Monday, May 10, 2010

A Volatile Wall Street

Rose asks, "Could you possibly do a blogpost on how to prevent or
adjust for something like Thursday's Wall Street panic attack?"

So I suppose my response to that would be that we shouldn't adjust
anything- people who are really affected (day traders, people who use
margin) know the risks. The necessary adjustments on Wall Street (and
there are many) regard fairness foremost (I'll address this in a
second) and risk-taking (less obvious), neither of which are reflected
in a one day panic.

That said, defusing volatility without harming the regular functions
of markets or raising the cost of capital is (obviously) a good thing.
Thus, like financial reform, procedures for halting, slowing or
defusing intraday crises would be valuable. I was wondering if they'd
close the exchange for a bit but A) it didnt hit the built in
circuitbreaker for the whole market, though some stocks did, and B)
that wouldn't necessarily have stopped the plunge on other networks
like ECNs. In fact, stopping primary trading could have even worsened
things bc of less buying liquidity but continued selling panic.

Thus, a number of big improvements come to mind. Firstly, make the
circuit breaker time dependent, as well as magnitude dependent.
Clearly, a 10% drop in ten minutes is more serious than a 10% drop in
a day. Circuit breakers that factored in speed of drop would be good.
Secondly, coordinate the networks such that when the primary exchange
closes down, the secondary ones do, too. Incidentally, this should
also apply to market opens and closes - currently, the NYSE opens 30
seconds later than a lot of the secondary NASDAQ or ECN networks, so
anyone who submits trades before the open can get cheated out of a
portion of the spread. The secondary networks should not be permitted
to trade when the primary listing exchange is closed. This has both
fairness and anti-panic benefits. There are also decent reasons to
implement the circuit breaker on an upward basis, as well.

Additionally, improving liquidity makes it harder to have a panic.
High-frequency trading can exacerbate panics and drives liquidity from
the market by functionally stealing a piece of every spread (for more
information, see my prior comments at number 7 here:

Flash trading (basically, where some people see trades a fraction of a
second before others) is blatantly unethical and drives liquidity from
the market because large block sellers or buyers don't want to be
jumped by hundredths of a penny by people who see their trades before
the market does. Dark pools make liquidity invisible and getting rid
of them would make the market deeper and less prone to panics (though
it's difficult to eliminate dark pools without getting rid of flash
trading first). To the Obama administration's credit, it has been
working to illegalize flash trading, and is looking at dark pools.

Some more ideas I've heard or had to combat UHFT and flash trading
(they're related but not the same) are to tax cancelled transactions
(instead of a turnover tax), or re-widening spreads away from pennies
to nickels (the market functioned perfectly fine in eighths until
1999). Even more radically, partially ECN-ing the market, where
matching and execution are not continuous but instead execution
happens at certain predetermined points (perhaps every 5 minutes, or
something like that) and only matching can happen in the interim,
would help to reduce a lot of the egregious HFT and flash trading

Finally, relaxing mark-to-market rules somewhat can, in some
instances, reduce liquidity crunches and, depending on the assets,
margin calls for some more sophisticated securities.

1 comment:

  1. Implementing the circuit breaker in the upward direction would be fantastic, but I think the odds are slim that the people running the exchanges (or the politicians who might try to browbeat them) would do it. Whatever the market throws at us is okay... so long as it's not falling, right?