<<

A Look at Bear Raids: Testing the bid test

By

Michael G. Ferri George Mason University School of Management/ MS 5F5 Fairfax, VA 22030 Tel. (703) 993-1858, FAX 993-1870 [email protected]

Stephen E. Christophe George Mason University School of Management/ MS 5F5 Fairfax, VA 22030 Tel. (703) 993-1767, FAX 993-1870 [email protected]

and

James J. Angel Georgetown University The McDonough School of Business Room G-4 Old North Washington, D.C. 20057 Tel. (202) 687-3765, FAX 687-4031 [email protected]

March 2004

Direct all correspondence to: James J. Angel, Georgetown University. We thank the Nasdaq Market for providing the data. Christophe gratefully acknowledges financial support from George Mason University. Angel and Ferri gratefully acknowledge financial support from the Nasdaq Educational Foundation, and especially thank Laura Levine of the NEF. The views expressed in this paper, however, are those of the authors and do not necessarily reflect the views of the Nasdaq , Inc., the Nasdaq Educational Foundation, or anyone else.

A Short Look at Bear Raids: Testing the bid test

Abstract

The U.S. market applies a variety of restrictions to short sales. Some of these restrictions, such as the NYSE uptick rule and the Nasdaq bid test, seek to inhibit short selling during declining markets to deter “bear raids.” We examine a natural experiment to see if the bid test is effective:

In the Nasdaq market, the bid test applies only to Nasdaq’s National Market and not to stocks listed in the Nasdaq SmallCap Market. We compare the experience of SmallCap stocks with comparable National Market stocks during the turbulent market decline of 2000 and 2001.

Using previously unavailable daily data on short selling, we examine the frequency of short selling, and the association of short selling with rapid declines stock prices. We find that, if anything, there is actually less short selling for the SmallCap stocks, and that abnormal short selling is not more correlated with price declines than for the matching National Market stocks.

This indicates that other regulatory protections in place are sufficient to protect investors, and that a bid test or uptick rule is unnecessary.

2

A Short Look at Bear Raids: Testing the bid test

I. Introduction

Short selling has often been blamed for many ills in securities markets.1 Those who lose money

from price declines often blame short sellers for unfairly manipulating prices downward,

regardless of whether short sellers were actually active in the stock or not. Profiting from the losses of others often causes deep resentments, especially among those who have lost money.

Issuers often detest short sellers because they may spread negative information about the issuer.

The United States imposes a variety of regulations on short selling in the equity markets.2 These regulations include:

1) Disclosure rules: All sale orders are required to be marked as long or short. Thus, the

broker, market maker, or specialist executing the trade knows that it is a short transaction.

Furthermore, the level of short interest is disseminated by the markets once a month.

2) The location rule. Investors are generally required to make an “affirmative

determination” that they can indeed borrow the stock before they sell it.

3) Margin requirements. Under Federal Reserve Regulation T, short sellers are generally

required to post a 50% margin before engaging in a short sale. Brokerage firms are

typically free to set their own maintenance margins, which are usually in the 25% range.

3

4) The uptick rule. In 1938, the SEC adopted rule 10a-1, also known as the uptick rule.

This rule, which applies only to NYSE and AMEX-listed securities, permits short sales

only at a price above the previous price, or at the same price as long as it was above the

last different prior price.

5) The bid-test: In 1998, Nasdaq implemented a bid test for Nasdaq National Market stocks.

This rule prohibits short sales at the bid price if the bid was less then the previous bid

price.

Recently, the U.S. Securities and Exchange Commission (2003) has proposed replacing the

uptick rule, which applied only to the NYSE and AMEX, with a uniform bid test that would apply uniformly to all NYSE, AMEX, and Nasdaq National Market stocks. The proposed rule

would prohibit any short sales at the bid price.

As the SEC (1998) has indicated, one of the motivations for the uptick or similar rule is to

prevent “bear raids.” In a bear raid, a trader or group of traders allegedly manipulates the price

of a stock downward through aggressive and repeated short selling. By exhausting all of the

liquidity available at the bid and below, repeated short selling rapidly drives the price down. Of

course, there is a fine line between price manipulation and price discovery. If a stock’s price is

truly overvalued, a bear raid actually helps the market get to the right price faster, and thus

enhances market efficiency.

The problem occurs if a bear raid can push stock prices below the “correct” level by taking

advantage of, or even creating, frictions in the market. Suppose, for example, that bear raiders

4

without any information target a less liquid stock. The repeated informationless selling sends a false signal that others in the market interpret as informed selling. The other investors falsely infer that the bear raiders possess negative information, and thus reduce their own estimates of the value of the stock. This leads to a lengthy period of undervaluation in the stock before the market recovers. Such undervaluation raises the cost of capital for the firm, leading it to forgo investment in projects that otherwise would have a positive NPV for the firm and for society.

However, it is an open question about whether such informationless bear raids are a threat in the modern world. If so, what rules, if any, are needed for investor protection? While the anecdotes from 70 years ago are entertaining, they are not necessarily the best basis for making public policy. It may well be the case that the other restrictions on short selling have made any type of uptick rule or bid test unnecessary. For example, the disclosure rule warns market makers or specialists that a trade is short, revealing information about the potential motives (or information level) of the seller. The market maker and/or specialist can then incorporate this information into her or his trading decisions.

The location rule inhibits short selling by forcing shorts to actually locate the shares before shorting. Otherwise, it would be easier to engage in “naked” short selling in which the shares are sold short without actually borrowing the shares. The naked short might expect to cover the shares later in the day, borrow the shares later, or else reply upon the leniency of the settlement system for failures to deliver.3

5

The requirement for 50% margins also inhibits short selling by forcing short sellers to put up

significant capital. Although the 50% margin may appear at first glance to be symmetric with

the 50% margin for long buying, it is actually very different. On the long buying side, margin

requirements protect the banking system from loan losses in the event of sudden declines in

value. Indeed, the protection of the banking system is one of the reasons why Congress, in the

Securities Exchange Act of 1934, charged the Federal Reserve, not the SEC, with regulating

margins. With short selling, the commercial banking system is far less involved, and is certainly

not making cash loans.

Advances in computers and communications technology over the last 70 years since the advent

of the uptick rule have increased the availability of information to market participants. The rapid

dissemination of information via the internet (not to mention radio, TV, and cheap long distance

calls) makes it easier for investors to be well informed. Because information is much more readily available, a false signal from an informationless bear raid is much more likely to be detected quickly, or just plain ignored with all of the other good information out there.

Improved regulation also makes it harder to pull off a price manipulation through a bear raid.

Improved accounting disclosures, disseminated rapidly over the internet, puts more factual

information in the hands of investors and makes it harder for informationless bear raiders to

cause havoc. Using computers, regulators can quickly determine who bought or sold a stock at a

given time using the so called “Blue Sheets” reports. This ability to track down and punish

miscreants helps to deter unlawful manipulation.

6

With these improvements in the market over the last 70 years, it is an open question whether an uptick rule or bid test is needed. Fortunately, there is a natural experiment which sheds light on this issue. When Nasdaq instituted its bid test in 199, it was only applied to National Market

stocks – not the stocks listed on the Nasdaq SmallCap market. Although stocks on the SmallCap tend to be smaller than National Market stocks, there is still quite an overlap in terms of price, market capitalization, and trading volume. It is thus possible to look at differences in the trading

of similar stocks in the two markets to see whether the bid test makes any difference.

During the time period of this study, the trading mechanisms that Nasdaq used to trade the stocks

for both the SmallCap and National Market systems were identical. (In earlier years, SmallCap

stocks had less transparency.) Real time quotations for both classes of stocks were disseminated over the Nasdaq quote system, and trades for both classes of stocks were required to be reported

to the tape within 90 seconds. The only major difference in the trading rules between the two markets was the bid test. The only other difference between the SmallCap and the National

Market is that National Market stocks are automatically exempted from all 50 state “Blue Sky” laws, whereas the SmallCap stocks were not. This meant that SmallCap stocks generally had to file registration papers with securities commissions in all 50 states upon their initial listing,

which added additional paperwork to the listing process.4 However, as most SmallCap stocks have filed in all 50 states, this does not generally affect the trading of their stocks.

7

II. Daily Short Sales Data

We obtained daily short sales data for each Nasdaq-listed stock from the Nasdaq Stock Market,

Inc. This data contains the number of trades, along with the reported volume, which was marked short or short exempt. These data are much more detailed than the monthly short interest data which have been used in prior studies of short selling. This data actually shows the number of shares shorted each day, while prior studies only had access to short interest levels once a month.

The data are from September 1, 2000 though August 16, 2001, a tumultuous period in which the

Nasdaq composite index dropped over 50%, including a large number of large price declines on individual stocks. If bear raids are ever a problem, they should have been during this period of volatile decline.

For each day, we selected all Nasdaq SmallCap-listed stocks that met the following selection criteria:

• Stock price greater than $5.

• At least 10 trades in that day

• At least one trade that was reported as a customer short trade.

• Market capitalization greater than $25 million.

We chose $5 as the cutoff because other institutional barriers, such as excessive margin requirements, impede short selling below $5. Many brokerage firms impose higher margin

8

requirements on stocks below $5, thus reducing the supply of lendable shares and increasing the

capital needed in order to sell short.5 Because stocks are sometimes “hard to borrow”, and we

did not have access to historical lists of hard to borrow stocks, we used the criteria of at least one

customer short trade to indicate that a stock could be borrowed and shorted. To further reduce

the chance that our stocks were hard to borrow, we required a minimum of $25 million market

capitalization on both the samples and controls. In addition, we limited the pool to National

Market stocks with a market capitalization less then $500 million so that the controls would be

close in market capitalization as well as price and industry to the sample stocks.

For each day, we then matched each SmallCap sample stock with a Nasdaq National Market

control stock in the same industry (measured by the two digit SIC code) and with the same

quotation tick size.6 Because the quotation tick size prior to decimalization changed at $10, we

further required that the control stocks be in the same price bin as the sample stocks, either under

$10 or above $10. We then chose those control stocks with market capitalization closest to the

sample stocks.

This process yielded 2,275 observations. Table 1 displays descriptive statistics on the SmallCap sample observations and the associated National Market control observations. The mean price of the sample ($10.19) and control ($11.09) stocks match closely, $10.19 compared with $11.09.

Mean market capitalization of the sample ($122 million) is still below, however, that of the controls ($180 million).

Please insert table 1 here.

9

III. Methodology and results

If the bid test were effective in restricting short selling, we would expect to see more short

selling in SmallCap stocks than in the National Market stocks. Furthermore, if there is abusive

short selling that could be prevented by a bid test, such as “bear raids”, the evidence of such

abuses, such as short-selling induced price declines, should be visible in the SmallCap market.

This section will present the empirical results comparing the Small Cap sample with the National

Market controls. Our results show no evidence of abusive short selling in the SmallCap market.

In particular, we examine the following indicators:

Overall level of short selling

We first examine overall levels of short selling, for both the SmallCap stocks and the controls. If the bid test in the National Market deterred short selling, one would expect to see more short selling in the SmallCap stocks. Our results indicate that the frequency of short selling is virtually the same in the SmallCap stocks as in the controls.

Table 2 presents results looking at the percentage of trading volume reported as customer short sales, and at the percentage of total trades reported as customer short trades, for both the

SmallCap and the controls. For the SmallCap sample, customer short trades averaged 3.54% of reported volume, compared with 3.43% of reported volume for the sample stocks. The difference of 0.11% has a statistically insignificant t-statistic of 0.58. The results are similar for

10

the percentage of total transactions, 3.46% for the sample and again 3.43% for the controls, with

a t-statistic of 0.18. To make sure that the results are not driven by outliers, we also examine the

medians. The median for the customer short volume was 1.00% for the sample compared with

1.17% for the controls.

Please insert table 2 here.

In order to see if there is an effect based on price level, we bifurcated the sample into two

categories, stocks between $5 and $10 and stocks above $10. The results, also reported in table

1, are similar: Rates of customer short selling are virtually indistinguishable between the sample

and the controls.

Impact of Decimalization

The sample period coincides with the decimalization of the Nasdaq Stock Market. During the

early part of our sample, Nasdaq displayed quotations for stock quotations higher than $10 in

increments of 1/16th and quotes between $1 and $10 in increments of $1/32. However, some

electronic communication networks (ECNs) could and did quote stocks in increments of

$1/256th. Even though the Nasdaq quotation system rounded quotes, there was no Nasdaq

regulation that required trades to take place on the rounded price increments. Nasdaq systems

permitted participants to report trades to the tape in intervals of $/256th.

On March 12, 2001, Nasdaq began to decimalize with a pilot of 15 securities. On March 26,

2001, Nasdaq added 199 additional securities, and finished decimalization on April 9, 2001. As

11

is well reported in the literature, bid-ask spreads fell dramatically subsequent to decimalization.

It could be the case that the reduction in the tick size to one cent may have reduced the efficacy of the bid test. We thus partition our sample into the pre-and post decimalization status of the stocks involved. The results, again reported in table 2, are the same: There is no significant difference in the rate of customer short selling between the SmallCap stocks and the controls.

Frequency of extreme events

Even though the rate of short selling may not be any higher in the SmallCap, perhaps there might still be some abuses that show up only in rare, but extreme events. These would be observable in the tails of the distribution. We thus examine whether there are more extreme observations of large negative returns and high degrees of short selling. Surely, if a bid test deterred bear raids in the National Market, then the SmallCap sector should see an increase in extreme observations in which there are highly negative returns along with high rates of short selling.

Table 3 displays the frequency breakdown of the daily return compared with the rate of short selling in the SmallCap sample and the National Market controls. We look for evidence of the frequency of bear raids by examining events in which a high negative return occurred on the same day as a high rate of customer short selling. There are 13 such events, 0.57% of the total observations, in the sample in which customer short selling accounted for more than 10% of the reported trading volume and in which the stock declined by more than 10%. If the bid test in the

National Market deterred such raids, then one would expect far fewer such events in the National

12

Market control stocks. This is insignificantly different from the 11 events, 0.48%, in the

National Market control sample.

Please insert table 3 here.

If one looks at all days of negative returns, the results are similar: 115 events (5.05%) compared with 112 (4.92%) in the controls.

In table 4, we also look at events in which only 5% of volume was reported as customer short sales, with similar results: 25 (1.10%) observations with returns worse than -10% in the sample and 24 (1.05%) in the controls. If we examine all observations with negative returns and customer short selling more than 5%, we find 212 (9.3%) observations in the sample and even more, 219 (9.6%), in the controls. Thus, there are no telltale “bear tracks” that would indicate that the there are more bear raids in the SmallCap sample than in the controls.

Please insert table 4 here.

Short selling and intraday volatility

Because short selling is sometimes blamed for increasing volatility, it is useful to examine the relationship between short selling and intraday volatility, and to see whether this relationship is different in the SmallCap and the National Market. If short selling did in fact increase volatility,

13

one would expect to see a positive correlation between intraday volatility and short selling.

Table 5 demonstrates the opposite: There is a strongly significant negative correlation between short selling and intraday volatility, which we measure by the coefficient of variation in the prices for the day.7 Times of high degrees of short selling are associated with less intraday volatility, indicating that short sellers were actively stabilizing prices.

Please insert table 5 here.

Short selling and price declines

The major evidence of abusive short selling would be a higher level of short selling when stocks decline more. Thus, the correlation between short selling and the return on the stock should be more negative for the SmallCap market than for the National Market when stocks are declining.

However, we see a strong positive correlation when stocks are declining. A higher proportion of short selling is associated with a more positive and hence less negative return. Even on days when shares are declining, short selling is correlated with stabilizing, rather than destabilizing, events.

Please insert table 6 here.

14

IV Conclusions

Using previously unavailable data, we explore the differences in short selling behavior between

comparable stocks in the Nasdaq National Market, which has a bid test and the Nasdaq SmallCap

Market, which has no bid test. If a bid test added additional protections against abusive short

selling, this should be visible in differences in short selling activity in these two markets. Alas,

we find no evidence that short selling is any more frequent in the SmallCap market. Nor do we

find any evidence that shows that short selling increases volatility or more rapid price declines in

the SmallCap market. Indeed, short selling appears to have a stabilizing effect.

The conclusion is that a bid test is unnecessary for investor protection. Advances in the dissemination of information, along with other regulatory protections, such as the disclosure and location rules, are apparently enough to prevent the fabled “bear raids” of yore.

15

References

Aitken, Michael J., Alex Frino, Michael S. McCorry, and Peter L. Swan, 1998, Short sales are almost instantaneously bad news: Evidence from the Australian , Journal of Finance 53, 2205-2223.

Alexander, Gordon J. and Mark A. Peterson, 2002, Implications of a reduction in tick size on short-sell order execution, Journal of Financial Intermediation 11, 37-60.

Asquith, Paul, and Lisa Meulbroek, 1996, An empirical investigation of short interest, Working paper, Harvard Business School, Harvard University.

Brent, Averil, Dale Morse, and E. Kay Stice, 1990, Short Interest: Explanations and Tests, Journal of Financial and Quantitative Analysis 25, 273-289.

Chen, Honghui, and Vijay Singal, 2003, Role of speculative short sales in price formation: Case of the weekend effect, Journal of Finance 58, April

Choie, Kenneth S., and S. James Hwang, 1994, Profitability of short-selling and exploitability of short information, Journal of Portfolio Management 20, 33-38.

D’Avolio, Gene, 2002, The market for borrowing stock, forthcoming: Journal of Financial Economics.

Dechow, Patricia M., Amy P. Hutton, Lisa Meulbroek, and Richard G. Sloan, 2001, Short sellers, fundamental analysis and stock returns, Journal of Financial Economics 61, 77-106.

Desai, Hemang, K. Ramesh, S. Ramu Thiagarajan, and Bala V. Balachandran, 2002, An Investigation of the Informational Role of Short Interest in the Nasdaq Market, Journal of Finance 5, .2263-2287.

Diamond, Douglas W., and Robert E. Verrecchia, 1987, Constraints on short-selling and asset price adjustment to private information, Journal of Financial Economics 18, 277-311.

Geczy, Christopher C., David K. Musto, and Adam V. Reed, 2002, Stocks are Special Too: An Analysis of the Equity Lending Market, forthcoming: Journal of Financial Economics.

Harris, Milton, and Artur Raviv, 1993, Differences of opinion make a horse race, Review of Financial Studies 6, 473-506.

Safieddine, Assem Jr., and William J. Wilhelm, 1996, An Empirical Investigation of Short- Selling Activity Prior to Seasoned Equity Offerings, Journal of Finance 51, 729-749.

16

Securities and Exchange Commission, 1999, Concept Release: Short Sales, 17 CFR 240, Release No. 34-42037; File No. S7-24-99, http://www.sec.gov/rules/concept/34-42037.htm.

Securities and Exchange Commission, 2003, Proposed Rule: Short Sales, 17 CFR 240 and 242, Release No. 34-48709; File No. S7-23-03 , http://www.sec.gov/rules/proposed/34-48709.htm

Senchack, A.J. and Laura T. Starks, 1993, Short-sale restrictions and market reaction to short- interest announcements, Journal of Financial and Quantitative Analysis 28, 177-194.

Solomon, Steve, 1997, Follow the Money: The Cost of an IPO, Inc Magazine, June. http://www.inc.com/magazine/19970601/1255.html

17

Table 1 SmallCap Sample and National Market Controls Descriptive Statistics

This table contains descriptive statistics for the Nasdaq SmallCap sample stocks and their associated control stocks from the Nasdaq National Market over the period from September 1, 2000 through August 16, 2001. The sample contains for each day all Nasdaq SmallCap stocks with a price greater than $5 and a market capitalization greater than $25 million and with at least 10 trades on that particular day and one trade reported as a customer short trade. The control stocks were matched by industry at the two digit SIC code level, tick size, and market capitalization from a universe of all Nasdaq National Market stocks that met the inclusion criteria and were less than $500 million in market capitalization. All statistics are mean unless otherwise stated. Standard errors are in parentheses.

Variable SmallCap Sample National Market Controls Number of observations 2,275 2,275 Price ($) 10.19 (0.11) 11.09 (0.14) Price – median 8.40 8.76 Market Capitalization ($ 122.34 (2.06) 179.84 (2.41) millions) Market Capitalization – median 93.36 152.25 Bid-ask spread (%) 1.70% (0.03) 1.50% (0.03) Number of trades per day 244 (6.21) 271 (11.84)

18

Table 2

Frequency of Short Selling in SmallCap Stocks versus Controls

This table presents data regarding the frequency of customer short selling in the Nasdaq SmallCap Market compared with matching control firms in the Nasdaq National Market from September 1, 2000 through August 16, 2001. Firms were matched as described in the text.

Median Mean Number of SmallCap National Median SmallCap National Difference t- observations Sample Market Difference Sample Market statistic Controls Controls Percent Percent Entire Sample Short 2,275 1.00% 1.17% -0.17% 3.54% 3.43% 0.11% 0.58 Volume Short 2,275 1.58 1.96 -0.10 3.46 3.43 0.03 0.18 Trades By Price Level: $5-$10 Short 1,430 0.92 1.07 -.15 3.44 3.25 0.19 0.78 Volume Short 1,430 1.64 1.96 -.20 3.47 3.49 -.02 -0.11 Trades By Price Level: Above $10 Short 845 1.08 1.37 -.011 3.72 3.73 -.01 -0.03 Volume Short 845 1.53 1.96 -.31 3.45 3.33 0.11 0.43 Trades By decimalization status: Pre decimals Short 1,536 1.00 1.18 -.08 3.39 3.50 -.012 -0.53 Volume Short 1,536 1.54 1.82 -.18 3.26 3.28 -.02 -0.11 Trades By decimalization status: Post decimals Short 739 1.01 1.15 -.15 3.87 3.28 0.59 1.57 Volume Short 739 1.76 2.30 -.39 3.89 3.76 0.13 0.43 Trades

19

Table 3

Frequency of extreme 10% short selling days with high negative returns SmallCap versus National Market

This table presents data regarding the frequency of customer short selling combined with days of high negative returns of more than -10% in the Nasdaq SmallCap Market compared with matching control firms in the Nasdaq National Market from September 1, 2000 through August 16, 2001. Firms were matched as described in the text.

SmallCap Stocks National Market Controls Percent Percent customer short customer short Daily return volume Daily return volume) Frequency Frequency Percent Percent Row Pct > Row Pct Col Pct 0 - 10% 10% Total Col Pct 0 - 10% > 10% Total < - 10% 120 13 133 < - 10% 117 11 128 5.14 0.48 5.63 5.27 0.57 5.85 Cell freq Cell freq 90.23 9.77 91.41 8.59 Row freq 5.83 6.05 Row freq 5.65 5.39 Column freq Column freq -10% - 0 833 102 935 -10% - 0 925 101 1026 36.62 4.48 41.10 40.66 4.44 45.10 89.09 10.91 90.16 9.84 40.44 47.44 44.66 49.51 0 96 16 112 0 85 18 103 4.22 0.70 4.92 3.74 0.79 4.53 85.71 14.29 82.52 17.48 4.66 7.44 4.10 8.82 0 – 10% 791 76 867 0 - 10% 810 65 875 34.77 3.34 38.11 35.60 2.86 38.46 91.23 8.77 92.57 7.43 38.40 35.35 39.11 31.86 > 10% 220 8 228 > 10% 134 9 143 9.67 0.35 10.02 5.89 0.40 6.29 96.49 3.51 93.71 6.29 10.68 3.72 6.47 4.41 Total 2060 215 2275 Total 2071 204 2275 90.55 9.45 100.00 91.03 8.97 100.00

20

Table 4 Frequency of extreme 5% short selling days with high negative returns SmallCap versus National Market This table presents data regarding the frequency of customer short selling combined with days of high negative returns of more than -5% in the Nasdaq SmallCap Market compared with matching control firms in the Nasdaq National Market from September 1, 2000 through August 16, 2001. Firms were matched as described in the text.

SmallCap Sample National Market controls Percent Percent customer customer short Daily return short volume Daily return volume Frequency Frequency Percent Percent Row Pct Row Pct Col Pct 0 - 5% > 5% Total Col Pct 0 - 5% > 5% Total < - 10% 108 25 133 < - 10% 104 24 128 4.75 1.10 5.85 4.57 1.05 5.63 81.20 18.80 81.25 18.75 5.82 5.98 5.57 5.90 -10% - 0 748 187 935 -10% - 0 831 195 1026 32.88 8.22 41.10 36.53 8.57 45.10 80.00 20.00 80.99 19.01 40.28 44.74 44.49 47.91 0 83 29 112 0 71 32 103 3.65 1.27 4.92 3.12 1.41 4.53 74.11 25.89 68.93 31.07 4.47 6.94 3.80 7.86 0 - 10% 713 154 867 0 - 10% 738 137 875 31.34 6.77 38.11 32.44 6.02 38.46 82.24 17.76 84.34 15.66 38.40 36.84 39.51 33.66 > 10% 205 23 228 > 10% 124 19 143 9.01 1.01 10.02 5.45 0.84 6.29 89.91 10.09 86.71 13.29 11.04 5.50 6.64 4.67 Total 1857 418 2275 Total 1868 407 2275 81.63 18.37 100.00 82.11 17.89 100.00

21

Table 5 Correlation between short selling and intraday price volatility

This table presents data regarding intraday price volatility, measured as the coefficient of variation of the price, and the frequency of customer short selling in the Nasdaq SmallCap Market compared with matching control firms in the Nasdaq National Market from September 1, 2000 through August 16, 2001. Firms were matched as described in the text.

Intraday Volatility Correlation SmallCap National Sample Market Controls Percent -0.091 -.068 Customer Short (p < .0001) (p < .0001) Volume Percent -0.114 -0.148 Customer Short (p < .0001) (p < .0001) Trades Number of 2,275 2,275 observations

22

Table 6 Correlation between short selling and returns

This table presents data regarding the correlation intraday price volatility, measured as the coefficient of variation of the price, and the frequency of customer short selling in the Nasdaq SmallCap Market compared with matching control firms in the Nasdaq National Market from September 1, 2000 through August 16, 2001. Firms were matched as described in the text.

Correlation with return Correlation with return On days with negative returns only All observations SmallCap National SmallCap National Sample Market Sample Market Controls Controls Percent -0.066 -.038 .061 -.016 Customer Short (p = .002) (p = .069) (p = .04) (p = .60) Volume Percent -0.051 -0.016 .087 -.002 Customer Short (p = .014) (p = .45) (p = .004) (p =.94) Trades Number of 2,275 2,275 1,068 1,068 observations

23

Endnotes

1 In a short sale, an investor first borrows the stock and then sells the stock. Later, the investor purchases shares to repay, or cover the stock loan. If the stock has declined in value, the investor earns a profit, at least before transactions costs. If the stock has gone up, the investor has loses money.

2 See SEC Rule 10a-1, Short Sales and NASD Rule 3350.

3 Brokerage firms have traditionally been reluctant to force buy ins when there are failures to deliver unless the failures are particularly large or long lived. In a few extreme cases, this has led to short positions outstanding far greater than the number of shares issued.

4 Solomon (1997) estimated that Blue Sky registration in all 50 states added about $35,000 to the cost of an IPO on the SmallCap market.

5 For example, E*trade imposes a 100% margin requirement on short selling for stocks less than $5.00. See https://us.etrade.com/e/t/estation/help?id=201081200. Some brokerage firms treat all securities less than $5.00 as not marginable and thus not shortable.

6 Just prior to decimalization, the quotation tick size was $1/16th for quotes above $10 and $1/32nd for quotes below $10. Subsequent to decimalization, which was phased in during the sample period, the tick size became $.01 for all stocks in our sample.

7 We use the coefficient of variation in price rather than the variance of trade to trade returns due to the nature of trade reporting in Nasdaq. Because trades are reported with different reporting lags by multiple entities using different clocks, it is not always possible to determine the precise ordering of trades.

24