Prof. Jayanth R. Varma’s Financial Markets Blog

A blog on financial markets and their regulation

Monthly Archives: May 2010

Am on vacation

I am on vacation for the rest of this month. There will be no posts during this period. I shall try to moderate comments during this period, but there are bound to be delays.


Grand daddy of algorithmic trading bites the dust?

The ten minutes of mayhem in the US stock market last Thursday may have involved the oldest form of algorithmic trading – the stop loss order. We do not often think of the stop loss order as algorithmic trading, but that is what it is. If its conditions are satisfied, it executes without seeking any confirmation from the person who placed the order.

These days, it is usually a computer which implements the stop loss order, but even in the old days when a human broker implemented it, the stop loss was an algorithm. Whether the hardware on which an algorithm runs is made of silicon or of carbon is totally immaterial.

When I blogged last week about the dangers of market orders, I did not realize that many of the market orders that executed at absurd prices on Thursday afternoon in the US might have been stop loss orders. When the stop loss limit is breached, the stop loss order becomes a market order and executes blindly against whatever bids are available. This can be a prescription for disaster in fast moving markets as an avalanche of stop losses can eat through the entire order book and execute at penny prices as well.

Ironically, stop loss orders might not only have been a big victim of Thursday’s “flash crisis,” but might have been one of its major causes as well. Stop losses are inherently destabilizing as they aggravate the current trend. By demanding liquidity when it is least available, they degrade market quality. By making investors complacent about risks (the stop loss order is supposed to limit losses!), they tend to make investors reckless. All the more reason why these people ought not to have been bailed out by cancelling trades.

Market panics and bailout manias

People seem to be debating whether it was computers or humans that panicked and caused a temporary intraday drop of almost 10% in the US stock market yesterday. What we do know is that the bailout mania that followed was due entirely to humans. This is what the Nasdaq media release says:

The NASDAQ Stock Market had no technology or system issues associated with the trading that occurred between 2:00 and 3:00 p.m. ET today. The NASDAQ Stock Market operated continuously and its close process ran successfully.

In addition, there is no indication at this time that a NASDAQ market participant experienced a technological failure in connection with this event. NASDAQ has coordinated a process among U.S. Exchanges and therefore, pursuant to rule 11890(b), NASDAQ, on its own motion, will cancel all trades executed between 14:40:00 and 15:00:00 greater than or less than 60% away from the consolidated last print in that security at 14:40:00 or immediately prior. This decision cannot be appealed. NASDAQ has coordinated this decision with all other UTP Exchanges. NASDAQ will be canceling trades on the participant’s behalf.

Make no mistake. This is a bailout as bad and sordid as all the bailouts that we saw in the financial sector in 2008. It is bad because it reduces incentives for the firms to discipline their traders (or redesign their computer algorithms) to reduce the risk of such problems.

Anybody who uses a large market sell order instead of a marketable limit order during a falling market (and that too without looking at the order book) is begging to receive a price of zero for the stock. The market is happy to oblige. These people deserve the price that they got. There is no need for the exchange to ride to their rescue by cancelling their trades.

I wrote a few posts about this kind of thing four years ago:

  • Mizuho’s $335 million trading error: here and here
  • BSE Tulip trading error: here

Why have things not changed? Because, it is so easy to bailout everybody, it is much harder to change things.

Goldman Sachs and the economic function of investment banks

I wrote a column in the Financial Express on what the securities fraud case against Goldman Sachs tells us about the economic function of investment banks.

Regardless of its ultimate outcome, the SEC’s case against Goldman Sachs alleging securities fraud has already transformed the debate on financial sector reforms in the US. More importantly, I believe the case has raised disturbing questions about the economic function performed by investment banks in modern financial markets.

At the centre of the SEC case is the Abacus deal that Goldman brought to market in early 2007. The structure was created at the request of the hedge fund, Paulson & Co, which wanted to bet on the collapse of the US housing market by taking a short position on subprime securities. Goldman created synthetic subprime securities through the Abacus vehicle and sold these to the German bank, IKB. Paulson took the opposite (short) position on these securities.

The prospectus of Abacus highlighted the role of a reputed CDO manager, ACA Management, in selecting the portfolio of subprime assets underlying the Abacus deal and gave full details of this portfolio. But the 196-page prospectus did not mention the fact that Paulson had played a role in selecting the portfolio. This non-disclosure is a key element in the SEC case against Goldman.

In addition, the SEC alleges that Goldman misled ACA about Paulson’s intentions. Apparently, ACA believed that Paulson intended to buy the equity (first loss) piece of Abacus and was, therefore, motivated to exclude truly bad assets from the portfolio. In reality, Paulson planned to take a short position in Abacus and wished to stuff it with the worst possible assets.

It is difficult to predict the outcome of the SEC case because there are few precedents for invoking the anti-fraud provisions of US securities law in similar situations. The SEC might be in uncharted waters here, but it is pursuing a civil case where the standards of proof are lighter. Moreover, Goldman would certainly not relish having to defend its unsavoury conduct in a jury trial.

It is true that during the crisis the SEC acquired a reputation for incompetence (for example, Madoff and Stanford), which makes people sceptical about the Goldman case as well, but the new director of enforcement, Robert Khuzami, whom the SEC hired last year, has a formidable reputation from his days as a federal prosecutor.

Interestingly, Goldman in its defence thinks of itself more as a broker-dealer in complex derivatives and less as an issuer or underwriter of the Abacus securities. Broker-dealers have no obligation to disclose the identity or motivations of either counterparty to the other. It is true that Goldman could have achieved the same economic effect as Abacus by intermediating a credit default swap (CDS) between IKB and Paulson, but that is not what it chose to do. It chose to issue securities in which a CDS was embedded.

I am, however, less interested in whether the SEC wins this case or not. I am more concerned about the role of investment banks like Goldman in modern financial markets. In an ideal, perfectly efficient market, buyers and sellers would deal with each other directly through an electronic limit order book without any gatekeepers or intermediaries. In reality, intermediaries are needed to solve the problem of information asymmetry where one side knows a lot more about the transaction than the other.

It follows that the value added by an investment bank is measured by the extent to which it reduces information asymmetry. Otherwise, it is only exploiting oligopolistic rents or earning the rewards of excess leverage made possible by implicit ‘too big to fail’ guarantees.

From this perspective, the major banks of the 19th century or early 20th century like Rothschilds, Barings or JP Morgan did serve an economically useful function. Academic studies have shown that sovereign bonds underwritten by these major banks during that period had significantly lower default rates than other sovereign bonds (Flandreau, et al, 2009, The End of Gatekeeping: Underwriters and the Quality of Sovereign Bond Markets, 1815-2007, NBER Working Paper 15128).

Similarly, financial historians tell us that 19th century investment banks like JP Morgan played a critical role in bridging the information asymmetry between US railroads and their British investors. To their contemporaries, rich and powerful bankers like the Rothschilds and the Morgans were among the many ugly faces of capitalism. But the hard facts show that while they did not claim to be doing God’s work, they did do something useful.

The Abacus case makes one wonder whether modern investment banks do play any such useful role. The Goldman defence asserts that sophisticated investors like ACA and IKB were capable of looking after their own interests and did not need help from Goldman or anybody else. If there are no information asymmetries to be resolved or if modern investment banks have too little reputational capital to resolve them, then it is not at all clear what economic function they perform in today’s highly liquid and sophisticated markets.