A Stock Returns Liquidity Volatility

Published Date: 02 Nov 2017

Martin Tamborsky

Abstract

In August 2011 the financial authorities in France, Spain, Italy and Belgium adopted a covered short sale ban on financial institutions. In this paper, I study the impact of this short-selling ban on stock returns, liquidity and volatility. I analyze various test periods before and after the introduction of the ban to quantify changes over time. The stocks subjected to the ban suffered degradation in market quality, measured by the relative bid-ask spread. The positive effect on stock return was only short-term. There was an especially strong short-term positive effect on stock prices in the Developing and PIGS country subgroup. However, there is no statistically significant positive effect on stock returns in the long-term horizon. The impact on intraday volatility is inconclusive. The short-term jump of excess volatility of the banned stocks is compensated in the long-term horizon.

Introduction

Short selling has become a much discussed topic in recent years. We have observed a turbulent stock market development from the year 2007 until now. Short selling influences a downturn in the stock market and expresses popular opinion. If politicians or the stock exchange authority want to stop a market crash, they ban short selling. A sort selling ban has been imposed a few times since 2007. I explore the effect of the short-selling ban during the debt crisis period of 2011. There was a short-selling ban only on financial institutions in my research sample. Which strategy was better? Was the short-selling ban helpful? The examination of this question is based on stock prices, bid-ask spreads, and a volatility analysis. I made several thousands of various empirical observations to justify my conclusions. As we will see, the results change based on different time periods.

Short selling is the selling of shares that the seller does not own. The seller usually borrows the stock from a broker’s account and sells it with the anticipation of a downturn. In the future, this trader buys the stock at a lower price on the market. Short selling has not been common only in recent years; short selling has a long history. The first famous short sale was made by the Dutch trader Isaac Le Maire in 1609. He sold more shares of the Dutch East-India company than he owned. He was blamed for causing a drop in the share price, and short selling was forbidden by the Dutch government. Short selling was perceived as a magnifying effect in the violent downturn in the Dutch tulip market in the 17th century. In another well-referenced example, George Soros became notorious for "breaking the Bank of England" on Black Wednesday of 1992 when he went short more than $10 billion worth of pounds sterling. Short-sellers were blamed for the Wall Street Crash of 1929. Regulations governing short selling were implemented in the US in 1929 and in 1940. The political fallout from the 1929 crash led Congress to enact a law banning short-sellers from selling shares during a downtick; this was known as the uptick rule [1] and was in effect until 2007. Today short-selling is very common. Diether, Lee and Werner (2009) show that short selling accounted for 24% of the New York Stock Exchange trading and 31% of the NASDAQ volume in 2005. In other words, approximately half of all seller-initiated trades are short sales. The popularity of short selling has increased dramatically in recent decades. 95% of developed countries allowed short sales in 2002. This ratio was 31 percent in emerging countries. Before 1990, the respective figures were 64% in developed countries and 10 percent in emerging countries.

On August 11, 2011, European regulators, in response to the continent’s debt problems, decided to limit short selling. The existing bans on short selling in Greece and Turkey were supplemented by bans on the short selling of financial stocks in France, Belgium, Italy and Spain. The local financial authorities of France, Belgium, Italy and Spain placed a ban on creating a net short position [2] or increasing the previous one, including intraday trades. The short-selling bans were lifted in the beginning of 2012.

This master’s thesis is structured as follows: Chapter 2 consists of a research overview. Chapter 3 describes the sample of all the stocks and data I worked with. Chapter 4 shows the reasons why financial authorities prefer to impose bans on financial institutions. There is also an empirical analysis on US stocks in Chapter 4. All other sections are focused solely on the European stock market. Chapters 5 and 6 present the results related to stock returns and bid-ask spread changes respectively. Chapter 7 provides a volatility analysis. Finally, some concluding remarks and policy recommendations are made in the final chapter of the paper.

Evolution of short-selling research

Research papers related to short selling often present opposing results. Opponents of short selling argue that short selling disrupts orderly markets by causing panic selling, high volatility, and market crashes. However, supporters of short-selling argue that it increases information efficiency and the liquidity of stock markets, and also improves the risk sharing mechanism in economic systems.

The research related to short selling has evolved over time. Seneca (1967) published a research paper with an unsurprising conclusion: an increase in short interest is a bearish indicator. Academic literature on short selling has had a renaissance since the 1970s.

According to Miller (1977), when short-selling is restricted and investors have heterogeneous beliefs, the private information of optimistic investors is slowly fed into prices through trading, but the private information of informed investors, who are bearish and do not own stocks, is not revealed in prices. Later, when market information is published through an announcement, because bad news has not been disseminated, there are stronger stock market adjustments for bad news than good news.

Diamond and Verrechia (1987) developed a theoretical model with rational traders. According to their conclusion, a short-sale ban does not lead to biased prices. Short-sale constraints influence the degree to which private information is revealed in the prices. In other words, these constraints reduce informational efficiency especially with respect to negative news. Option trading serves as an alternative to short selling, and it increases the informational efficiency.

Ofek and Richardson (2003) showed that short selling played an important role during dot.com crisis. In the February 2000 index, NASDAQ reached maximal values. Short interest was considerably higher for Internet stocks than for their corresponding old economy counterparts at this time. Short interest here is defined as the total amount of shares of stock that have been sold short relative to the total amount of shares outstanding. For example, the short interest measures were 2.8 percent (for Internet stocks) versus 1.8 percent for the mean and 1.6 percent (for Internet stocks) versus 0.7 percent for the median. To gather additional evidence, the authors collected proprietary rebate rates for the universe of stocks on a selected number of dates from a financial institution. They assumed that a low rebate rate represents a weak supply of stocks for lending. The mean and median rebate rate was, respectively, 1.08 percent and 1.45 percent less for Internet stocks than other stocks, and these differences are statistically significant. Note that 46 percent of the Internet stocks lie in the 10 percent tail of all rebate rates. According to the authors of this research paper, a Lockup agreement is the modification of short sale constraints. Some investors are prohibited from selling and short selling in case of a Lockup agreement. This prohibition usually takes effect some period after an IPO. After the expiration of this period, stock performance of Internet stocks is different compared to other stocks. The drop of Internet stocks was approximately 1% stronger compared to the non-Internet firms during the days following the lockup expiration. The negative performance remained for the next 10 days (the whole research period). This fact illustrates the negative pricing effect of sale and short sale constraints. The contribution of this paper to the field of financing is the knowledge that short sale statistics can predict price movement or market crashes.

With the exception of a few papers, most available empirical research papers examine the impact of short-sale restrictions at the individual stock level. Charoenrook and Daouk (2005) examined short-selling issues from an aggregate market point of view. The impact of market-wide restrictions on market returns can differ significantly from the impact of short-sale restrictions on individual stocks due to diversification. They examined the effect of a market-wide short-sale restriction on volatility, probability of market crashes, liquidity, and expected market returns. Their data stem from 111 countries. In contrary to other papers which are focused on a critical point-in-time on the stock exchange, the paper by Charoenrook and Daouk (2005) is focused on the very long-term impact of short sales. The data range from 1969 through 2002. The authors constructed a binary variable that reflects the ability of investors to take short positions either through the existence of short-selling or put option trading. The coefficient and statistical significance of this variable is critical in the empirical analysis. Based on the variance of daily returns and monthly returns from the ARCH model, the aggregate market return is less volatile when short-selling is possible. Their next analysis shows that the feasibility of short-selling has no relation to the probability of a market crash. The authors used turnover as proxy for liquidity. When short-selling is possible, the turnover is 15 percentage points higher. As far as returns are concerned, the authors argue that investors should require a lower expected return when variance risk is lower and liquidity is greater, which is when short selling is possible. Additionally, when investors can diversify their risks in a more efficient manner, they require a lower rate of return for their investment. This fact is proven empirically in the paper. The theory, however, is not consistent with what happens with prices when a short-selling prohibition is lifted. On one hand, Miller´s overpricing effect during a short-selling ban should cause a decrease in market prices when short selling becomes possible. On the other hand, if expected returns are lower when short-selling is allowed, stock prices should increase when the prohibition of short-selling is lifted. The empirical evidence illustrates that market prices increase when short-selling restrictions are lifted. Paper of Charoenrook, Daouk (2005) shows the positive side of short selling in the long-term. Banning bets on decreasing markets may deteriorate volatility and liquidity and increase expected returns.

Another reputable paper focused on the financial crisis period (from August 1st until October 31st, 2008) was published by Boehmer, Jones, and Zhang (2009). Their research sample consisted of 465 short restricted stocks traded on the NYSE and NASDAQ. 404 stocks from this sample were short restricted on 18 September, 2008. An additional 61 stocks later became subject to the shorting ban. They also created a matched sample of 465 stocks where shorting was never banned. The authors found a significant abnormal return on the 404 stocks banned on 18 September, 2008. These results were deformed by the TARP announcement. To try to neglect the confounding news about TARP, the authors looked at the subset of firms that were added to the ban list at a later date. This second sample of stock performed differently compared to the first 404 stock sample. Except for the first day outperformance, the short restricted stock performed poorer compared to the freely traded sample. There are two possible arguments for this. One is an illiquidity discount related to deteriorating market quality. The second one is that the investor can understand the addition to the ban list as a negative signal about the company´s prospect. To measure the impact of short selling on market quality, the authors analyzed the effective and quoted spread. The results indicate a dramatic increase of spread on the stocks subject to the shorting ban. The volatility, calculated as the difference between the highest and lowest transaction price recorded for a given stock on a given trading day, divided by the stock´s volume-weighted average trade price for the day, increased in the short-term too. This research paper achieved a similar conclusion as Beber and Pagano (2009).

Appel and Fohlin (2010) used a non-traditional difference-in-difference estimator comparing non-US stocks with their American Depository Receipts (ADR). ADRs corresponding to the non-U.S. shares were not short sale restricted for most of the duration of the non-U.S. bans. Therefore, ADRs are the optimal control group for the short restricted non-U.S. shares. Their research period was from September 2007 until July 2010. The research sample, unfortunately, consisted of only 35 ADR stocks. Liquidity issues were measured by a relative effective spread and Amihud ratio. The short-selling ban improved market liquidity or at least had a neutral impact. The authors justify these empirical results through the modified theoretical model by Glosten and Milgrom (1985). They assumed that short sellers were more likely to hold private information relevant to valuation, and pure liquidity traders were less likely to sell short. Applying the Bayes´ rule, the difference between the non-ban and ban spread is always positive. In other words, the bid-ask spread during the non-ban period is larger than the spread during the ban period, according to the Glosten and Milgrom (1985) model adjusted by the assumption that short sellers hold private information. The authors focused on volatility changes as well. Volatility was empirically measured by the 20-day rolling standard deviation of returns. The results indicate that volatility decreased in equity markets by imposing a short-selling ban. All in all, Appel and Fohlin (2010) found that banning short selling might prove useful in times of crisis.

Beber and Pagano (2011) published a reputable and very extensive research paper. I will describe this paper in detail as it examines similar issues as my master’s thesis, just in a different period. It enables me to compare my empirical evidence to the empirical conclusion of Beber and Pagano (2011). Their sample consists of daily data for 16,491 stocks in 30 countries, from January 2008 to June 2009. This research paper focused on stock return, liquidity and price discovery. The bid-ask spread increased worldwide with the salient moments of the crisis (the collapse of Bear Stearns, collapse of Lehman Brothers, etc.). Short-selling bans were introduced in the wake of the bad news about the state of U.S. banks in September 2008. These short-selling bans contributed to the deterioration in liquidity. Stocks affected by a short-selling ban feature a significantly larger median bid-ask spread during the ban period. Based on the Wilcoxon test for the difference between the median spread, the difference is statistically different from zero at the 1 percent level for all the countries. For example, the median bid-ask spread for stocks affected by the ban increased by 127 percent. The bid-ask spread of the control group increased by only 49 percent. The analysis by Beber and Pagano (2011) shows that the ban on naked short sales is associated with an increase of 1.28 percentage points in the bid-ask spread, and covered short sales with an increase of 1.98 percentage points. The bid-ask spread is negatively correlated to the obligation to disclose short sales. This suggests that the disclosure of a short sale position may reduce adverse selection problems in the market. In the subset of financial stocks, short-selling bans are associated with a larger bid-ask spread as well. Very similar results are also obtained using the Amihud illiquidity measure. It is well known that, even in the absence of short-selling constraints, market makers provide less liquidity to small-cap and riskier stocks than for other stocks. The impact of a short-selling ban on small-cap stocks was stronger but not statistically significant. During short-selling bans, investors could still bet on falling stock prices by trading in the option markets because ban regulations did not impose any direct restrictions in derivative markets. As expected, the authors found a significantly stronger effect of short-selling bans on liquidity for stocks without listed options. The next part of their paper is related to the analysis of stock returns. Beber and Pagano’s (2011) methodology is focused on the countries where the ban did not apply universally, and compares the post-ban median cumulative excess returns for stocks subject to bans with those of exempt stocks, where excess returns are defined as the difference between individual stock returns and the respective country equally-weighted market indices. The median cumulative excess return of U.S. financial stocks which were subject to a ban exceeded free traded stocks after the ban inception. But there is no such outperformance in other countries: the median excess return on stocks subject to short-selling restrictions is very close to that for other stocks. We can observe only a short-term outperformance of banned stocks in countries outside the U.S. The U.S. stock market response to short-selling bans is positive and significant. The authors highlight that the positive effect for the U.S. may be driven by the TARP announcement rather than from the ban itself. Therefore, in countries other than the U.S., short-selling bans are associated either with no significant change or with a decline in stock returns.

All in all, their empirical analysis suggests that the short-selling ban was detrimental to market liquidity, especially for stocks with a small market capitalization, high volatility and no listed options. Moreover, it slowed down price discovery, and it was at best neutral in its effects on stock prices. They found excess return for the banned stocks in the U.S. Researchers argue this could have been caused by the TARP announcement to support financial institutions. The ban was announced on September 18, 2008, and on September 19, 2008 Treasury Secretary Henry M. Paulson proposed TARP. They found short-term excess return for European banned stocks.

One of a few research papers related to the 2011 debt crisis short-selling ban was published by Alves, Mendes and Silva (2012). They made an empirical analysis but they used a different sample and different methodology than I used. Their sample consisted of 170 financial stocks purely in Western Europe (basically all the listed financial institutions and insurance companies in Western Europe). Stock return and volatility were analyzed by the cumulative abnormal return over the event period. This statistical technique follows the MacKinlay (1997) and Brown and Warner (1985) methodology. The evidence presented suggests a short-term outperformance of banned stocks. Volatility was measured by an f-test, t-test and Beaver´s U. The short-selling ban did not contribute to reducing volatility according to their empirical analysis. Their analysis shows that the number of stocks that exhibit a decrease in volatility is higher for freely traded stocks after the introduction of a short-selling ban. In the next step, the permanent impact of a short-selling restriction was described. The data in this analysis was collected on a weekly basis. Liquidity, measured by bid-ask spread and the Amihud indicator, deteriorated after the initiation of the ban.

Nowadays, not only stock returns, liquidity, volatility or price discovery are connected with short-selling research. Other issues, like the impact of a short-selling ban on information production, come into discussion. The findings of other important papers are described in the following chapters.

Data

I made empirical observations to justify my conclusions. This chapter describes the data used in my master´s thesis. Chapter 4 is related to US stocks. The sample of stocks consists of the 127 biggest financial US titles with a market capitalization of at least 2 bn USD quoted on NYSE. The last row of Table 2 shows the descriptive statistics of this sample. The empirical data stem from Wharton Research Data Service-CRSP.

The next chapters describe stock returns and liquidity issues in Europe. In the statistical analysis, I worked together with almost 500 different European stocks from 12 countries: France, Spain, Italy, Germany, the UK, Austria, Poland, Belgium, Holland, Russia, Portugal, and Switzerland. The sample consists of 59 financial institutions with a short sale ban and 61 financial institutions without a short-selling ban. The rest of the sample stocks are members of stock indices: CAC 40, FTSE MIB, IBEX, FTSE 100, DAX, BEL20, ATX, AEX, MICEX, SLI, PSI20, and WIG20.

Local financial authorities in Spain, Italy, France, and Belgium published the list of stocks with a short sale restriction in 2011. There was no ban in the UK, Germany, Austria, Poland, Holland, Russia, Portugal and Switzerland, so the local financial authorities did not publish any lists of relevant financial institutions. However, it is necessary to collect a sample of financial institutions from countries without a short-selling ban as well in order to compare it to the countries with a short-selling prohibition. Short selling was banned in those countries during the 2009 financial crisis. I assume the list of short restricted financial institutions would be similar if those countries imposed a ban again in 2011. Some of the financial institutions from 2009 are not listed anymore because of bankruptcy or mergers, but the core of this list remains. There was no list of banned financial institutions during the debt crisis or financial crisis in Russia and Poland. I chose all financial institutions contained in the Micex and WIG 20 for the bank sample. The list of financial institutes and relevant stock market indices analyzed can be found in Appendix 1.

I made the statistical analysis with Stata Statistics 10.0 and the graphs and tables with MS Excel. Stock prices (daily closing price, daily price low, and daily price high), shares outstanding, and daily volume data stem from Wharton Research Data Service-Compustat. Bid and ask prices were downloaded from Datastream. Bid and ask prices are measured at the market close. A test for heteroscedasticity was undertaken for each regression and was passed with flying colors. My research periods consist of the period from July 1, 2011 to the end October 30, 2011 and its subperiods. A 1% statistical significance is marked with three stars, a 5% statistical significance is marked with two stars and a 10% statistical significance is marked with one star.

Why are short selling bans imposed on financial stocks?

Financial institutions are the first vote if a short-selling ban comes into discussion. A short-selling ban is usually extended into other sectors in the second step. There are several possible explanations for this. The first explanation is that financial institutions have the biggest impact on the whole economy and it is necessary to protect them. Others argue that financial institutions are more sensitive to bearish markets and short selling.

To analyze the impact of short-selling on stock prices, I regress U.S. financial institutions stock returns from July 29, 2011 – August 15, 2011 on a normalized measure of the change in short interest over the period. The US did not ban short selling, so the short interest position changed freely. In addition to debt crisis, the U.S. rating was downgraded [3] during this period. Battalio, Mehran and Schultz (2011) have already done the analysis for the whole market. I added the regression for financial institutions. I chose all financial institutions traded on the NYSE with a market capitalization higher than 2 billion USD, altogether the 127 biggest US financial institutions. I do not distinguish between weak and strong banks; the analysis is made for the financial sector in general. Table 1 lists the result for all stocks published by Battalio, Mehran and Schultz (2011). The last row of this table contains comparable results for financial institutions.

These results are very similar for the whole market and for the financial institutions. Paradoxically, stocks with a larger increase in short interest had higher returns over this period. This paradox is a little stronger for financial institutions. Regardless, R2 is always very small and coefficients are not statistically significant. The impact of short selling on stock return is not different by financial institutions according to this empirical analysis. From the short selling point of view, there is no strong reason to prefer a short-selling ban on the financial sector to other sectors (utilities, basic materials, etc.). I focus only on financial institutions in the following chapters because my research sample contained no other short sale restricted sectors.

As I proved, there is no strong difference between the financial and other sectors in general. Brunnermeier, and Oehmke (2008) presented a theoretical model about the negative impact of short selling on the fundamental value of financial institutions. This model proves that the impact of short selling on some types of financial institutions may be enormous. Financial institutions with weak balance sheets are especially victim to short selling according to this model.

Brunnermeier and Oehmke (2008) show that short sellers can cause huge troubles to financial institutions. This action of short sellers does not work in other sectors. The reason is that financial institutions are subject to leverage constraints which are given by financial authorities. If the leverage is high, they are forced to reduce it somehow. There are several possibilities how to do it. The model assumes the only possibility is to sell some assets at a fire sale price. Additionally, the model assumes that the market capitalization of the company represents the necessary equity which is given by the financial authority. The model implies that banks with weak balance sheets are vulnerable to short selling.

The theoretical model of Brunnermeier, Oehmke (2008) has three time periods, t=0,1,2. At t=0, a financial institution has invested in X units of a long-term asset. The financial institution has debt with face value of. This debt is due to be paid off at t=2. This debt can be paid at t=1 if the financial institution is forced to reduce leverage. The expectation at t=1 is that the long-term asset will pay off R at t=2. If the financial institution is forced to repay some debt at t=1, it has to sell some long-term assets at t=1. Early liquidation is subject to a discount; the early liquidation decreases the pay off, where. Therefore, the liquidation decreases the fundamental value of the financial institution. The essence of the model is that short sellers pressure equity, which decreases the leverage ratio below the critical level, and so the financial institution is forced to sell some long-term assets at a fire sale price to satisfy its capital requirements. This unprofitable sale of assets decreases the fundamental value of the financial institutions, and therefore short selling is profitable. A deeper analysis of this model is described below:

The maximal leverage prescribed by the financial authority is :

There are two types of investors in the model: the long-term investor and short-sellers. The long-term investors offer demand to the short sellers. In other words, long-term investors form a demand curve that short sellers can sell into. The scope of this demand curve is given by the formula: . Long-term investors determine the intercept and the scope. The activity of short sellers, i.e. the amount of shares sold short, is expressed in S. There is some similarity between this curve and the curve of Kyle (1985) related to information asymmetry.

In the beginning, we assume no leverage constraints given by financial authorities. In the second step, I add these constraints into the model.

The fundamental value of the financial institution is calculated by:. The intercept of this equation with the demand curve determine the price of the stock. This situation is illustrated in Figure 1. There is only one fundamental value and equilibrium. Short sellers cannot change the fundamental value of the financial institution. Equilibrium is highlighted with a black point.

Figure 1

The situation changes if we add leverage constraints into the model. If the leverage of the financial institution breaks a critical value, then it is forced to sell of long-term assets. As I mentioned above, this sale price is worth and is not profitable. The fundamental price of the company with leverage constraints is lowered by. The reduction in equity value stems from the fact that long-term assets can only be sold at a discount. The price of the financial institutions is in this case.

Figure 2 Figure 3

From Figure 3 it is obvious that when short sellers take a large enough position, the fundamental value drops as a result of forced selling of long-term assets at fire sale prices.

Based on the model parameters, the financial institutions can be positioned into three different regions.

Well-capitalized region (Figure 2). There is no danger of predator short selling in this case. The condition needs to be satisfied. The financial institution does not have to unwind any of its long-term holdings in any case.

Vulnerability region (Figure 3). The financial institutions especially suffer from short selling in this region if the payoff of the long-term assets occurs in the interval. The leverage constraint is initially satisfied and the financial institution does not sell the assets if short selling is forbidden. If short selling is allowed, there are three possible equilibriums (prices) in this state. The middle equilibrium is not stable. If the short sellers know (and have market power), they can influence the price and the equilibrium with P=0 will prevail.

Constrained region. The leverage constraints are violated even in the absence of short selling in the region. Payoff of the assets has to be low: . There is a unique equilibrium in this state. If short selling is restricted, the financial institutions have to sell part of their long-term assets. If short selling is allowed, it unwinds all long-term assets and P=0.

This paper shows how short selling can destroy the fundamental value of a financial institution. On the other hand, some assumptions of this model are inconsistent.

Nguyen and Tang (2011) performed an empirical analysis related to financial institutions. They focused on short restricted US financial stocks in the 2009 financial crisis. Together 753 stocks were analysed by linear regressions and a Wilcoxon test. Their research period is a few days around the introduction of the short-selling ban. Their results indicate that the ban has a positive impact on all financial institutions [4] . However, the subgroup analysis is interesting. Medium/large commercial banks and brokerage firms benefited most from the ban, and small commercial banks and insurance companies benefited the least from this ban. The next subgroup analysis is relevant to the theoretical model by Brunnermeier and Oehmke (2008) described above. Nguyen and Tang (2011) proved that financial firms with a higher leverage and greater likelihood of financial distress experience especially high abnormal returns. Unfortunately, this analysis was made on a very short-term period around the ban. The average frequency of extreme negative daily returns is lower in the ban period than in the pre-ban period. The frequency of extreme positive daily returns increases from the pre-ban to the ban period. On the other hand, the authors admit volatility and liquidity problems caused by the ban. The S&P volatility index increases significantly from the pre-ban to ban periods as well.

Stock return analysis

On 11 August, 2011, the official justification by the Belgium Financial Authority for a short-selling restriction was as follows: "One of the aims of the measure is to limit the possibility of making a profit by disseminating misleading information." The European Supervisory Authority emphasized the requirements of the Market Abuse Directive ("MAD"), referring to "the prohibition of the dissemination of information which gives, or is likely to give, false or misleading signals as to financial instruments, including the dissemination of rumors and false or misleading news".

Research has an opposite opinion regarding this information. The theory considers information as a true signal which reflects the fair fundamental value of the stock. The ban for short sellers with bad fundamental information slows down price discovery. Another interpretation of slow price discovery is that a short-selling restriction temporarily increases stock prices. The theory is quite ambiguous about the long-term effect of a short-selling ban. Shkilko, Van Ness and Van Ness (2011) presented another argument against the financial institutions authorities’ justification statement. According to their empirical study, short sales may increase downward pressure on prices even in the absence of negative information. In any case, a short-selling prohibition is not as strong as it seems to be. Market participants could use ETFs, put options, CDS, or other derivative instruments to bet on falling stock prices. Based on Boehmer, Jones and Zhang’s (2009) empirical research, shorting activity drops by about 65% during a financial crisis short-selling ban. Recall that market-makers are able to short as part of their market-making and hedging activities, and these are probably the short sales that we observe during the ban period.

Haruvy and Noussair (2006) did some non-traditional empirical research. They made an experiment with students, who represented the traders on the stock exchange. The students submitted their orders. The researchers made detailed statistics of price development, bubble creation and so on. The data showed that short selling has the effect of reducing market price. Prices in the markets with relatively loose short-selling constraints exceeded the fundamental value only occasionally, for brief intervals, and by relatively small magnitudes. However, allowing a sufficiently large short-selling capacity reduces prices to levels below fundamental values. The authors admitted the availability of short selling reduces transaction prices.

Bid-ask spread analysis

During a short-selling ban period, an increase in the bid-ask spread is an often mentioned fact in the literature. It is necessary to check the origin of the bid-ask spread to justify this increase. A bid-ask spread consists of three basic components: Adverse selection costs, inventory holding costs and order processing costs. Logically, some of these components have to increase in case of a bid-ask spread increase. I will now briefly examine the connection between each of the components and the short-selling ban.

Order-processing costs are those directly associated with providing the market making service and include items such as the exchange seat, floor space rent, computer costs, informational service costs, labor costs, and the opportunity cost of the market maker’s time. Order processing costs are fixed costs per share. Charoenrook and Daouk (2005) proved that a short-selling restriction reduces volume by 15%. Because these costs are largely fixed, at least in the short run, their contribution to the size of the bid-ask spread should fall with trading volume; that is, the higher the trading volume, the lower the bid-ask spread. There are factors that cause market makers costs per share to increase as the scale of trades is decreased. This effect is generally called economies of scale and in our specific case diseconomies of scale. On the other hand, a diseconomy of scale is a long-run concept and short-selling bans are usually in force for only a few months. It is very difficult to quantify this and the final effect would be small if at all.

Inventory-holding costs are the costs that a market maker incurs while carrying positions acquired in supplying investors with an immediacy of exchange (liquidity). There are two obvious points to take into consideration: the opportunity cost of funds tied up in carrying the market maker’s inventory and the risk that the inventory value will change adversely as a result of security price movements. With respect to the opportunity cost of funds, Demsetz (1968) argues that price per share is a reasonable proxy. His argument is that the relative spread (bid- ask spread divided by the bid-ask midpoint) should be equal across stocks, holding other factors constant, or the higher the share price, the higher the spread. Market makers, however, try to reduce or close out positions before the close of trading each day. If positions are opened and closed on the same day, the marginal cost of financing is zero. Moreover, even if inventory is carried overnight, it is not clear whether it represents a cost or a benefit. If, during the day, most customer orders are buys, the market maker may be short inventory, in which case he will earn (not pay) interest overnight. With respect to security price movements, market makers frequently hold undesired portfolio positions that do not lie on their efficient frontier. It is necessary to note that the decision to ban short selling did not apply to financial intermediaries acting as market makers or liquidity providers in France, Italy, Spain and Belgium. The market makers could freely change their inventories like before the short-selling ban. Despite no restriction for market makers, there are some reasons to increase inventory holding costs. The inventory holding costs are a function of volatility and price discovery as well. If the market makers believe that stocks are not fairly priced during a short-selling ban period and these market makers do not have information about a fair price, they will increase the bid-ask spread because of the higher inventory holding risk.

The assumption of asymmetric information generates adverse selection costs, which force dealers to open the bid–ask spread. Such a spread is the premium that dealers demand for trading with traders with superior information. The market maker does not know if the order received stems from somebody with better information (insider) about the company or from a liquidity trader [5] . The market maker’s loss caused by insider traders is compensated by profit earned by liquidity traders. There are different opinions on the adverse selection effect. As mentioned in Chapter 2, Appel and Fohlin (2010) argue that a ban on short sales disproportionately restricts informed traders from selling, mitigates the adverse selection problem and thereby lowers spreads.

In any case, my empirical research and most other empirical studies have found evidence about a statistically significant increase in the bid-ask spread for short restricted stocks. Therefore, the increase in the order processing costs and inventory holding costs outweigh the decrease in adverse selection.

Liquidity costs estimation

Battalio, Mehran and Schultz (2011) and Boehmer, Jones and Zhang (2009) calculated the costs imposed by short sale bans in the US market during the 2009 financial crisis. They calculated these costs for equity and derivative markets separately.

The average dollar trading volume for banned stock during the ban was $66,749,000 in the 2008 financial crisis. There were 404 financial stocks subjected to the short sale ban. The increase of the effective half-spread was 0.0016%. The duration of the short sale ban was 14 days. If we multiply all figures, the total costs were $604,051,750. In other words, the excess liquidity costs of one stock per day were $106,789. This estimate does not include the costs associated with mutually beneficial trades that did not occur because of the inflated liquidity costs.

Furthermore, the options market suffered higher costs because of the short sale restriction during the 2008 financial crisis. Battalio and Schultz (2011) and Battalio, Mehran and Schultz (2011) estimate the inflated trading costs paid by liquidity demanders in option markets during the short sale ban at more than $110 million in inflated liquidity costs on the first day of the ban. The inflated costs did not disappear once the option market makers were granted their exemption from the short sale ban. Battalio and Schultz (2011) attribute these inflated costs to the regulatory uncertainty that prevailed during this period. Overall, Battalio and Schultz estimate that liquidity demanding investors paid more than $505 million in inflated liquidity costs during the 14-day short sale ban. Altogether, the estimated inflated cost of liquidity attributable to the short sale ban in the U.S. equity and option markets exceeds $1 billion.

Volatility analysis

Economic theory and empirical analyses are not absolutely consistent in their prediction of volatility consequences, according a study by the Financial Service Authority UK. Based on theoretical models, a short-selling ban would lead to lower volatility for restricted shares (when compared to the market). Empirical results for the relative change in volatility of restricted stocks compared to the market after the introduction of the temporary ban are inconclusive.

Chen and Zheng (2005) examined the impact of short selling on the volatility and liquidity of the stock market. Their empirical analysis stems from the Hong Kong Stock Exchange. They used a Granger causality test to provide the empirical analysis. The notion of Granger causality is based on a criterion of an incremental forecasting value. Variable X is said to "Granger-cause" the Y variable if Y can be predicted better from the past of X and Y together than the past of Y alone, with other relevant information being used in the prediction. Therefore, from the results of Granger causality tests and the figures from an impulse response function, they could conclude that short-selling volumes do not Granger-cause market volatility, but market volatility Granger-causes short-selling volumes. With the increase of market volatility, short-selling volumes increase too. To sum up, they found no evidence that short selling will make the market more volatile. In a word, the empirical results indicate that when short selling is possible, aggregate stock returns are less volatile and there is greater liquidity in the market. The authors provide a logical explanation for their results; for example, when a stock price is too high to be reasonable, some rational investors will undoubtedly short sell this stock, and therefore there will be an obvious increase in the stock supply on the market. On the one hand, it will relieve the severe imbalances of supply and demand for this stock. On the other hand, it will reveal the higher priced signal of this stock to other investors and make the price return to a reasonable one in the end. Furthermore, when the bubble of a higher priced stock bursts, the short sellers will repurchase the stock, which has been short sold, to close their short positions. While it will increase the demand for this stock, and the price of it, it will also reveal the lower priced signal of this stock to other investors, and then make the price reasonable in the end. It is through this mechanism that short selling can make the market more stable and liquid.

Schwartz and Norris (2010) examined whether the market capitalization of a company influences its reaction to a short-selling ban. The sample consisted of 60 US financial stocks held by the Vanguard Index mutual fund [6] . The firms were divided into small, medium and large capitalization firms. Small firms had a market value of $1 billion or less. Medium-sized companies had a capitalization of $1 billion-$4 billion and large firms had capitalizations over $4 billion. The firms were divided into restricted and non-restricted portfolios as well. The restricted ones were financial firms whose stock was included in the temporary short-selling ban the SEC ordered from September 19 to October 8, 2008. An analysis of variance was performance to study the effect of the short-selling restrictions on the volatility of the stock portfolios. Variance measures risk, so an increase in variance implies an increase in risk for investors. During the 30 days before the introduction of the ban, small-sized companies had the lowest variance in the restricted and non-restricted sample. The variance of the restricted portfolio was similar to the non-restricted portfolio. On September 19th, when the temporary ban was placed on the stocks, there was a significant increase in variance for each size portfolio. The small firms kept the lowest level of volatility in the 14 days after the introduction of the short-selling ban. The variance of the restricted portfolio was twice the variance of the non-restricted portfolio during this period. This difference is primarily caused by large companies. Restricted large companies had a significantly higher volatility than large companies in the non-restricted portfolio. The large restricted firms in this analysis were closely followed by the media and were regularly in the news. The additional media coverage may have produced this significant increase in volatility for those firms. This ban on financial institutions lasted only 14 days in the US. When the ban was lifted, the small firm portfolio continued to witness a significant increase in variance. Unlike the small firms, both the mid-size and large-size firms saw a decrease in variance. The variance of restricted and non-restricted portfolios returned to the same level in the period after the short-selling ban.