The Concentration of Undelivered Shares Among Threshold Securities: Prospects of Stock Manipulation Using Naked Short Sales
Robert J. Shapiro November 2005
American public companies and their shareholders face a significant threat. Last year, researchers determined that naked short sales - short sales in which the shares are credited to buyer, but the short seller fails to borrow and deliver those shares - occur on a large scale, often extending for months at a time. New data now suggest that these "failures to deliver" or "fails" are concentrated in a relative handful of companies. This raises the prospect of naked short sales being used to manipulate some companies' stock prices. The enormous extent of naked shorting and its likely use in stock manipulation could threaten the integrity of our financial markets and international confidence in them.
Since the November 2004 release of research by visiting SEC economist Leslie Boni, it has been well established that on any given day, some 120 to 180 million shares that have been sold have gone undelivered for at least 3 days. The Boni research further established that these failures to deliver are widespread and often extended: On any day, about 80 percent of listed stocks have some undelivered shares. Moreover, more than 1,000 of about 6,100 listed stocks, or about 17 percent, have fails that have persisted for at least one month (20 business days), and more than 700 listed firms, or 11.5 percent, have fails at least two-months old (40 business days). It is virtually certain that the vast majority of these failures to deliver involve naked short sales, since extended uncompleted trades involving sales are typically canceled when the price changes.
In response, the SEC established Regulation SHO in late 2004, directing the exchanges to
use DTCC data on undelivered shares to publish daily "threshold security lists" of companies with outstanding fails equal to at least 10,000 shares and 0.5 percent of their total outstanding shares, and requiring that broker dealers locate and provide shares to back up subsequent short sales of threshold list stocks.
The Boni study did not establish whether the 120 million to 180 million daily failures-to¬deliver are concentrated among a relatively small number of firms or more randomly distributed across the 80 percent of stocks with some fails. The DTCC holds data on the precise number of fails for each stock, including every threshold list company; but has refused to release these data even to the affected companies. This question is critical for the integrity of U.S. equity markets, because as recent allegations in the Refco scandal attest, large-scale naked short sales can be used to manipulate the prices of stocks by flooding the market with sell orders that go undelivered for many days or months.
In response to a request under the Freedom of Information Act (FOIA), the SEC recently released the aggregate number of fails for listed NASDAQ and NYSE securities from
April 30, 2004 to April 30, 2005. As these data cover only fails of 10,000 shares or more -- the minimum for inclusion of the threshold-security list - they approximate the total fails for listed companies on those lists. Based on these data and on public information on trading volume and outstanding short interest for each threshold security, we can generally estimate the level of fails for each threshold security.
Summary of Findings
This new analysis strongly suggests that fails are highly concentrated among a relative handful of threshold security and are not in any way evenly distributed. This raises the prospect of large-scale naked short sales being used to manipulate stock prices.
.On any given day, we estimate that about half of some 125 to 160 NASDAQ¬NM and NYSE threshold securities account for 95 percent or more of all fails¬to-deliver, averaging about 1.5 million-to-2 million fails per-company.
.Based on a combination of trading volume and short interest, we estimate that on any given day, 10 or fewer threshold companies account for two-thirds of all fails, and 20 or fewer threshold companies account for 75 percent of fails
.We further estimate that on any given day, some 20 to 30 threshold companies carry fails of at least 1,000,000 shares each.
The analysis also indicates that designation as a threshold security does not appear to prevent a firm from bearing large numbers of undelivered shares for months at a time.
.Companies can remain on the t eshold list for months: Of the top 25 hreshold securities listed on February 15, 2005, 11 or 44 percent were also listed as threshold securities on both March 22,2005 and on April 26, 2005.
.Designation as a threshold security does not necessarily prevent subsequent naked short sales: In all 11 cases of persistent threshold securities, estimated failures-to-deliver rose from February to March; in five of the 11 cases or 45 percent, estimated fails in also were greater in April than in February.
.Designation as a threshold security does not greatly reduce subsequent short sales: In five of the 11 cases of persistent threshold securities or 45 percent, shorts continued to rise over the nine weeks studied. In the other six cases of persistent threshold securities, shorts fell by less than 12 percent.
.More important, designation as a threshold security does not necessarily prevent subsequent naked sort sales: In all 11 cases of persistent threshold securities, our estimated failures-to-deliver rose from February to March; in five of the 11 cases or 45 percent, our estimated fails also were greater in April than in February.
2. Our estimates rest on assumptions about links between trading volume, short interest, and fails. While these factors are certainly correlated, their precise relationship is unknown because, once again, the DTCC has refused to provide data on the .number of fails for particular securities. Even so, the analysis establishes that fails are highly concentrated, making it imperative that the DTCC both inform companies and investors of the extent of fails affecting each company, and crack down on large-scale, extended fails.
Procedure and Analysis:
Average fails per-threshold security. We analyzed the threshold security lists for three days - February 15, 2005, March 22, 2005, and April 26 2005. The undelivered shares for these three days totaled 136.8 million, 132.7 million and 131.6 million, respectively, based on the SEC release of aggregate fails. The FOI data cover listed securities and not OTC pink sheet and bulletin board companies. Setting aside those companies left the analysis with the following number of threshold companies: 159 securities on February 15,2005; 142 securities on March 22,2005; and 125 securities on April 26, 2005.
Based on the proposition that very thinly-traded stocks are not expected to account for large numbers of failures-to-deliver traded shares, we identified two groups of companies that were thinly traded. First, NASDAQ exchange data show that NASDAQ small-cap stocks account for only 5 percent of trading in NASDAQ-listed companies, with National Market stocks (NM) accounting for 95 percent. Based on those data, we excluded NASDAQ small cap securities from the analysis and appropriately adjusted the aggregate fails reported by the SEC.l By setting aside the small caps, the number of threshold securities declined to 122 for February 15,2005; 109 for March 22; and 93 for April 26.
Second, of the NYSE and NASDAQ-NM threshold-security companies, we identified those that treaded so lightly during the period of their inclusion on the threshold list, as to constitute "statistical outliers": We calculated the average trading volume for the NYSE and NASDAQ NM threshold securities during the month that included the threshold security list in question, and set aside those companies with trading volume equal to less than 10 percent of that average. In fact, our analysis showed that these companies accounted for less than 3 percent of all trades in threshold securities; and we adjusted the aggregate fails reported by the SEC to reflect their trading volume.2 By setting aside these statistical outliers, the number of threshold securities declined to 89 companies on February 15, 2005; 70 companies on March 22; and 64 companies on April 26. The analysis showed that these companies, comprising about half of the original NYSE and NASDAQ-listed threshold securities, could account for more than 95 percent of the aggregate fails reported by the SEC. Based on that analysis, these threshold securities averaged an estimated 1.5 million to 2 million fails each (Table 1).
1 The methodology used to adjust the aggregate fails will be provided if requested.
2 To ensure that "statistical outliers" in the other direction - i.e., very heavily-traded stocks -- did not distort the averages, we did not include any security with trading volume of 10 times or more of the average in calculating those averages. .
securities. In terms of both volume and short interest, the other dominant threshold company was Level 3 Communications (L VL T), accounting for 16.6 percent of trading volume and 16 percent of thresholds shorts. L VL T was followed by American Italian Pasta Company (PLB), a NYSE company which had more than 10 percent of all threshold-company shorts; followed in turn by Delta Airlines (DAL) and Taser International (TASR). All told, 21 companies on the February 15th threshold list had estimated fails exceeding 1,000,000 shares.
Table 3. Distribution of Fails for Threshold Securities, February 15, 2005
Trading Share of Outstanding Share of Estimated Share of
Volume Volume Shorts Shorts Fails Est Fails
_.....
KKD 48,343,600 3.06% 27,900,000 3.59% ¬
GG 30,234,900 1.91% 5,740,000 0.74% 1,762,398 1.33%
"'CYBX""". 29,858,m
.. .. . .:. .
SGI 21,281,000 1.35% 5,820,000 0.75% 1,392,817 1.05%
GLBC 5,116,582 0.33% 10,806,027 1.39% 1,140,023 0.86%
AIJ!'I 4,078,064 0.26% 7,332,628 0.95% 799,065 BSTE 3,472,978 0.2.2% 5,766,270 0.74% 639,558 0.48%
; ==: := t t= : '. " ::4. i : = m_::.. : -.m =:i iit :; = ¬
CREO 12,952.844 0.82% 822,212 0.11% 614,921 0.46%
UL TEQ 4,320,563 0.27% 2,667,206 0.34% 409,933 0.31 %
VLCCF 6,814,054 0.43% 873,099 0.11 % FICC 7,420,777 0.47% 253,450 0.03% 333,668 ACTT 5,071,104 0.32% 200,180 0.03% 230,326 0.17%
NAT 2,908,100 '0.18% 1,030,000 0.13% 210,419 0.16%
VOCL 3,954,447 0.25% 211,297
- EML 3,231,080 0.20% 419.,939 0.05% 171,010 0.13° _¬
NMGC 2,777,153 0.18% 31,494 0.004% 119,449 0.09%
. MH JJ70,700 0.11% 69,000 0.009% 80,347 0.06%
OTIV._1.,Q88,74§ 0.07% - t?Z!36L 0.02 59,241 .05%_¬
. : :. m_-_m
AEPPRA 471,500 0.03% 376,250 0.05% 52,026 0.04%
r Q:9° 1..OZ ..._..........
DA 2,30Q 9.:93% O,QOO Q.003%
Total 1,580,394,600 100% 776,242,438 100%
Adj Fails 132,881,617
Threshold Security List for March 22, 2005. Table 4 provides the level and share of
trading volume, outstanding shorts, and estimated fails-to-deliver for NYSE and NASDAQ-NM threshold companies for March 22, 2005. The two dominant threshold securities from the February list, the QQQQ Fund and Level Three Communications, do not appear on the March 22nd list. This time, the list is led by the NYSE-listed AMR Corp. (AMR) with an estimated 9.9 percent of all fails, based on its position as the most heavily-shorted and third most heavily-traded threshold security in March. The threshold securities with the second and third largest estimated fails were Delta Airlines (DAL) and Taser International (TASR) with an estimated 7.9 percent and 7 percent of all fails, respectively. DAL and T ASR ranked number four and five in the February 15, 2005 threshold list. Completing the top five on March 22, 2005 were TIVO (TIVO) and Krispy Kreme Donuts (KKD), accounting for an estimated 6.8 percent and 6.4 percent of fails. All told, 33 securities had estimated fails of 1,000,000 or more shares on March 22,2005.
Table 4. Distribution of Fails for Threshold Securities, March 22, 2005
Trading Share of Outstanding Share of Estimated Share of
Firm Volume Volume Shorts Shorts Fails l7ails .
AMR 72,482,000 7.85% 4,6720,000 12.00% 12,511,638 9.93%
IMAX 15,275,973 1.66% 6,879,009 1.77% 2,156,473 L71%
9_ . .m._m m m.??,41 &§Q m?,?§r.om... ?,.?: . "(...m¬
AGEt:! 6,428,147 0.70% 8,853,204 2.27% 1,872,102 1:49%
_.
AAII 9,394,799 1.02% 5,369,985 1.38% 1,510,699 1.20%
IDBE . 4;881 ,91? 0.53% 6,69.? ??§"' .. .9,Q?!0}4?,44?. .. ? 2.!0!0
XNVA 4,134,435 0.45% 133,263 0.03% 303,806 0.24%
SHMR 2,613 0.0003% 1,800 0.001 % 470 0.0004%
L8Total: Adj Fails
923,196,952 126,041,445
100%
389,267,922
100%
126,041,445
100%
Threshold Security Listfor April 26, 2005. Table 5, below, provides the level and share of trading volume, outstanding shorts, and estimated fails-to-deliver for NYSE and NASDAQ-NM threshold securities for April 26, 2005. As in the February 15, 2005, threshold list, the NASDAQ-I00 Index Fund, QQQQ, again is the dominant security, accounting for more than half of all trading volume in non-thinly traded threshold securities and for one-quarter of all short interest for those securities. The threshold security with the second-largest estimated fails on April 26, 2005 was Calpine Corporation, which did not appear on the threshold-security lists for February 15, 2005 or March 22, 2005. On April 26, 2005, Calpine had the second highest volume of any security of this list of non-thinly traded threshold issues, and the largest short interest, accounting for nearly one-third of all shorts for non-thinly traded threshold securities. The other three securities in the top five for estimated fails were also among the top six in the lists for February 15, 2005 and March 22, 2005: Delta Airlines (DAL) with an estimated 3.6% of all threshold-security fails; Taser International (TSR) with an estimated 3.6 percent of those fails; and Krispy Kreme Donuts (KKD) with an estimated 2.7 percent of all fails. All told, 15 threshold securities had estimated fails of 1,000,000 or more shares on April 26, 2005.
Table 5.
Distribution of Fails for Threshold Securities, April 26, 2005
Trading ' Share of Outstanding "S-hare cl Estimated Share'-m
Firm Volume Volume Shorts Shorts Fails of Fails
DAL 32,156,500 1.96% 39,230,000 5.32% 4,654,519 3.64%
TZOO 17,858,165 1.09% 1,595,827 0.22% 833,600 0.65%
Ir. I
BBD 8,943,700 0.54% 3,130,000 0.42% 619,650 0.48%
JOR- 11 ,5?4,500 . 70% _..1 :J 000 0.04% -_. 473,865 ... -¬
.. ACI--I 1,3§5,43.4 0.1.1% !,o.
Totals 1,643,612,880 100% 737,650,993 100% 127,964,806 100%
Adj Fails: 127,964,806 -- _m ..
Regulation SHO and Companies' Persistent Status as Threshold Securities. Our analysis also indicates that Regulation SHO has not prevented companies designated as threshold securities from continuing to have large numbers of fails (Tables 6A and 6B). Many threshold companies remain on the list for months: Among the top 25 securities
L10
from the February 15, 2000 threshold list, 11 or 44 percent were also so designated on March 22, 2005 and on April 26, 2005. Nor does a firm's designation as a threshold security significantly discourage or reduce short sales of its shares. Among the 11
companies appearing on the threshold lists on all three dates, short interest increased each month in five cases and fell by an average ofless than 12 percent in the other six cases.
Our analysis further suggests that in many cases, the number of fails can rise while a company remains on the threshold list. Using the combination of trading volume and short interest to estimate fails, estimated fails increased in all 11 cases from February 15, 2005 to March 22,2005. This finding could reflect the fact that the NASDAQ-lOO index fund QQQQ, which dominated the February 15,2005 threshold list, was absent from the March 22, 2005 list. Therefore, we also compared estimated fails in persistent threshold securities for February 15, 2005 and April 26, 2005, both of which included the NASDAQ-I00 index fund. Using a combination of trading volume and short interest to estimate fails (Table 6A), the estimated fails for one company, Napster (NAPS), increased from February 15, 2005 to April 26, 2005. Using short interest alone to estimate fails (Table 6B), the estimated fails for five of the 11 persistent threshold securities increased from February 15,2005 to April 26, 2005.
Table 6A. Short Interest and Estimated Fails, Based on Volume & Shorts
for Persistent Threshold Securities
Shorts Est. Fails - Based on Volume & Shorts
15-Feb 22-Mar 26-Apr 15-Feb 22-Mar 26-Apr
DAL 50,160,000 39,730,000 39,230,000 5,639,996 9,991,193 4,654,519
TASR 22,064,356 18,984,274 17,213,876 5,304,344 8,884,041 4,622,377
KKD 27,900,000 29,910,000 30,620,000 4,420,437 8,112,966 3,498,712
NFLX 15,531,946 14,110,024 14,005,403 1,942,257 3,408,439 1,812,003
ALD 16,100,000 15,030,000 14,950,000 1,647,017 2,71 ,582 1,513,811
MSO 7,240,000 7,160,000 6,710,000 1 ,428,802 3,637,788 928,872
NFl 7,330,000 8,560,000 8,430,000 1 ,255,012 1,966,687 911,897
GLBC 10,806,027 10,782,624 10,392,037 1,140,023 1,986,729 1,013,162
OSTK 4,889,912 5,276,868 5,444,496 1,089,032 1,659,032 892,807
NAPS 6,259,615 7,964,105 6,561,601 984,754 2,775,107 1,260,220
AGEN 7,927,845 8,853,204 7,976,018 892,855 1,872,102 827,297
11
lTable 6B. Short Interest and Estimated Fails, Based on Shorts Alone, for Persistent Threshold Securities
DAL
;tASR
KKD NFLX ALD MSO NFl GLBC OSTK NAPS AGEN
15-Feb
50,160,000 22,064,356 27,900,000 15,531,946 16,100,000 7,240,000
7,330,000
10,806,027
4,889,912 6,259,615 7,927,845
Shorts
22-Mar
39,730,000 18,984,274 29,910,000 14,110,024 15,030,000 7,160,000
8,560,000
10,782,624
5,276,868 7,964,105 8,853,204
26-Apr
39,230,000 17,213,876 30,620,000 14,005,403 14,950,000 6,710,000
8,430,000
10,392,037
5,444,496 6,561,601 7,976,018
Est. Fails - Based on Shorts Alone
15-Feb 22-Mar 26-Apr
8,586,675 12,864,216 6,805,467 3,777,103 6,146,937 2,986,169 4,776,081 9,684,588 5,311,838 2,658,847 4,890,476 2,533,605 2,756,090 4,866,579 2,593,467 1,239,385 2,318,343 1,164,025 1,254,791 2,771,651 1,462,403 1,849,837 3,491,316 1,802,770
837,083 1,708,602 944,490
1,071,557 2,578,705 1,138,281 1,357,134 2,866,588 1,383,648
Once again, while our analysis fairly establishes that fails are concentrated among certain threshold securities, only the DTCC has the data to confirm the level of fails for any particular company. It should release those data to investors and affected companies. More important, the SEC and DTCC should finally crack down on every case of large¬scale, extended naked shorting.
