenous excess volatility induced by market participants' rational behavior.

Excess volatility is defined by Shiller (1981) as the movements in real stock prices that cannot be explained by new information about subsequent real dividends, and by LeRoy and Porter (1981) as the fluctuations in asset prices that are more than is consistent with present value models. They both found excess volatility using very different variance-bounds tests. Since then, however, their methods have been subject to many criticisms; from having little to none statistical significance, to having econometric problems, to their hypothesis requiring the risk neutrality assumption. At the same time, a lot has been done trying to reconcile financial models to the puzzling levels of volatility observed in financial markets.

I approach excess volatility in the three essays from a different angle. I define it as the movements in asset prices in excess of the changes in the fundamental value of the asset, and since it cannot be measured empirically, I have chosen a theoretical approach so I can examine which frictions or behaviors can cause excess volatility. Also, I have focused my attention on volatility inducing frictions or behaviors for which we can find remedies in the form of market design change or policy. The implications of the theories presented can then be tested by implementing pilot programs in any exchange. One such program, the Securities and Exchange Commission's tick size pilot program, is already underway.

In the first essay, I examine the impact of inventory pressure on a single market maker. I present a continuous-time model of liquidity provision with long-lived information and endogenous inventory control. I show that an (l,I_I,I_u,u) inventory control strategy is optimal. The optimal price depends on the inventory level. Furthermore, the instantaneous cost of holding a position (either long or short) to the market maker and the excess volatility in prices are in a direct relationship. The instantaneous cost can be interpreted as adverse selection cost (risk) of holding any non-zero position, cost of capital for holding long positions, and short selling cost for holding short positions. My result suggests reducing short selling costs to market makers, e.g. reducing borrowing cost or allowing naked short selling, decreases the excess volatility they induce as a result of their inventory control strategy.

My main focus in the second essay is on the characteristics of limit-order markets and the impact of market design changes on the incentives, actions, and payoffs of different kinds of liquidity providers. I show that agents with the lowest latency in the market have an incentive to front-run upcoming market orders. This behavior causes mismatch between expected execution prices and realized execution prices, excess volatility in spreads, and excess volatility in execution prices. In this model, the incentive depends on the latency of the fastest traders (not frequency) relative to the rest of the market and the tick size. Then, I examine several possible solutions; from changes to market latency or tick size to "taxing" front-running behavior through pricing of co-location based on order cancellations.

In the third essay, I switch my attention to the buy-side, and present a continuous-time model of trading on private information with uncertainty about the timing of information events. This uncertainty prevents partially-informed traders from knowing the "newness" of their private information. Their trades can cause the price to systematically diverge from fundamentals even when market participants are rational, there are no persistent exogenous demand (or supply) shocks, and there are no restrictions on trade. My result links the behavior of informed rational traders, i.e. the "smart money", to the seemingly manic episodes of price behavior, and suggests policy advice on the importance of transparency in maintaining informational efficiency and stability of financial markets.

Finally, I present recommendations aimed at standardizing a set of measures capable of capturing the multidimensional nature of the quality of financial markets. This is a first attempt to address this enormous subject, and it has not been my purpose to provide a "sufficient statistic" for market quality. I have aimed at providing a first draft to encourage further conversation and investigation. I have included only so much theory as I thought necessary, and have omitted altogether topics, although important, that did not seem to me to help with the comprehension of the problem at hand. Also, I have recorded seemingly unimportant details when I considered them illustrative of the nature of the problem.