top of page
Publications
Ambiguous price formation [Paper, Internet Appendix] (with X. He)
Journal of Mathematical Economics, 2023
Abstract: We study the price and liquidity of an asset in a model where market makers face ambiguity about the asset payoff. This ambiguity explains liquidity deteriorations and improvements in financial markets. We show that the ambiguity influences how market makers perceive adverse selection risk, and therefore, distorts market liquidity. The perceived adverse selection risk increases (resp. decreases) with the ambiguity when the market maker is sufficiently (resp. insufficiently) ambiguity averse. Our model also helps to understand how ambiguity and ambiguity aversion of market makers impact price and liquidity dynamics under various trading histories.
Toward a general model of financial markets [Paper] (with X. He)
Artificial Intelligence, Learning and Computation in Economics and Finance, 2023, 71-100, Springer
Abstract: This paper discusses the idea of reconciling efficient market hypothesis and behavioral finance using the literature of decision theories and information sciences. The focus is centered on the precision and reliability of information and the broad definition of rationality. The main thesis advanced is that the roots of behavioral anomalies are the imprecision and reliability of information. We propose a general framework that subsumes efficient and inefficient markets as special cases. We also show that the proposed framework helps to understand behavioral anomalies.
Working papers
Quants and market efficiency [Paper] (with X. Deng)
Abstract: We study the impact of quantitative trading strategies (quants) on price efficiency by developing a model that incorporates two defining features of quants: superior information processing capacity and limited strategic awareness. These forces have opposing effects on trading aggressiveness-better information processing increases it, while limited strategic awareness decreases it. As a result, an increase in quant participation deteriorates price efficiency when their information quality is low, improves it when information quality is high, and produces a non-monotonic effect in between.
Through stormy seas: how fragile is liquidity across asset classes? [Paper] (with M. Aquilina, K. Rzayev and S. Zhu)
Abstract: Market liquidity across asset classes has considerably increased in recent decades. Our study of stocks, foreign exchange (FX), and government bonds in the US, Europe, and Japan—using 25 years of high-frequency data—reveals a significant decline in both the average and standard deviation of bid-ask spreads across all asset classes. However, we also observe an increase in its skewness and kurtosis in equity and bond markets, indicating more frequent episodes of illiquidity. In contrast, FX markets do not show a significant increase in the higher moments of the distribution of bid-ask spreads. We identify structural breaks in the time series of spread distributions across regions and asset classes, associate these breaks with macroeconomic shocks and changing market conditions, and quantify the cost of this fragility to investors.
Have financial markets become more liquid? [Paper] (with J. Brogaard, K. Rzayev and T. Putnins)
Abstract: As investors shift their focus from average liquidity to liquidity risk, relying solely on average measures to claim markets are more liquid can be misleading. We document that as the mean of bid-ask spreads decreased over recent decades, their skewness increased. To understand why this has occurred, we examine how two key features of modern markets-high-frequency trading (HFT) and dark trading (DT)-affect bid-ask spread skewness. We find HFT increases skewness, while DT decreases it. The mean bid-ask spread is priced in NYSE stocks only, and only before 2008. In contrast, bid-ask spread skewness is priced in Nasdaq stocks throughout the 1996-2022 period, with an average monthly premium of 22 bps. We further analyse the drivers of illiquidity premiums associated with the mean and skewness of the bid-ask spread.
Can blockchain-based atomic settlements improve traditional financial markets? [Paper] (with R. Gaudiosi, and T. Putnins)
Abstract: Atomic settlement involves the conditional exchange of two assets. We model atomic settlement in permissionless and permissioned blockchains and apply it to traditional financial markets. We find that a permissionless blockchain is optimal for US equities and foreign exchange, and a permissioned blockchain is optimal for US corporate bonds and treasury bills, and the gold market. We estimate that transitioning to optimal blockchain settlement could potentially improve gains from trade in foreign exchange by $17 billion, Nasdaq by $12 billion, US corporate bonds by $6 billion, gold market by $4 billion, and US treasury bills by $420 million annually.
Equilibrium staking rewards: Implications for PoS blockchain security [Paper] (with K. Nguyen, T. Nguyen and T. Putnins)
Abstract: The security of PoS blockchains is at the mercy of external return opportunities. We model how capital moves among staking opportunities. Consistent with the model, we find that higher opportunity costs of capital decrease the staking ratio in PoS blockchains, thereby reducing their security. On average, a 1% increase in Fed fund rate equates to a 0.089% drop in staking ratio. Moreover, a 1% increase in DeFi yields in other blockchains may lead to a 0.096% drop in a chain’s future staking ratio. We derive economically grounded security measures for PoS blockchains, estimate them for major PoS blockchains, and use them to assess the overall security of the PoS ecosystem.
A taxonomy of tokenisation methods for real-world assets [Paper]
Abstract: This note outlines four distinct methods for tokenising real-world assets—1-for-1 ownership, 1-for-1 custodial, collateralised pool, and algorithmic issuance—and discusses the key conceptual, economic, and legal implications of each. We show how these models differ in terms of infrastructure design, collateral backing, and price stability mechanisms, and highlight parallels with traditional financial instruments such as ADRs, ETFs, and banking operations.
Once bitten twice shy: Learning about scams [Paper] (with I. Allahverdiyeva and T. Putnins)
Abstract: Do investors learn from their mistakes after falling victim to a scam? Experiencing a scam in the recent past (resp. a 1% increase in the scam investment ratio) decreases the probability of investing in a scam again by approximately 4% (resp. 0.30%). However, experiencing a scam also decreases subsequent non-scam investment returns by around 14%. Victims increase their portfolio values, but the cash allocation and the standard deviation of daily portfolio returns suggest that victims become more (resp. less) risk-tolerant in the short (resp. long) term after being scammed. Furthermore, less experienced investors tend to fall victim to multiple scams.
Short selling constraints and their impact on corporate investment decisions [Paper] (with X. Deng)
Abstract: How do short selling constraints impact corporate investment decisions? We analyze two forms of short-sale constraint (short-sale ban and cost) in a model where firm value is endogenous to trading, due to feedback from the financial market to corporate investment decisions. We show that, compared to an economy with no constraint on short selling, introducing a cost on short sellers can increase firm value, but a large cost or a short-sale ban always harms it. Our model suggests that the impact of short-sale friction extends beyond trading in secondary markets.
Scam alert: Can cryptocurrency scams be detected early? [Paper] (with I. Allahverdiyeva and T. Putnins)
Abstract: Decentralized exchanges (DEXs) have given rise to a new type of scam known as a "rug pull." We find these scams are widespread: 44% to 50% of tokens on major DEXs have “scam-like” characteristics, resulting in approximately $1.7 billion in investor losses. Using these characteristics, we develop a scam index that can detect scams early, providing a practical tool for surveillance and scam prevention. Scammers are increasingly using more subtle methods to exploit investors, while victims tend to learn from their past mistakes, suggesting a dynamic “cat-and-mouse” game between them. Furthermore, we find that scams become more prevalent during periods of increased market activity and that they discourage market participation.
The good and evil of algos: Investment-to-price sensitivity and the learning hypothesis [Paper] (with Khaladdin Rzayev and Fariz Huseynov)
Abstract: We investigate how firm managers’ learning from share prices is influenced by two different types of algorithmic trading (AT) activities in their shares. We find that liquidity supplying AT enhances managers’ ability to learn from share prices by encouraging information acquisition in markets, leading to increased investment sensitivity to share prices. However, liquidity-demanding AT impairs this learning process by discouraging information acquisition. Firm operating performance correspondingly improves with liquidity-supplying AT and deteriorates with liquidity-demanding AT in firm’s shares. Using NYSE’s staggered Autoquote implementation as an exogenous variation in AT, we establish causality. Our findings demonstrate AT’s significant impact on real economic outcomes.
The real effects of market manipulation [Paper] (with Inji Allahverdiyeva and Talis Putnins)
Abstract: Market manipulation distorts financial market prices, but does it have real economic effects on listed companies? We show that it does. Increased manipulation makes stock price signals less useful for firm managers seeking to learn about potential investment opportunities, thereby decreasing the sensitivity of firms' investments to stock prices. This leads to a decline in the quality of firms' investment decisions, and consequently, firm operating performance also decreases. Our findings suggest that the real economic consequences of market manipulation extend beyond the direct effects on secondary markets.
Is public distrust of the finance sector warranted? Evidence from financial adviser misconduct [Paper] (with I. Allahverdiyeva and T. Putnins)
Abstract: Using data from over one million financial advisers in the U.S., we estimate that up to 30% of advisers are involved in misconduct, but only about one-third of these cases are reported by regulators. Advisers with a high propensity for misconduct oversee approximately $6.9 trillion assets under management (AUM). The rates of adviser misconduct and unreported cases increase during the GFC, paralleling the decline of trust in financial institutions. One misconduct costs a firm nearly 9 clients and $14 million AUM annually. We provide a list of characteristics for consumers, advisory firms, and regulators to help identify adviser misconduct.
Learning about adverse selection in markets [Paper] (with Xue-Zhong He and Talis Putnins)
Abstract: How does a market learn about the number of informed traders and thus adverse selection risk? We show that trade sequences convey information about adverse selection risk. Consequently, buy/sell order imbalances can destabilize markets, triggering extreme price movements, flash crashes, and liquidity evaporation. The increasing prevalence of these effects in markets can be explained by more active learning about adverse selection by competitive, high-frequency market makers. We use our model to estimate the uncertainty in adverse selection risk for US stocks and show that it decreases market liquidity and increases extreme price movements.
Ambiguity and information tradeoffs [Paper]
Abstract: We model investors facing ambiguity about the number of informed traders and characterize equilibrium in both financial and information markets. In the financial market, this ambiguity leads to a premium that can be positive or negative, depending on traders’ ambiguity attitude. The premium always increases with ambiguity aversion but only increases with ambiguity level when traders are sufficiently ambiguity averse. We show that traders’ effective ambiguity aversion increases with the number of informed traders, resulting in a non-monotonic relation between the equity premium and informed traders. In the information market, we determine a unique proportion of informed traders given the cost of information.
A visual guide to market microstructure models [Paper]
Abstract: In this note, I provide geometric interpretations of the classic asymmetric information models of Kyle (1985) and Glosten and Milgrom (1985). This visual approach helps to better understand how market makers infer private information from order flow and how prices adjust in response.
bottom of page