How to regulate financial volatility

6/14/2013 01:50:00 PM
Last week we learned that Thompson Reuters pays the University of Michigan over \$1 million annually to get their consumer sentiment survey five minutes before it was released to the general public. Reuters, in turn, sells this information to select customers who use it to profit by making bets on how the public will, five minutes later, react to the same information.

Neil Irwin has a proposal that, I think, expertly explains why this is wrong. And as Paul Samuelson put it, the economic benefit of having information moments before everyone will have it cannot be anything other than negligible (otherwise, why not release the info to everyone five minutes earlier?), yet those who do have this information early will surely make millions in profit as a result. Of course, Neil Irwin leaves one detail out: studies show that consumer sentiment surveys are basically meaningless. To be sure, consumer sentiment is co-incident--it falls when unemployment rises and incomes fall--but does not provide any additional information about the economy that can't be gleaned from "real" data variables like employment, GDP, inflation, etc. To put that differently, if you have an econometric forecast model that includes GDP, employment, inflation, and other common indicators, adding consumer sentiment to the model will not improve your forecast by a statistically significant margin.

I think the time has come to regulate the kind of insider trading that the University of Michigan and high-frequency traders have been getting away with for far too long. And here's the easiest way to do it: simply record all market trades as occurring at 5:00 pm each day. The way this would work is that traders would negotiate all day long, but the purchases and sales would not happen until precisely 5:00 pm, after the market has closed. A particular stock would only be able to be bought or sold once per day, and there would be only one price recorded for the stock per day. If that sounds unreasonable, I'd like to point out that, actually, this is how basically all macro finance theoretical models work. They are all discreet-time models where people trade once per period, and there is only on price for a given asset in a given period.

I think that such a system would do a number of things. One is that it would eliminate high-frequency trading once and for all--you can only buy a stock if you plan to own it for at least 24 hours. That, in turn, will reduce market volatility since it forces everyone to think about their choices for at least 24 hours before making them. It would also help put investors on equal footing, so that unsophisticated investors would be able to time their investments as well as the biggest wall street firms, eliminating market failures due to asymmetric information in the process. But the most important thing is that it would force investors to stop thinking about the timing of trades, and start thinking about prices. Ultimately, the only thing an investor should be considering when deciding whether to buy or sell is "what is the actual fundamental value of this company?" Under this system, one source of uncertainty--the timing of everyone's purchases--is completely removed, so that investors instead have to decide what prices at which they want to buy and sell for that day.

On the other hand, such a proposal might but Thompson-Reuters and CNBC out of business, but I consider that a social benefit in itself.