© Bloomberg

ETFs, ETNs and ETPs are among the most shorted instruments in the market, with short interest regularly hitting double-digit percentages and in some unique cases triple-digit percentages. 

Among the reasons they are so popular for shorting is their convenience and low cost, especially compared to alternatives like futures or options. Other advantages include the fact they are not subject to the uptick rule, which can bring trading advantages (or so we are told).

The way shorts are initiated in ETFs, however, has a significant influence on their assets under management. How so is not well understood by the market, and leads to poor signalling of trading sentiment.

As market experts confirmed to me this week, an ETF short position can be created in two different ways. The first is the traditional way whereby shares are located and borrowed from prime brokers. The second, and often cheaper, way is by using authorised participants (APs) through something known as the create-to-lend function.

It is this secondary system that can lead to the manifestation of large net inflows into a fund whenever shorting demand is particularly high. 

Short interest data indicates this sort of shorting was the probable driver of outsized inflows into the United States Oil Fund (USO), an exchange traded product based on derivative contracts, in recent weeks.

The inflows have generated much controversy and debate due to their disruptive impact on the underlying WTI oil futures market, which the ETP tracks, at a time of exceptionally weak fundamentals in the oil market.


The USO’s mandate obliges it to take on an increasingly dominant position in the front-month WTI contract as its assets under management grow. This causes market disruption because its investment mandate also obliges it to predictably roll those positions into the subsequent contract each month. Currently, prices of contracts further into the future are dearer than near-term ones due to bearish fundamentals. This causes a structure known as a contango. In the context of a derivative-backed ETP, such as the USO, contango allows traders to profit from positioning themselves in anticipation of a fund’s rolls, making it even dearer for the fund to reposition itself in the subsequent contract. The growing size of the fund only heightens its visibility and thus makes its losses worse.

John Hyland, now retired, is the former investment officer of United States Commodity Funds, the manager that operates the USO. He was in charge of the products back in 2007 and 2008 when they were afflicted by a similar outsized growth which led to comparable market misunderstandings.

The first myth Hyland was keen to correct in emailed correspondence with FT Alphaville was the notion that ETPs like the USO are predominantly the domain of the retail investor, which leads to the assumption it is retail buying that is causing the fund’s epic expansion:

Generally 80 per cent of USO’s shares are held by non-retail. For convenience sake call them hedge funds although that is not 100 per cent accurate. [They] include energy trading desks, hedge funds, and other professional players.

According to Hyland, sophisticated investors are particularly attracted to ETPs because of their convenience. USO, he explained, already fits into their portfolio management software or accounting systems, and there is no need to run parallel futures accounts or to manage both their collateral and exposure separately. Other times, ETPs are a useful fudge to get around investment mandates which would otherwise restrict trading to equities only. 

But the ease with which ETPs can be borrowed for shorting purposes, largely thanks to the create-to-lend function, is a factor behind their counterintuitive inflows at times of general market weakness . 

The ease comes about because prime brokers are able to generate brand new ETF units via creation requests to authorised participants whenever demand for borrowing stock is running high. All they need to do to initiate them is deliver the underlying collateral tracked by the products to the APs, which can often be easier to source than the ETF units themselves.

Regardless of whether the securities delivered to the AP are borrowed or purchased, the broker must then hedge their position to remain market neutral. If the PB opts to buy rather than borrow the underlying securities, they must also find a cost-efficient way to fund those securities (usually by charging lending fees that more than compensate for the cost of money). 

Among the advantages of meeting clients’ shorting demand with create-to-lend is the ability to benefit from the phenomenon of optimised baskets. This allows APs to deliver a representative sample of stocks tracked by the ETF rather than an identical representation of the index. This can reduce expenses by limiting purchases to the most liquid shares of all.

The create-to lend mechanism

In a 2016 paper, entitled Short Selling ETFs, Frank Weikai Li and Qifei Zhu, of the Singapore Management University and Nanyang Technological University respectively, noted that shorting via the create-to-lend mechanism allows investors to benefit from the well-connected nature of the financial intermediaries involved, especially with respect to cheaply sourcing hedges or collateral. This is another driver of efficiency and low cost.

There is very little public literature, however, with respect to how the create-to-lend function operates in ETPs backed by derivatives, such as the USO.

According to Hyland, it is likely the vast majority of shorting in the USO does represent create-to-lend activity:

A hedge fund wants to borrow 10,000,000 shares to short and it is easier and quicker for Merrill Lynch or whoever to just create them (the shares now belong to Merrill), Merrill will hedge their shares by being short the futures, and lend the shares to the hedge fund for the short.

Hyland estimated 90 per cent or more of short positioning in the ETP would involve professional rather than retail investors.

He further noted AUM almost always correlated inversely to the price of WTI:

As soon as WTI started falling in late February, I knew they would see major inflows.

Outflows, he explained, tend conversely to occur whenever the market is rallying. When oil prices hit record levels in 2008, for example, the USO was struck hard by redemptions making it a net seller of futures contracts in the market. Ironically, the creations it did attract, says Hyland, were driven mostly by brokers doing create-to-lend activity:

Back then only one or two brokers really dominated this activity because futures-based ETFs were still a bit of a new thing and the others had not yet wrapped their heads around the idea, but today I bet that just about any major broker can accommodate a major client here.

That there is evidence retail investors also bought heavily at the same time as shorts were being created is a function of short-sellers having to sell their shares to someone. Since the sellers occupy the role of the informed party, counterparts must de facto be less informed. But these uninformed buyers are unlikely to be the ones triggering or indeed funding the majority of the inflows. That role is more likely being played by brokers using the create-to-lend function to create new shares.

The following chart showing short interest in the USO from financial analytics firm S3 Partners seems to corroborate the theory. Short interest in the USO shot up as WTI prices began to sell off sharply:

The growth in the short interest, meanwhile, has mirrored the expansion of the AUM of the USO fund almost perfectly.

As of Wednesday short interest represented some 15.47 per cent of the ETP’s total float. Mark-to-market calculations imply the positions will have made up to $347.7m in profits since the beginning 2020 (net of financing costs).

According to S3 Partner’s head of predicative analytics, lhor Dusaniwsky, the cost to borrow “existing” USO shares was averaging about 3.8 per cent on Wednesday. In some bespoke cases, borrow rates were being quoted as high as 30 per cent and 35 per cent, with smaller clips reaching 75 per cent.

As can be seen in the chart below (orange line), that’s a multi-fold increase over the average price since the beginning of the year and marks an abrupt reversal of what was until then a steadily decreasing borrowing rate:

The data further shows borrowing rates began to spike higher the day before the ETP announced it could no longer honour creation requests because it had run out of pre-registered shares. As Dusaniwsky noted:

Stock borrow rates were declining for most of the year as USO’s ETN shares created grew to 1.4 billion and stock loan supply increased in lockstep. By mid-April, rates dropped below 1.0% fee after being above 2.0% fee at the beginning of the year. The recent spike in stock borrow fees occurred over the last couple of days wit rates going from 0.68% on 4/15 to 1.42% fee on the 20th and 3.80% fee on the 21st .

The falling borrowing rates for USO might indicate that prime brokers offering create-to-lend services were themselves benefiting from falling costs (or even windfalls) associated with funding and hedging their stock-lending positions.

The derivative factor

Faced with large levels of incoming short positioning, prime brokers – fully aware of how that growth would impact open interest in WTI futures market and the rolls of the fund – might have chosen to hedge their long USO positions by both selling the front month and buying the second to profit from the shifts in the curve structure driven by the USO’s rolls. This would have lowered their costs and added to their competitiveness.

But it may also have exacerbated the contango and invited the wider market to participate in pre-positioning ahead of the USO’s rolls as well. 

To offset this vulnerability, the USO has in recent days announced it will be shifting its holdings beyond the front-month WTI contract into the second and third month as well.

But many traders argue those contracts are not sufficiently liquid to absorb the USO’s flows meaning further price disruptions — to the detriment of the fund — are still likely. They add the fund’s tendency to buy through “Trade-at-Settlement” orders (a type of order which turns buyers into price takers) only adds to its visibility and losses.

Faced with ongoing liquidity constraints, the USO is likely to experiment with increasingly exotic instruments to achieve its investment objective. The terms of its prospectus, for example, allow it to diversify into everything from heating oil and RBOB gasoline and to Brent crude futures. As Hyland noted:

The next stop to my mind would be to start to use Brent contracts as well. It is almost as deep and liquid a market as WTI, correlates better to WTI, and better fits their overall mandate. In fact if I was still running USCF I might have already announced a move into Brent as well.
I would also not be surprised if USO lengthens their roll period another day or so to smooth out that process. Once again if I was running USCF I likely would have announced that at the same time I announced were going to both multiple months AND to Brent. All in one fell swoop.

For now, the market can only speculate. But a clue to what might happen next comes in the shape of something that’s already happened. When natgas prices cratered in the summer of 2009, the USO’s sister product, the United States Natural Gas Fund (UNG), experienced eerily similar stresses.

First its AUM exploded, then it started experimenting with exotic instruments, then its registered shares got depleted, then its creations were suspended, then it continued to trade at a premium despite creations being suspended, then the SEC failed to grant it permission for the issuance of new shares, then the SEC changed its mind, and then regulators got involved with position limits, and then, finally, the fund reopened and lived to fight another day . 

Related links:
The problem with commodity ETFs – FT Alphaville
More weirdness at the UNG – FT Alphaville
The United States Oil Fund mystery, revived – FT Alphaville 
What’s driving the USO ETF’s sudden AUM growth? – FTAV Screencast

Copyright The Financial Times Limited 2020. All rights reserved. You may share using our article tools. Please don’t cut articles from FT.com and redistribute by email or post to the web.

Source link