In light of the news today that the SEC might not permanently apply its naked short selling “emergency order” to the entire market, and media reports that a study by Professor Arturo Bris is influencing this decision, we republish the following Deep Capture installment, which shows that Professor Bris quite blatantly fudged his numbers.
“The SEC’s public data say that on any given day over the first three months of this year, there were more than one billion shares that had been sold and failed to deliver (within the allotted 3 days) and that 70% of those fails were concentrated in just 100 companies. That’s a real red flag for the SEC that naked short selling is very widespread, is highly concentrated, and consequently might be being used today to manipulate the price of scores of stocks.”
-Former Deputy Secretary of Commerce Robert Shapiro on CNBC
It’s great that CNBC allowed someone to report this news. It seems pretty interesting – criminals manufacturing piles of phantom stock in order to systematically manipulate the share prices of perhaps 100 companies. Come to think of it, it sounds like a really big financial scandal.
Strange that in the week since Secretary Shapiro’s CNBC debut, naked short selling has not been mentioned even once in any mainstream news publication. And last we heard from some publications, they were arguing that the SEC should allow hedge funds to continue selling stock that they have not borrowed or purchased.
Which was different from a few months ago, when hedge funds and journalists were telling us that there was no such thing as naked short selling. As of last week, the new line was that naked short-selling happens all the time, but cracking down on it would cause irreparable harm to “market efficiency.”
This line even appeared in an editorial by the Economist. If it was in the Economist, it must have been true. Or maybe not. As someone who spent several years writing for the Wall Street Journal editorial page, which is similar to the Economist, I can tell you that the opinions of these places are informed by paradigms, not reporting. String together the words “market” and “efficiency,” throw in a threat of “regulation,” and they’ll be on your side, even if you’re defending criminals.
As it were, the articles in the Economist and every other publication were based almost entirely on a report by a guy named Arturo. As I noted in an earlier blog, some of these same publications reported that the SEC’s emergency order banning naked short selling in 19 financial stocks had caused the stocks to lose value, “according to Arturo Bris, a professor in Switzerland,” even though Arturo’s numbers showed quite clearly that the performances of those stocks had improved dramatically.
I don’t mean to pick on Professor Arturo, but when even the Economist is giving this guy the last word on naked short selling, it seems worth noting that the professor, with considerable help from the American hedge fund lobby, has poured into the media’s credulous gullets a mind-bending brew of cherry-picked numbers and calculated balderdash. Nearly every single number in his report contradicts his thesis that the SEC’s emergency order “significantly” harmed “market efficiency.”
I doubt any journalists read the report, but it should be obvious on the surface that its thesis is absurd.Markets are efficient when prices properly reflect supply and demand. You’d think that preventing people from diluting supply with a bunch of phantom stock would improve “market efficiency.” But apparently there is a “debate” over whether the market can efficiently set prices without criminals manipulating prices, so I hope some journalist, somewhere, will join me as I trudge through the only “expert” report on the planet that makes such a claim.
The report’s relevant section, “The Effect of the Emergency Order,” analyses the 19 affected stocks compared to a sample of 59 U.S. financial stocks not directly impacted by the order, and to a sample of non-U.S. financial stocks that, obviously, will be unaffected by any current or future SEC regulations. Professor Arturo chooses to focus his analysis on the following:
- Volatility (measured by “open-to-close” price volatility; “close to close” price volatility; and the so-called “trade price range”).
- Liquidity (measured by “Quoted Spreads” and “Relative Quoted Spreads”).
- Pricing Efficiency (measured using five statistics that show the extent to which there is a correlation between stock prices and swings in the overall market).
Professor Arturo would have us believe that the emergency order increased volatility, decreased liquidity, and increased market correlation (suggesting less efficient pricing). In fact, his numbers (and, indeed, his words, if you read them closely) suggest precisely the opposite.
I will work, line by line, through the report’s section titled “The Effect of the Emergency Order,” addressing Professor Arturo’s claims in order.
Professor Arturo begins by referring to table IX. Take a look.
According to this table, Professor Arturo writes,the SEC’s emergency order caused “significant volatility increases: open-to-close and close-to-close volatility [of the 19 affected stocks] increased 158 percent and 188 percent respectively. Trade price range increases 4.37 percent in the post-EO period.The table reports similar results for other measures.”
Professor Arturo has determined that open-to-close volatility of the 19 stocks “increased 158%” merely by subtracting the pre-EO open-to-close volatility (168.1%) from the post-EO volatility (327.4%) to get the “difference” of 158.23%.
Similarly, for close-to-close volatility, he merely subtracts the pre-EO number (216.18%) from the post-EO number (404.67%) to get the “difference” of 188.49%.Same for the trade price range: he subtracts pre-EO (2.74%) from post-EO (7.11%) to get the 4.37 number.
He highlights these “differences” throughout his text and in the above table, clearly intending for us to believe that they are important.
But the “differences” are completely irrelevant. They do not tell us by what percentage these numbers increased. And what is important is the whether the percent increases of the 19 stocks exceeded the percent increases of the U.S. and non-U.S. samples.
So let’s compare the increases of open-close volatility, close-close volatility, and trade price range.
Open-to-Close Volatility: For the 19 stocks, the increase is (difference) / (pre-EO volatility) = 158 / 168.1 =94.4%.For the U.S. sample, the increase is 79.68 / 123.65 =64%. For the non-U.S. sample, the increase is 70.60 / 64.34 = 109.7%.
In other words, open-close volatility of the non-U.S. stocks increased far more than that of either the 19 stocks or the U.S. sample. Given that the non-U.S. stocks are in no way affected by an SEC action in the U.S., we can assume the professor’s open-to-close volatility numbers say nothing about the effects of the emergency order.
Close-to-Close Volatility: For the 19 stocks, the increase is (difference) / (pre-EO close-close volatility) = 188.49 / 216.18 = 86.7%. For the U.S. sample, the increase is 170.37 / 93.65 = 182%.And for the non-U.S. sample, the increase is 120.87 /125.52 =102.5%.
In other words, the close-close volatility of the 19 affected stocks increased far less than that of both the U.S. and the non-U.S. sample.
Trade Price Range: For the 19 affected stocks, the increase is (difference) / (pre-EO trade price range) = 4.37 / 2.74 = 159%.For the U.S. sample, the increase is 4.20 / 2.79 =151%.For the non-U.S. sample, the increase is 1.53 / 2.39 = 64.01%.
The increase in the trade price range of the 19 stocks increased more than the U.S. and non-U.S. samples. But given that this contradicts the close-close and open-close volatility numbers, we certainly are not able to conclude that “volatility increased” as a result of the emergency order.
This might explain why, a few paragraphs after pointing to “significant volatility increases,” which is the line that was apparently fed to the press, Professor Arturo admits that “we do not find significant differences in volatility in either the pre or post EO period between G19 and US financial firms.”
Moving on to liquidity, Professor Arturo writes that “differences in liquidity [of the 19 affected stocks] significantly deteriorate.”
As noted, Professor Arturo measures liquidity by looking at quoted spreads and relative spreads. Again, he seems to see some significance in the “differences,” but what matters is the relative increases.
Look back up at that table. What you see is that for the 19 stocks, quoted spreads increased from $0.08 to $0.12 (50%). For the U.S. financial institutions, the increase was from $.0.04 to $0.06 (50%).
Isn’t 50 the same as 50?What “significantly deteriorated”?
As for relative quoted spreads, Professor Arturo writes in the introduction to his report that“from the pre-EO period to the post EO period, relative quoted spreads for G19 stocks have increased from 18 to 48 percent, but they have increased only from 11 percent to 29 percent for comparable US financial stocks. “
“Only from 11 percent,” he writes. Only? If something increases from 18 to 48, that’s a 166 percent increase. If something increases from 11 to 29 percent, that is a 163 percent increase. Isn’t 163 and 166 pretty much the same? This does not suggest that the 19 protected stocks “significantly deteriorated” relative to the sample of U.S. financials.
It is true that the spreads increased a lot more in the U.S. than they did overseas, but by this point Professor Arturo’s picture of what he calls “market quality” is looking pretty fuzzy.
Indeed, a bit further down, he writes that “controlling for firm and market characteristics, the EO has led to a significant increase in market liquidity.”
You read that right. Before he said there was a “deterioration” in liquidity. Then he said that the EO led to a significant increase in liquidity.
By the way, why am I wasting my time with this? Who cares about these screwball statistics?The SEC is talking about protecting companies from getting clobbered by illegal market manipulation. The SEC is talking about stopping a crime and upholding the basic tenet of capitalism and correct human conduct that says that someone who sells something had darn well better deliver it.
If some economist sees a change in some decimal point – big deal!If some blogging media critic had the stupid idea to stare cross-eyed at the economist’s decimal points until he noticed that they’d been completely fudged – well, big deal!Unless these numbers measure radioactivity, I don’t know why we’re even discussing them.Criminals are destroying market value and ruining lives. Decimal points be damned!
Sorry. Onwards with the report.
I notice that Professor Arturo throws some “semi-variance” numbers into the table that I posted above. Semi-variance increased dramatically for the 19 stocks. That must mean that “market quality” got really bad, right?
Wrong. Semi-variance isn’t a measure of volatility or liquidity. It is a measure of how much stocks could fall, based on their performance during a previous period. Perhaps Professor Arturo stuck the semi-variance numbers in the table to create a misimpression, but he doesn’t include them in his written discussion of the effects of the emergency order, no doubtbecause he knows they are not particularly relevant to “market quality” and “market efficiency” (though they do suggest that the performance of the 19 stocks soared during the emergency order, relative to the previous period, which is the opposite of what the professor and his media mimics said they did).
Moving on, Professor Arturo measures the correlation between the movement of stock prices and the movement of the market as a whole. If there’s a higher correlation, it supposedly suggests that the market isn’t efficiently processing information – that stock prices are determined by general market sentiment, rather than specific data points about the companies’ track records.
Professor Arturo measures correlation using five statistics, shown in the table below.
Referring to this table, the professor writes that there has been “an important deterioration of market efficiency as a result of the EO. The R squared increases from 22 to 33 percent for US financial firms (an absolute increase of 11 percent). R-squared increases 12 percent for G19 firms.”
You see? He did it again. He subtracted 22 from 33, which is 11. Since that’s less than 12, we’re supposed to believe that the R-squared for the 19 protected firms increased more than the R-squared of the U.S. financial firms. The table similarly displays these “absolute differences” as if they were the key to understanding the effects of the SEC’s ban on naked short selling.
But, again, the “difference” numbers are irrelevant. The relevant number, cited nowhere, is the percent increase of R-squared.For the sample of U.S. financial firms, the increase is 10.82 / 22.46 = .481, or 48%. For the 19 affected firms, the increase is 11.5 / 32.22 = .356 or 35.6%.
So the R-squared for the 19 firms increased less than the R-squared for the sample of U.S. financial firms, which is the opposite of what the professor would have us believe.
In the next line, Professor Arturo writes that “Cross-autocorrelation increases by the same magnitude in G19 and US financial stocks. However, cross-autocorrelation increases much more for G19 than for U.S. financial stocks.”
Yes, he said it increases by “the same magnitude.” Then he said the opposite–that it“increases by much more.” If this were the first time, I’d call it a mistake.
In any case, look at the table, and you will see what happened to the cross-autocorrelation of the 19 stocks. Before the emergency order there was a correlation of 8.24%. After the emergency order there was an inverse correlation of -24.4%.
In other words, the emergency order made it exceedingly less likely that the stocks would move with the market, which by professor Arturo’s standards, means the market became more efficient.
The professor goes on to say that “downside cross-correlation increases 3.26 percent for G19 stocks, while it decreases 4.05 percent for U.S. financial institutions.” So market efficiency “significantly deteriorated,” right?
No. Look at the table. Before the emergency order, downside cross-correlation for the 19 stocks was an insignificant -1.61%. After the emergency order, it was an insignificant 1.65%. In other words, there was never much of a correlation. By this standard, the market in the 19 stocks was almost perfectly efficient before the emergency order. And it remained almost perfectly efficient after the emergency order.
As for the sample of U.S. financials, downside cross-correlation was inversely correlated (-1.79%) before the emergency order. After the emergency order it was even more (-5.84%) inversely correlated. By this standard, the market for the U.S. sample of stocks became more efficient.
The downside cross-correlation numbers for foreign stocks show that they became less inversely correlated to the market after the emergency order. So, according to this statistic, the market in foreign stocks, but not U.S. became less efficient as a result of the emergency order in the United States. I don’t know what to conclude from that, but it certainly isn’t that U.S. market efficiency “significantly deteriorated” as the result of a ban on naked short selling.
Lastly, the downside R-squared of the 19 stocks increased 26.9 / 21.94 = 123%. That is significantly more than the increase in the U.S. financial stocks.However, the downside R-squared for overseas companies increased 56%, so obviously something other than an American regulatory action can affect downside R-squared.
All in all, most of the statistics in this report contradict the hedge fund party line that a ban on naked short selling harmed “market efficiency” – and that speaks volumes about the way in which our financial media processes and delivers information.
As for the data showing that criminals are hammering around 100 companies – destroying not just stocks but the lives of employees and small investors…Well, if you’ve read this far, you give a hoot, and that sets you apart from a great many journalists.