Three essays on short selling bans, asset pricing and market quality
- Autores: Oscar Hernández
- Directores de la Tesis: José Penalva Zuasti (dir. tes.), Miguel Ángel Tapia Torres (codir. tes.)
- Lectura: En la Universidad Carlos III de Madrid ( España ) en 2021
- Idioma: español
- Tribunal Calificador de la Tesis: Belén Nieto Domenech (presid.), Pedro Jose Serrano Jimenez (secret.), Isabel Figuerola Ferreti Garrigues (voc.)
- Programa de doctorado: Programa de Doctorado en Empresa y Finanzas / Business and Finance por la Universidad Carlos III de Madrid
- Materias:
- Enlaces
- Tesis en acceso abierto en: e-Archivo
- Resumen
In this Doctoral Thesis we perform a detailed study of SEC Rule 201 short selling ban and its effects on asset pricing and market quality. In the first chapter, we analyze thoroughly the econometric opportunities of the regulation and how it can be exploited as an quasi-exogenous shock. In the second chapter, we build on the econometric knowledge of the first chapter and conduct a daily event study focusing on price efficiency. The last chapter covers an intradaily study that focus on the immediate consequences of the rule activation in the main intraday market quality metrics. Next, we expose more in detail the contribution of each chapter, its methodology and its main results.
CHAPTER 1 The question of the true effects and consequences of short selling restrictions are still debated today among researchers, regulators and practitioners. Under times of extraordinary market volatility or sustained price declines, most regulatory authorities across the globe have historically reacted by limiting short-selling, in the expectation that these measures helped markets to return to a status of operating normality and reduced investors' distrust. The regulatory authorities justify their position as an answer to a growing concern among investors.
Despite short selling bans are typically imposed as a mechanism to foster price stabilization and help financial markets recover under times of distress, there is no clear consensus among researchers about the true effects of such regulation. The classic models regard short selling limitations as market frictions that harm market efficiency and reduce price informativeness. Recent policy changes have allowed for an extensive literature on empirical studies of the classic hypotheses, but there exist some methodological concerns about the research frameworks employed that could question the representativeness of the samples used and/or the extrapolability of the results.
As a market-wide short selling restriction in the American markets was active for more than seven decades, there was a major difficulty in conducting empirical tests in this framework due to the lack of comparable samples. The early 2000s opened the field for empirical research with the introduction of a pilot program in 2005, a repeal of the previous regulation in 2007 and the temporary reinstatement of short selling ban via an emergency order by 2008. In less than five years, there were more policy changes than in the preceding fifty. Most of the main empirical papers in the literature build on this specific framework of the early 2000s with unclear results about the role of short selling bans on asset pricing, liquidity or price informativeness (among others). However, the particular characteristics under which the majority of the studies are conducted could be a source of concern with respect to the relevance and extrapolation of the conclusions.
In this chapter, we document a set of historical shortcomings of prior research on the topic of short selling bans and analyze the usability of the new regulation in the US since 2011 (the Rule 201) as a solution for these caveats. Our results indicate a substantial potential of this new rule to offer heterogeneous long-term event studies that could isolate the ban effect accurately. Moreover, we document that Rule 201 short selling bans are not fully random, listing the set of factors significantly correlated with the propensity of receiving the treatment that would be useful for future research in the topic to guide the correct empirical design of the studies on the matter. Overall, Rule 201 poses a significant opportunity to solve the documented flaws, but it is also limited to a demanding empirical design to establish clean causality. We believe the discussion, results and analyses gathered in the paper will represent a substantial contribution and be helpful for future research building on this particular framework.
Through an econometric development and a thorough analysis of the Rule 201 circuit breakers for more than six years, we show that Rule 201 events exhibit potential for studying many of the still unresolved questions. Its main advantages arise from its unique design plus the fact that it has been active for a long period of time, offering a large, rich and heterogeneous sample that covers many different market states, rather than very specific frameworks as was the case with the pilot program or the 2008 emergency ban.
Throughout the chapter, we offer advice on the design of an empirical setup that combining features of regression discontinuity, experimental studies and differences-in-differences, can effectively identify causality in this framework. We believe this should be of special relevance for those researchers interested in employing Rule 201 as their research framework, guiding them in the process of designing the correct strategy to obtain insightful analyses that help academics, practitioners and regulators to understand the implications of short selling bans more accurately, hence contributing to a better understanding of the topic and the pursue of the optimal regulation level.
CHAPTER 2 There still exists an unresolved debate about the true consequences of short selling constraints, their effectiveness and the consequences on market microstructure. The main controversy lies around whether short sellers are active contributors to market efficiency or rather act as predatory traders which induce unjustified bearish pressure. Often, regulators have opted for short selling constraints as a measure to stabilize markets, especially, as a reaction in times of extraordinary volatility and uncertainty. However, this restrictive approach is not fully supported in the literature. Only in some contexts, such as financial stocks, a restriction on short selling can help prevent predatory behavior. On the other hand, a wide extent of the literature highlights the valuable role of short sellers as effective channelers of information about future underperformance, fostering price efficiency and reducing overreaction to news.
The evidence is combined with the main theories building on the efficient market hypothesis, which indicate that the removal of short sellers should be associated with worse market quality in terms of illiquidity and inefficient overpricing due to the over-weighting of optimistic trading. Recent changes in the regulations across the world have opened the field for testing these classical results under frameworks of event studies, showing evidence in favor of constraints being related with unjustified stock returns and less liquidity, especially for more vulnerable and smaller stocks.
In this chapter we examine how the Rule 201 short lived restriction on short selling affects stock pricing in two principal dimensions: price efficiency and overpricing. We conduct a daily event study of more than six years using a differences-in-differences methodology combined with a matched pairs sample that ensures causal identification of the ban effect. We find evidence of the Rule 201 short selling ban supporting prices above their economic value, which is associated with Miller's (1977) overpricing theory. As the short selling restriction is repealed, short sellers drive the price down towards the true asset value. We argue that the removal of short sellers could defer the convergence of the stock's price and economic value, generating a sudden adaptation of prices when short sellers return to the market. This sudden adjustment would be associated with higher short-term volatility and distrust in the financial markets, which are two of the issues that the Rule 201 was designed specifically to target. Our results support this argumentation and highlight the importance of correctly assessing the pricing dynamics that surround the Rule 201 triggers and bans.
Our concerns are supported by our analysis of price efficiency. Using the Boehmer and Wu (2012) specification we assess the speed with which assets incorporate publicly known information. Our tests of price delay show a statistically significant effect of the ban related to a delayed speed of price discovery for stocks subject to the ban. The effect is not limited only to the day during which short sellers are removed, revealing a sticky effect that lasts for up to three days after the actual trigger. This result raises questions about the true effectiveness of Rule 201 in improving the situation of stocks under distress.
Our event-study design of the Rule 201 ban, with a matched pairs analysis that allows for contemporaneous comparison is currently one of the largest empirical studies analyzing the issue of short selling bans, with a wider time window that prior studies using short lived one time bans such as the emergency ban of 2008 or the Russell 3000 pilot program lacked. We also show empirically the validity of the Rule 201 events as candidates for a quasi experimental study pointed out in the first chapter. Our research design and robustness should be helpful for future research about the Rule 201 framework that contributes to a better understanding of the implications of the new regulation. Given the importance of the topic under question and the substantial degree of dimensions to analyze, we consider there is still a major field to be explored in this setting and we expect that our methodology offers an insightful guidance for future researchers in this framework.
CHAPTER 3 After establishing the main methodological foundations (in Chapter 1) and exploring the classical hypotheses of asset pricing using a daily event study (in Chapter 2) we are now interested in exploring the dynamic effects of the current regulation at a granular intraday level.
The U.S. has experienced several rounds of regulation of short sales, and we are The current regulation is significantly different from the type of short sale restrictions observed in the past. In particular, the current regulation imposes a short sale ban only on aggressive short sales and comes into effect only after a stock suffers a significant price drop (10\% from the previous day's closing price). The short selling restrictions affect only that stock, and for the rest of the day and the subsequent 24 hours. This is a temporary, endogenously triggered regulation that does not ban all short-sales, but only ``aggressive'' short sales, defined as those at (or below) the current bid price, and applies to all but a handful of exempt types of short sales. In contrast, the most commonly observed bans in the past have been bans on all short sales, imposed for a prolonged period of time, and on a significant fraction, if not all, traded stocks.
Our analysis splits the effect into an overall and a local effect. With the local effect we look at the effects within a window of five minutes before and after the minute in which the restrictions are activated, while the overall effect captures what happens during the remaining time of day in which the restrictions are in place. We find that the observed market conditions are consistent with a situation in which the stock is going through a strong downward price pressure, with strong order imbalance driven by sales of the asset, which results in a price drop that is partly reversed by trading over the remainder of the day after the restrictions are introduced. The restrictions appear to be accompanied by a partial shift to competition in liquidity provision using hidden orders, an increase in off-exchange trading, and off-exchange exempt short sales.
In particular, following the introduction of the restrictions we find that volume continues to be greater than before, and it continues to reflect an order imbalance driven by sales. This is also accompanied by increased off-exchange short sales, where non-exempt short sales continue at pre-restrictions level, and the volume of exempt short sales increases. We observe a displacement towards off-exchange trading, but not from the NASDAQ exchange, which retains its pre-restrictions market share. On NASDAQ, liquidity, as measured in terms of the quoted spread and depth (particularly on the bid side), is lower than before, but the effective spread improves. We explain this from an observed increase in hidden volume that suggests that after the restrictions are introduced, aggressive short-sellers convert their executable sales into aggressively priced hidden orders inside the spread. Liquidity demanders facing higher quoted spreads can thus obtain lower effective spreads by executing against these hidden orders inside the spread. This increased competition via hidden liquidity is paired with a reduction in algorithmic activity (measured in terms of trade-to-order ratio) on the ask side of the order book, and an increase on the bid side. For market makers we observe lower realized spreads, consistent with lower visible liquidity, and with a slowly rebounding of the price, as well as potentially substantial profits for long-term liquidity providers willing to hold on to the asset over longer time horizons.
We also observe that the introduction of the restrictions leads to unusual activity in the immediate vicinity of the transition to active short sale restrictions. We find that the introduction of the short sale restrictions appears to act as a catalyst for price pressure to turn into a price drop (albeit only in the short-term) in a way that reminds us of the bubbles bursting in the coordination model of Brunnermeier and Oehmke (2013). This is accompanied by unusually high volume, larger sell order imbalance, larger volatility (wider ranges of price movement), and wider effective and quoted spreads. We also observe a temporary increase in buy volume on NASDAQ, and away from other (off-exchange) venues but only in the minute of the event. This coincides with a shift in algorithmic trading on the bid side of the order book, and higher realized spreads.
This third chapter provides a detailed analysis of the stock-days in which the short sale restrictions of Rule 201 are put into effect. With this, we aim to provide additional evidence on the ongoing discussion over the positive or negative role played by the short sales restrictions in general, and the current US Rule 201 in particular. We focus on the effects of the Rule 201's ban of aggressive short sellers on market quality, the level and distribution of traded volume (visible, hidden, and short-sales), returns and volatilities, algorithmic activity, and other dimensions of market microstructure during the day the short-selling restrictions are activated.
The first key take-away from our analysis is that off-exchange short selling activity increases, primarily due to increases in exempt short-sales, while at the same time hidden volume in NASDAQ also increases. The combination of these suggests that short-selling continues to take place in less transparent venues through the channels permitted via Rule 201 exemptions, or is modified to continue to be aggressive but not immediately executable in the form of aggressively posted hidden limit orders inside the spread.
Second, in terms of market quality, the effect of the ban has a negative impact when we look at depth or quoted spreads. However, the presence of aggressively posted hidden limit orders leads to an improvement in the effective spread, so that the effective cost of transactions decreases (at least on NASDAQ). The transformation of aggressive short-sales into aggressive limit orders is consistent with the general evidence in Comerton-Forde et al. (2016) that short sellers become liquidity providers when the quoted spread is wider than usual.
The possibility of increased competition for liquidity from hidden orders is also consistent with the observed decrease in algorithmic activity on the ask, both in terms of the trade-to-order ratio and ultra-fast activity, despite an observed increase in the arrival of visible aggressive buy orders. Furthermore, realized spreads (and associated expected profits) across our three horizons of analysis (100 miliseconds, one, and five minutes) are lower which further justifies the observed wider quoted spreads and lower depth.
We also find evidence consistent with previous studies on the dynamics of the price. As expected, returns are negative in the minutes after the ban and we observe a `rebound' during the rest of the day. Accompanying the price movements, we find that general market conditions around the time of the activation of the restrictions are similar to those in early trading, where high volume is accompanied by wider quoted spreads and higher volatility. However, we find that overall trading volume, although constant in NASDAQ, shifts from other exchange to off-exchange venues.
Our results are obtained using standardized variables and trade direction in TAQ data and hidden volume is obtained signed using the Lee-Ready algorithm. In short, this chapter sheds some light on the immediate consequences of the short selling restrictions under Rule 201, and provides direct evidence of the impact and possible costs and benefits of this particular form of regulating short sales.
