Market MicrostructureNBER Reporter: Summer 2000
Barclay and Hendershott examine the trading process outside of normal trading hours. Although their trading volume is small, after-hours trades are more informative than trades during the day and are associated with significant price discovery. Spread-related trading costs are also more than twice as large after hours than during the trading day. Barclay and Hendershott observe two separate trading processes for Nasdaq-listed stocks in the after-hours market: larger less informative trades are negotiated directly with market makers, and smaller more informative trades are executed anonymously on electronic communications networks. Although both trading processes are active after the close and before the open, the non-anonymous, liquidity-motivated trades are more prevalent after the close, and the anonymous, information-motivated trades are more prevalent before the open.
Edelen and Gervais model an exchange as a collection of specialists, each a monopolist who makes a market in a subset of the stocks listed on the exchange. They show that specialists can obtain net private benefits at the expense of the exchange by quoting a privately optimal pricing schedule. Conversely, a coordinated pricing schedule makes all specialists and customers better off. However, coordination requires a system of "monitor jog and punishment," which can break down when information asymmetries between the exchange and the specialist are high. This breakdown can cause the specialist to seek a temporary halt to trading to alleviate unjustified punishment or can cause the exchange to halt trading to prevent the quoting of damaging, privately optimal pricing schedules. To test this theory, Edelen and Gervais use a sample of over 2,000 New York Stock Exchange (NYSE) halts and a proxy for the exchange-specialist information asymmetry. As predicted, they find a dramatic increase in estimated information asymmetry immediately preceding trading halts, which is far larger than the abnormal volatility or volume preceding the halt.
Christie, Corwin, and Harris study the impact of Nasdaq trading halts on prices, transaction costs, and trading activity. These halts, which stem from impending news, have a median duration of slightly less than one hour and produce a median absolute return of 5.5 percent. The authors find that the period after the halt is associated with unusually high volatility, share volume, and number of trades that are slow to decay. Transaction costs are also extremely high immediately after halts. In particular, inside-quoted spreads more than double after Nasdaq halts, decreasing to normal levels within 40 minutes. Overall, these results suggest that the price discovery process associated with Nasdaq halts may be inefficient.
The existence and enforcement of insider-trading laws in stock markets is a phenomenon of the 1990s. Of the 103 countries that now have stock markets, 87 have insider-trading laws, but in only 38 of them has enforcement -- as evidenced by prosecutions -- taken place. Before 1990, the respective numbers were 34 and 9. Bhattacharya and Daouk ask if this matters, because no study has yet documented empirically whether prohibitions against insider trading affect the cost of equity. They find that indeed it is the enforcement, not the existence, of insider-trading laws that matters. The cost of equity in a country (after controlling for risk factors, a liquidity factor, and other shareholder rights) is reduced by about 5 percent if insider-trading laws are enforced, they conclude.
By estimating the location and magnitude of price discovery across three informationally linked stock exchanges, Harris, McInish, and Wood can detect trades that permanently move the markets. They argue that after coincident but unequal price changes in synchronous trades, discovery of the new equilibrium price occurs in that trading venue to which other markets error correct. Common factor estimation (Gonzalo and Granger, 1995) summarizes each market's proportion of the price discovery in this error-correction sense. Across the Dow Jones Industrial Average (DJIA) stocks in 1988, the average common factor weight for the NYSE (72 percent) closely matched its share of the trades. The centralized market was "information dominant." However, by 1992 the proportion of the price discovery attributable to the NYSE had declined precipitously for 27 of the DJIA stocks, averaging only 49.6 percent. The NYSE's share of price discovery recovered substantially by 1995. The authors' tests confirm that these changes in the location of price discovery over time are statistically significant.