In a recently published research paper, ‘Fails-to-Deliver, Short Selling and Market Quality’, three academics from Warwick University — Veljko Fotak, Vikas Raman and Predeep Yadav — attempt to measure the impact on market quality of regulatory measures to prohibit settlement failures in the US equity market. It is an excellent piece of practical research and a genuine ‘must-read’ paper for anyone engaged in the debate about financial market regulation.

While the conclusions of this study are not new, the empirical foundations are especially strong and the study is a powerful addition to the extensive body of empirical evidence suggesting that settlement failure does not damage market quality but actually serves to enhance price discovery and market liquidity. Such conclusions point squarely in the opposite direction to that which has been taken by regulators, who have taken fails to be a proxy for naked short selling and sought to control the latter by prohibiting the former.

Of course, intentional naked short selling, to the extent that it exists, is undesirable for reasons of market integrity, trust and confidence. The problem is that policy and regulation are unable to distinguish between intentional and accidental fails. As market practitioners know, but as politicians and regulators apparently refuse to believe, the prohibition of fails as a tool to suppress intentional naked short selling is a blunt instrument. It adds significantly to the cost of trading across the board and discourages legitimately-intended transactions, particularly in less liquid markets. The fear of penalties and other sanctions has the effect of sapping liquidity. This reduces the ability of arbitrageurs to realign overpriced securities with their true value and generally clogs up the channels of financial intermediation. It should therefore be no surprise that short selling regulations, mandatory buy-ins and unpredictably severe fines damage market quality, the very thing that regulation is intended to protect.

The authors of the study articulate very well the positive contribution of fails to market quality. Echoing previous research, they argue that ‘the difficulty of borrowing shares when short selling demand is high leads to costly frictions which negatively impact market quality; and the ability to fail is a release valve that helps to protect traders from the worst effects of these frictions: it is particularly valuable when stock-borrowing is so costly that short selling rebate rates become negative, which is exactly when liquidity is needed in the stock-borrowing market’. (Another important point is that fails are typically delays in settlement rather than permanent disruptions.)

Although the study investigates the equity market, there are no good reasons to suppose its conclusions do not apply to fixed income. In bond markets such as Italy and Spain, where fails have long been subject to penalties, the desire to avoid such costs (combined with a lack of securities lending) has had serious consequences for the market quality experienced by foreign investors. In the case of Italy, the penalty regime has actually exacerbated the problem for foreign investors. In the case of Spain, a high rate of settlement efficiency is belied by a high degree of latency in settlement by foreign investors.

What the study studied

The study examined settlement data from the US equity market provided by the DTCC (Depository Trust and Clearing Corporation). Specifically, there were two datasets covering 1,492 NYSE and 2,381 NASDAQ ordinary shares from January 2005 to June 2008.

The authors sought to measure whether fails impacted the efficiency of price discovery and market liquidity of the samples of securities by looking at changes in prices, pricing errors, intraday volatility, bid/offer spreads and order imbalances.

The study distinguished between market-makers and other traders by looking at the effect of the introduction, in September 2008, of SEC Rule 204T, which required brokers to borrow or purchase shares by the morning after a fail but only by the fourth morning in the case of market makers or failures of long sales.

A fail was defined as an insufficiency of a particular share in the accounts of a clearing member with a net delivery obligation in that share. Fails could therefore result from short selling without the timely arrangement of stock borrowing or sales of long positions where, for example, the shares had been lent out and not been recalled in time. To this extent, the analysis could not directly distinguish failed short sales from failed long sales. Nor did it directly distinguish intentional naked short selling (no intention of delivery) from accidental fails (beyond the control of the seller).

However, these shortcomings were addressed indirectly in a number of ways.

• To distinguish failed short sales, the study proxied short sales (both those setting first time and those failing first time) by daily changes in short interest net of fails. Secondly, use was made of the imposition of an SEC Emergency Order between 21 July and 12 August 2008 which prohibited the short selling of 19 financial shares. Given that the Order did not prohibit long sales, it provided an opportunity to measure quality without short selling.
• Intentional naked short selling was proxied by sub-sets of shares compiled on the basis of their high or persistent fail rates (and therefore most likely to be the targets of intentional naked short selling).

Finally, the study assessed the claims made by the media, investors and management that short selling caused crashes and distortions in the prices of certain financial sector shares during the height of the crisis in 2008, eg Bear Sterns, Lehman Brothers, Merrill Lynch and AIG. The data were tested for high fail rates before the dramatic price collapses in the shares of these companies.

What the study discovered

The study estimated that, for the NYSE sample, an increase in fails equivalent to 0.1% of the sample shares led to statistically significant reductions of 3% in pricing errors, 0.2% in intraday volatility and 1.7% in bid/offer spreads.

The beneficial impact of fails did not depend on whether market-makers or other traders failed.

The results were also similar for the proxy measure of short sales and for sub-sets of shares affected by high or persistent rates of settlement failure. And during the period subject to the 2008 SEC Emergency Order, market quality deteriorated as the number of fails dwindled, with significant increases in intra-day volatility, absolute pricing errors and bid/offer spreads. (Interestingly, the fall-off in the number of fails suggested that most arise from short sales.)

The authors’ conclusion was that failed trades, including failed short sales and perhaps even intentional naked short sales, enhanced price discovery and liquidity. This empirical evidence is consistent with the absence of any theoretical argument as to why failed short sales should have a different market impact to failed long sales. Sellers will not know they are going to fail when they trade. And automatic stock borrowing to cover fails means that there is no reason why the market should distinguish between trades that fail first time and trades that settle first time.

On the question of whether fails could have been a significant factor in the collapse and distortion of the share prices of Bear Sterns, Lehman Brothers, Merrill Lynch and AIG, the study found that the fail rates were insignificant prior to large price declines and only became abnormally large afterwards. In other words, the evidence refutes the claim that fails — and by implication, naked short selling — contributed to the demise of major financial institutions.

Why regulators need to read this paper

The authors of the study recommend that, instead of trying to attack naked short selling by prohibiting fails, it would be more productive to improve the stock borrowing market. They also tentatively propose progressive fines for fails.

These are sensible suggestions but not ones being followed by regulators. In the EU, regulators have travelled even further in the opposite direction of travel to the one signposted by the evidence. In addition to the Short Selling Regulation, implemented in 2012, we now have the Central Securities Depositories Regulation (CSDR). Ostensibly, this seeks to improve market quality by imposing penalties on failed deliveries and mandatory buy-ins (the real reason is probably the wish to ensure that the soon-to-be-launched T2S eurozone settlement system will not be blamed for fails beyond its control).

The Warwick University study has highlighted the damage that the CSDR may do to the quality of markets. Regrettably, that damage has been compounded by poorly informed drafting. This has inadvertently prohibited forward repos from being traded on electronic platforms. An erroneous assumption that most repos are overnight transactions has resulted in a regulation that is likely to fragment the market between short and longer-term repos, and between the repo and cash markets. Lack of understanding of how the fixed-income market functions, particularly about the lack of involvement of trading venues in settlement, means that financial intermediaries are faced with the expensive and disruptive task of rewiring the settlement infrastructure. And all this is to replace a consensual fails management process that has worked well.

As a recent paper by the ICMA has demonstrated, the cost of fails is likely to be increased to such an extent that financial intermediaries will seek to avoid the risk by drastically repricing, rationing or completely withdrawing from market-making. Meanwhile, investors will be discouraged from lending securities.

It’s time for a regulatory rethink.