Recent booms and busts in commodity prices have placed renewed scrutiny on commodity futures markets as a mechanism for price discovery, the process of incorporating new information about the relative scarcity of the commodity into prices. Such concerns are not new; there has been some distrust of futures market price discovery since the inception of these markets. As these markets evolve, new market participants and institutions may influence price discovery. Using the Intercontinental Exchange (ICE) cotton futures market as a laboratory, I consider three such forces potentially responsible for poor price discovery during the 2007-2011 period of volatile cotton prices. These are financial speculation, electronic trading, and funding constraints on commercial hedgers. In Chapter 1, I study whether the increased presence of financial firms, particularly commodity index traders, drives cotton futures prices away from the levels implied by supply and demand under rational expectations. I estimate a structural vector autoregression model of the cotton futures market. My model develops a new method to point identify shocks to precautionary demand for cotton separately from shocks to current supply and demand and separately identifies the effects of two types of speculation: precautionary demand for the commodity and financial speculation. I show empirically that most cotton price variation stems from contemporaneous unanticipated shocks to current cotton supply and demand. However, the 2008 price spike came from an increase in precautionary demand due to projections of lower future production. I find no evidence in support of claims that financial speculation causes commodity booms and busts.Chapter 2 considers the introduction of electronic trading to the cotton futures market across three periods of floor trade, parallel floor and electronic trade, and electronic-only trade. I statistically decompose intraday variation in cotton prices into a component related to information about market fundamentals and a ''pricing error'' caused by frictions in the trading mechanism. Better market quality or price discovery is characterized by lower variance of the pricing error. Unlike previous studies of floor and electronic trading, I consider more than average measures of market quality. I calculate statistics for market quality for each trading day, and study their trend, variance, persistence, and relationship to other variables related to price discovery. I find that market quality improved, but became more variable under electronic trading. This relationship between electronic trading and market quality is robust to controls for changes over time in the number of trades, trading volume, and price volatility.My final chapter considers the role of funding constraints in exacerbating futures price spikes. I review the experience of commercial hedgers during the 2008 cotton futures price spike. In this period, commercial hedgers without access to credit were forced to close futures positions in an illiquid market. Losses incurred on these trades led some firms to exit the cotton merchandising business. I use facts from the cotton case to develop a dynamic model of futures market equilibrium in the short-run for cases where funding constraints for some hedging firms bind and do not bind. Analytical results show that observed futures price volatility can be explained by the relation between funding liquidity of trading firms and market liquidity. This relationship alters the trading behavior of hedgers and results in diminished price discovery.

My dissertation is a compilation of three separate research studies that explore how information diffuses in financial markets. The first chapter examines how non-uniform information diffusion through distinct networks segments U.S. financial markets. Using changes in newspaper ownership networks, I document that a network link between different geographic areas leads to increased comovement of turnover and returns between stocks headquartered in those areas. Consistent with delayed content sharing within a network, the largest increase in comovement is observed using weekly data. I show that the network-driven comovement is not driven by fundamentals and is weaker for large firms with high institutional ownership and decreases over time. I also document that a network link causes price levels of linked stocks to become more similar. My findings show that segmented information networks lead to segmented financial markets with implications for market efficiency, home bias, and the effects of changes in the U.S. media landscape on financial markets. The second chapter shows that investors do not fully monitor the information about directors available in the past prices of firms within the network the directors oversee. A long-short portfolio using this information yields an annual alpha of 6.6%. This predictability is limited to when firms share a director and is not driven by industry or previously identified economic links between firms. The predictability is largest in the long end, when small firms predict big firms, and when information on shared directors is costlier to obtain. Trading by the shared directors is a key mechanism: filtering on their trades increases the annual alpha to 15%. The third chapter studies the econometric properties of a commonly used network-based measure of information diffusion between economically linked firms. Previous studies use this measure to document failures of market efficiency with price discovery requiring up to a year. The measure is constructed as the long-short alpha of portfolios formed sorting on the preceding returns of firms economically linked to portfolio firms. We show that correlated alphas between linked firms bias these measures. Existing studies have monthly biases as large as a factor of two. This bias creates predictability even after price discovery completes. Subtracting the predicted return from the sorting firms' returns removes this bias. Eliminating this bias reveals a more efficient market than previously documented: price discovery takes one month.

This dissertation consists of four chapters. Chapter 1 studies the effect of online review manipulations on review systems. Chapter 2 and Chapter 3 are co-authored with Ali Kakhbod and focus on post-trade transparency in dynamic over-the-counter markets. Chapter 4 is co-authored with Umut Dur, Parag A. Pathak and Tayfun Sönmez and studies the effect of the Taiwan mechanism, a mechanism that allocates high school seats to applicants. Chapter 1 shows that the conventional impression holds in the short-run that review manipulation makes review systems less informative. In the long-run, however., a manipulated review system can contain the same level of information as an un-manipulated counterpart. I develop a dynamic programming model with fixed product quality and naive buyers who are unaware of manipulation. I then extend it to consider endogenous product quality and sophisticated buyers. I also identify an unexpected effect of a policy to target sellers and check for manipulation. Chapter 2 studies how mandatory transparency (through TRACE), along with the long-term incentive of informed dealers, affects market price informativeness, liquidity and welfare in dynamic over-the-counter (OTC) markets. We show that the public disclosure of additional information about past trades, paradoxically, makes the markets more opaque, by reducing the market price informativeness. Thus, surprisingly, transparency requirements such as U.S. Dodd-Frank Act may make markets more opaque. However, this market opacity creates liquidity and increases welfare. To enhance financial transparency and improve the price informativeness as well as the market liquidity and welfare, an effective approach is to randomly audit dealers. Chapter 3 then studies how public disclosure of past trade details affects price discovery dynamics under asymmetric information with heterogenous hedging motives. We model that an informed buyer (informed trader) sequentially trades with a series of uninformed sellers (hedgers). The informed buyer is forward-looking and risk-neutral, and uninformed sellers are myopic and heterogeneously risk-averse. We discover that sellers' price discovery over the underlying fundamentals is crucially affected by what they can observe about past trade details. Specifically, (i) post-trade price transparency delays price discovery, but once it happens, it is always perfect. (ii) In contrast, when only past order information is available, price discovery can never be perfect, and can even be in the wrong direction. (iii) The availability of past trade details, paradoxically, makes it easier for the informed buyer to hide her private information and offer opaque prices. We establish that, under some minor regularity conditions, our equilibrium characterization achieves the maximal degree of ignorance among all pure-strategy PBE. Hence, this chapter can be viewed as a worst case analysis for regulators who care about market transparency. Moreover, we show that our findings are robust when the informed party's bargaining power decreases along the length of past trade history. Finally, we extend our results to the case where the informed buyer has a non-zero outside option, and the case where both parties switch their trading positions. Chapter 4 analyzes the properties of the Taiwan mechanism, used for high school placement nationwide starting in 2014. In the Taiwan mechanism, points are deducted from an applicant's score with larger penalties for lower ranked choices. Deduction makes the mechanism a new hybrid between the well-known Boston and deferred acceptance mechanisms. Our analysis sheds light on why Taiwan's new mechanism has led to massive nationwide demonstrations and why it nonetheless still remains in use.

Automated trade execution systems are examined with respect to the degree to which they automate the price discovery process. Seven levels of automation of price discovery are identified, and 47 systems are classified according to these criteria. Systems operating at various levels of automation are compared with respect to age, geographical location, and type of securities traded. Information provided to market participants, and asymmetries of information between traders with direct access to the automated market and outside investors also are examined. It is found, for example, that the degree of asymmetric information increases with the level of automation of price discovery. The potential for trading abuses related to prearranged trading, noncompetitive execution, and trading ahead of customers is analyzed for each level of automation. Certain levels of automation widen the opportunities for trading abuses in some respects, but may narrow them in others.